[PATCH/RFC] checkout: print something when checking out paths

[PATCH/RFC] checkout: print something when checking out paths

[Nguyễn Thái Ngọc Duy]

One of the problems with "git checkout" is that it does so many
different things and could confuse people specially when we fail to
handle ambiguation correctly.

One way to help with that is tell the user what sort of operation is
actually carried out. When switching branches, we always print
something [1].  Checking out paths however is silent. Print something
so that if we got the user intention wrong, they won't waste too much
time to find that out.

Since the purpose of printing this is to help disambiguate. Only do it
when "--" is missing (the actual reason though is many tests check
empty stderr to see that no error is raised and I'm too lazy to fix
all the test cases).

[1] Knowing the number of paths updated could also be useful even in
    normal case.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 This is related to another patch in
 
 https://public-inbox.org/git/20181110120707.25846-1-pclouds@gmail.com/T/#u

 While writing that patch I thought printing something would also
 help. But if we get ambiguation right (in that particular case) then
 we don't actually need this. But it still seems a good idea...

 apply.c                  |  3 ++-
 builtin/checkout-index.c |  6 ++++--
 builtin/checkout.c       | 30 ++++++++++++++++++++++--------
 builtin/difftool.c       |  2 +-
 cache.h                  |  4 ++--
 entry.c                  | 10 ++++++----
 unpack-trees.c           |  6 +++---
 7 files changed, 40 insertions(+), 21 deletions(-)

diff --git a/apply.c b/apply.c
index 073d5f0451..5876b02197 100644
--- a/apply.c
+++ b/apply.c
@@ -3352,7 +3352,8 @@ static int checkout_target(struct index_state *istate,
 
 	costate.refresh_cache = 1;
 	costate.istate = istate;
-	if (checkout_entry(ce, &costate, NULL) || lstat(ce->name, st))
+	if (checkout_entry(ce, &costate, NULL, NULL) ||
+	    lstat(ce->name, st))
 		return error(_("cannot checkout %s"), ce->name);
 	return 0;
 }
diff --git a/builtin/checkout-index.c b/builtin/checkout-index.c
index 88b86c8d9f..bada491f58 100644
--- a/builtin/checkout-index.c
+++ b/builtin/checkout-index.c
@@ -67,7 +67,8 @@ static int checkout_file(const char *name, const char *prefix)
 			continue;
 		did_checkout = 1;
 		if (checkout_entry(ce, &state,
-		    to_tempfile ? topath[ce_stage(ce)] : NULL) < 0)
+				   to_tempfile ? topath[ce_stage(ce)] : NULL,
+				   NULL) < 0)
 			errs++;
 	}
 
@@ -111,7 +112,8 @@ static void checkout_all(const char *prefix, int prefix_length)
 				write_tempfile_record(last_ce->name, prefix);
 		}
 		if (checkout_entry(ce, &state,
-		    to_tempfile ? topath[ce_stage(ce)] : NULL) < 0)
+				   to_tempfile ? topath[ce_stage(ce)] : NULL,
+				   NULL) < 0)
 			errs++;
 		last_ce = ce;
 	}
diff --git a/builtin/checkout.c b/builtin/checkout.c
index acdafc6e4c..13ed041dc1 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -44,6 +44,7 @@ struct checkout_opts {
 	int ignore_skipworktree;
 	int ignore_other_worktrees;
 	int show_progress;
+	int count_checkout_paths;
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
 	 * should be updated accordingly.
@@ -165,12 +166,13 @@ static int check_stages(unsigned stages, const struct cache_entry *ce, int pos)
 }
 
 static int checkout_stage(int stage, const struct cache_entry *ce, int pos,
-			  const struct checkout *state)
+			  const struct checkout *state, int *nr_checkouts)
 {
 	while (pos < active_nr &&
 	       !strcmp(active_cache[pos]->name, ce->name)) {
 		if (ce_stage(active_cache[pos]) == stage)
-			return checkout_entry(active_cache[pos], state, NULL);
+			return checkout_entry(active_cache[pos], state,
+					      NULL, nr_checkouts);
 		pos++;
 	}
 	if (stage == 2)
@@ -179,7 +181,7 @@ static int checkout_stage(int stage, const struct cache_entry *ce, int pos,
 		return error(_("path '%s' does not have their version"), ce->name);
 }
 
-static int checkout_merged(int pos, const struct checkout *state)
+static int checkout_merged(int pos, const struct checkout *state, int *nr_checkouts)
 {
 	struct cache_entry *ce = active_cache[pos];
 	const char *path = ce->name;
@@ -242,7 +244,7 @@ static int checkout_merged(int pos, const struct checkout *state)
 	ce = make_transient_cache_entry(mode, &oid, path, 2);
 	if (!ce)
 		die(_("make_cache_entry failed for path '%s'"), path);
-	status = checkout_entry(ce, state, NULL);
+	status = checkout_entry(ce, state, NULL, nr_checkouts);
 	discard_cache_entry(ce);
 	return status;
 }
@@ -257,6 +259,7 @@ static int checkout_paths(const struct checkout_opts *opts,
 	struct commit *head;
 	int errs = 0;
 	struct lock_file lock_file = LOCK_INIT;
+	int nr_checkouts = 0;
 
 	if (opts->track != BRANCH_TRACK_UNSPECIFIED)
 		die(_("'%s' cannot be used with updating paths"), "--track");
@@ -371,17 +374,27 @@ static int checkout_paths(const struct checkout_opts *opts,
 		struct cache_entry *ce = active_cache[pos];
 		if (ce->ce_flags & CE_MATCHED) {
 			if (!ce_stage(ce)) {
-				errs |= checkout_entry(ce, &state, NULL);
+				errs |= checkout_entry(ce, &state,
+						       NULL, &nr_checkouts);
 				continue;
 			}
 			if (opts->writeout_stage)
-				errs |= checkout_stage(opts->writeout_stage, ce, pos, &state);
+				errs |= checkout_stage(opts->writeout_stage,
+						       ce, pos,
+						       &state, &nr_checkouts);
 			else if (opts->merge)
-				errs |= checkout_merged(pos, &state);
+				errs |= checkout_merged(pos, &state,
+							&nr_checkouts);
 			pos = skip_same_name(ce, pos) - 1;
 		}
 	}
-	errs |= finish_delayed_checkout(&state);
+	errs |= finish_delayed_checkout(&state, &nr_checkouts);
+
+	if (opts->count_checkout_paths)
+		fprintf_ln(stderr, Q_("%d path has been updated",
+				      "%d paths have been updated",
+				      nr_checkouts),
+			   nr_checkouts);
 
 	if (write_locked_index(&the_index, &lock_file, COMMIT_LOCK))
 		die(_("unable to write new index file"));
@@ -1064,6 +1077,7 @@ static int parse_branchname_arg(int argc, const char **argv,
 		has_dash_dash = 1; /* case (3) or (1) */
 	else if (dash_dash_pos >= 2)
 		die(_("only one reference expected, %d given."), dash_dash_pos);
+	opts->count_checkout_paths = !opts->quiet && !has_dash_dash;
 
 	if (!strcmp(arg, "-"))
 		arg = "@{-1}";
diff --git a/builtin/difftool.c b/builtin/difftool.c
index 544b0e8639..71318c26e1 100644
--- a/builtin/difftool.c
+++ b/builtin/difftool.c
@@ -323,7 +323,7 @@ static int checkout_path(unsigned mode, struct object_id *oid,
 	int ret;
 
 	ce = make_transient_cache_entry(mode, oid, path, 0);
-	ret = checkout_entry(ce, state, NULL);
+	ret = checkout_entry(ce, state, NULL, NULL);
 
 	discard_cache_entry(ce);
 	return ret;
diff --git a/cache.h b/cache.h
index 8b1ee42ae9..52fb6ba148 100644
--- a/cache.h
+++ b/cache.h
@@ -1540,9 +1540,9 @@ struct checkout {
 #define CHECKOUT_INIT { NULL, "" }
 
 #define TEMPORARY_FILENAME_LENGTH 25
-extern int checkout_entry(struct cache_entry *ce, const struct checkout *state, char *topath);
+extern int checkout_entry(struct cache_entry *ce, const struct checkout *state, char *topath, int *nr_checkouts);
 extern void enable_delayed_checkout(struct checkout *state);
-extern int finish_delayed_checkout(struct checkout *state);
+extern int finish_delayed_checkout(struct checkout *state, int *nr_checkouts);
 
 struct cache_def {
 	struct strbuf path;
diff --git a/entry.c b/entry.c
index 5d136c5d55..5f213c30fe 100644
--- a/entry.c
+++ b/entry.c
@@ -161,7 +161,7 @@ static int remove_available_paths(struct string_list_item *item, void *cb_data)
 	return !available;
 }
 
-int finish_delayed_checkout(struct checkout *state)
+int finish_delayed_checkout(struct checkout *state, int *nr_checkouts)
 {
 	int errs = 0;
 	unsigned delayed_object_count;
@@ -226,7 +226,7 @@ int finish_delayed_checkout(struct checkout *state)
 				ce = index_file_exists(state->istate, path->string,
 						       strlen(path->string), 0);
 				if (ce) {
-					errs |= checkout_entry(ce, state, NULL);
+					errs |= checkout_entry(ce, state, NULL, nr_checkouts);
 					filtered_bytes += ce->ce_stat_data.sd_size;
 					display_throughput(progress, filtered_bytes);
 				} else
@@ -435,8 +435,8 @@ static void mark_colliding_entries(const struct checkout *state,
  * its name is returned in topath[], which must be able to hold at
  * least TEMPORARY_FILENAME_LENGTH bytes long.
  */
-int checkout_entry(struct cache_entry *ce,
-		   const struct checkout *state, char *topath)
+int checkout_entry(struct cache_entry *ce, const struct checkout *state,
+		   char *topath, int *nr_checkouts)
 {
 	static struct strbuf path = STRBUF_INIT;
 	struct stat st;
@@ -506,5 +506,7 @@ int checkout_entry(struct cache_entry *ce,
 		return 0;
 
 	create_directories(path.buf, path.len, state);
+	if (nr_checkouts)
+		(*nr_checkouts)++;
 	return write_entry(ce, path.buf, state, 0);
 }
diff --git a/unpack-trees.c b/unpack-trees.c
index 7570df481b..17f1e601da 100644
--- a/unpack-trees.c
+++ b/unpack-trees.c
@@ -294,7 +294,7 @@ static void load_gitmodules_file(struct index_state *index,
 			repo_read_gitmodules(the_repository);
 		} else if (state && (ce->ce_flags & CE_UPDATE)) {
 			submodule_free(the_repository);
-			checkout_entry(ce, state, NULL);
+			checkout_entry(ce, state, NULL, NULL);
 			repo_read_gitmodules(the_repository);
 		}
 	}
@@ -450,12 +450,12 @@ static int check_updates(struct unpack_trees_options *o)
 			display_progress(progress, ++cnt);
 			ce->ce_flags &= ~CE_UPDATE;
 			if (o->update && !o->dry_run) {
-				errs |= checkout_entry(ce, &state, NULL);
+				errs |= checkout_entry(ce, &state, NULL, NULL);
 			}
 		}
 	}
 	stop_progress(&progress);
-	errs |= finish_delayed_checkout(&state);
+	errs |= finish_delayed_checkout(&state, NULL);
 	if (o->update)
 		git_attr_set_direction(GIT_ATTR_CHECKIN);
 
-- 
2.19.1.1231.g84aef82467

Re: [PATCH/RFC] checkout: print something when checking out paths

[Junio C Hamano]

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> Since the purpose of printing this is to help disambiguate. Only do it
> when "--" is missing (the actual reason though is many tests check
> empty stderr to see that no error is raised and I'm too lazy to fix
> all the test cases).

Heh, honesty is good but in this particular case the official reason
alone would make perfect sense, too ;-)

As with progress output, shouldn't this automatically be turned off
when the standard error stream goes to non tty, as the real purpose
of printing is to help the user sitting in front of the terminal and
interacting with the command?

And by this, I do not mean to say that "When -- is missing" can go
away.  I agree that "--" is a clear sign that the user knows what
s/he is doing---to overwrite the paths in the working tree that
match the pathspec.

> @@ -371,17 +374,27 @@ static int checkout_paths(const struct checkout_opts *opts,
>  		struct cache_entry *ce = active_cache[pos];
>  		if (ce->ce_flags & CE_MATCHED) {
>  			if (!ce_stage(ce)) {
> -				errs |= checkout_entry(ce, &state, NULL);
> +				errs |= checkout_entry(ce, &state,
> +						       NULL, &nr_checkouts);
>  				continue;

As we count inside checkout_entry(), we might not actually write
this out when we know that the working tree file is up to date
already but we do not increment in that case either, so we keep
track of the accurate count of "updates", not path matches (i.e. we
are not reporting "we made sure this many paths are up to date in
the working tree"; instead we are reporting how many paths we needed
to overwrite in the working tree).

>  			}
>  			if (opts->writeout_stage)
> -				errs |= checkout_stage(opts->writeout_stage, ce, pos, &state);
> +				errs |= checkout_stage(opts->writeout_stage,
> +						       ce, pos,
> +						       &state, &nr_checkouts);

Likewike.

>  			else if (opts->merge)
> -				errs |= checkout_merged(pos, &state);
> +				errs |= checkout_merged(pos, &state,
> +							&nr_checkouts);

Likewise.

>  			pos = skip_same_name(ce, pos) - 1;
>  		}
>  	}
> -	errs |= finish_delayed_checkout(&state);
> +	errs |= finish_delayed_checkout(&state, &nr_checkouts);
> +
> +	if (opts->count_checkout_paths)
> +		fprintf_ln(stderr, Q_("%d path has been updated",
> +				      "%d paths have been updated",
> +				      nr_checkouts),
> +			   nr_checkouts);

This one may want to be protected behind isatty(2).  Or the default
value of count_checkout_paths could be tweakedbased on isatty(2).

> @@ -1064,6 +1077,7 @@ static int parse_branchname_arg(int argc, const char **argv,
>  		has_dash_dash = 1; /* case (3) or (1) */
>  	else if (dash_dash_pos >= 2)
>  		die(_("only one reference expected, %d given."), dash_dash_pos);
> +	opts->count_checkout_paths = !opts->quiet && !has_dash_dash;

i.e. "&& isatty(2)" as well.

Of course, "--[no-]count-paths" could be added to override this.

Re: [PATCH/RFC] checkout: print something when checking out paths

[Duy Nguyen]

On Mon, Nov 12, 2018 at 7:21 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>
> > Since the purpose of printing this is to help disambiguate. Only do it
> > when "--" is missing (the actual reason though is many tests check
> > empty stderr to see that no error is raised and I'm too lazy to fix
> > all the test cases).
>
> Heh, honesty is good but in this particular case the official reason
> alone would make perfect sense, too ;-)
>
> As with progress output, shouldn't this automatically be turned off
> when the standard error stream goes to non tty, as the real purpose
> of printing is to help the user sitting in front of the terminal and
> interacting with the command?

I see this at the same level as "Switched to branch 'foo'" which is
also protected by opts->quiet. If we start hiding messages based on
tty it should be done for other messages in update_refs_for_switch()
too, I guess?

> And by this, I do not mean to say that "When -- is missing" can go
> away.  I agree that "--" is a clear sign that the user knows what
> s/he is doing---to overwrite the paths in the working tree that
> match the pathspec.

My other problem with deciding to print based on the presence of "--"
is also with branch switching code: it always prints something (unless
it actually does nothing). So it's a bit strange that the
checkout_paths() behaves slightly different based on "--".


> > @@ -371,17 +374,27 @@ static int checkout_paths(const struct checkout_opts *opts,
> >               struct cache_entry *ce = active_cache[pos];
> >               if (ce->ce_flags & CE_MATCHED) {
> >                       if (!ce_stage(ce)) {
> > -                             errs |= checkout_entry(ce, &state, NULL);
> > +                             errs |= checkout_entry(ce, &state,
> > +                                                    NULL, &nr_checkouts);
> >                               continue;
>
> As we count inside checkout_entry(), we might not actually write
> this out when we know that the working tree file is up to date
> already but we do not increment in that case either, so we keep
> track of the accurate count of "updates", not path matches (i.e. we
> are not reporting "we made sure this many paths are up to date in
> the working tree"; instead we are reporting how many paths we needed
> to overwrite in the working tree).

Yeah. I counted matched paths first, but when you do "git co .", you
get a huge (and meaningless, to me) number. Printing how many files
are updated makes more sense.

> >                       pos = skip_same_name(ce, pos) - 1;
> >               }
> >       }
> > -     errs |= finish_delayed_checkout(&state);
> > +     errs |= finish_delayed_checkout(&state, &nr_checkouts);
> > +
> > +     if (opts->count_checkout_paths)
> > +             fprintf_ln(stderr, Q_("%d path has been updated",
> > +                                   "%d paths have been updated",
> > +                                   nr_checkouts),
> > +                        nr_checkouts);
>
> This one may want to be protected behind isatty(2).  Or the default
> value of count_checkout_paths could be tweakedbased on isatty(2).

Another thing I'm going to change (if this v1 is in the right
direction): print different messages for "git checkout -- foo" and
"git checkout foo -- bar". The first one is "%d paths have been
reverted" but the second one does different things and probably should
say "%d paths have been updated from branch foo" or something like
that.
-- 
Duy

Re: [PATCH/RFC] checkout: print something when checking out paths

[Junio C Hamano]

Duy Nguyen <pclouds@gmail.com> writes:

> Another thing I'm going to change (if this v1 is in the right
> direction): print different messages for "git checkout -- foo" and
> "git checkout foo -- bar". The first one is "%d paths have been
> reverted" but the second one does different things and probably should
> say "%d paths have been updated from branch foo" or something like
> that.

I actually think that it is a bad idea to deliberately use such
misleading words like "revert", that will discourage users from
weaning themselves off of SVN.  "checked out N paths out of the
index", "checked out N paths out of the commit X" and "checked out N
paths out of branch B" would be clear enough and won't have such a
problem---otherwise you are encouraging them to make a mistake

	echo garbage >path
	... say oops ...
	git revert path

[PATCH v2] checkout: print something when checking out paths

[Nguyễn Thái Ngọc Duy]

One of the problems with "git checkout" is that it does so many
different things and could confuse people specially when we fail to
handle ambiguation correctly.

One way to help with that is tell the user what sort of operation is
actually carried out. When switching branches, we always print
something unless --quiet, either

 - "HEAD is now at ..."
 - "Reset branch ..."
 - "Already on ..."
 - "Switched to and reset ..."
 - "Switched to a new branch ..."
 - "Switched to branch ..."

Checking out paths however is silent. Print something so that if we
got the user intention wrong, they won't waste too much time to find
that out. For the remaining cases of checkout we now print either

 - "Checked out ... paths out of the index"
 - "Checked out ... paths out of <abbrev hash>"

Since the purpose of printing this is to help disambiguate. Only do it
when "--" is missing.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 v2 updates the messages a bit but it does not check isatty or add
 --count-paths, for consistency reason with how messages are printed
 in the branch switching case.
 
 Consistency is not always a good reason to follow. But I haven't
 seen a strong reason to go against it.

 apply.c                  |  3 ++-
 builtin/checkout-index.c |  6 ++++--
 builtin/checkout.c       | 39 +++++++++++++++++++++++++++++++--------
 builtin/difftool.c       |  2 +-
 cache.h                  |  4 ++--
 entry.c                  | 10 ++++++----
 unpack-trees.c           |  6 +++---
 7 files changed, 49 insertions(+), 21 deletions(-)

diff --git a/apply.c b/apply.c
index 073d5f0451..5876b02197 100644
--- a/apply.c
+++ b/apply.c
@@ -3352,7 +3352,8 @@ static int checkout_target(struct index_state *istate,
 
 	costate.refresh_cache = 1;
 	costate.istate = istate;
-	if (checkout_entry(ce, &costate, NULL) || lstat(ce->name, st))
+	if (checkout_entry(ce, &costate, NULL, NULL) ||
+	    lstat(ce->name, st))
 		return error(_("cannot checkout %s"), ce->name);
 	return 0;
 }
diff --git a/builtin/checkout-index.c b/builtin/checkout-index.c
index 88b86c8d9f..bada491f58 100644
--- a/builtin/checkout-index.c
+++ b/builtin/checkout-index.c
@@ -67,7 +67,8 @@ static int checkout_file(const char *name, const char *prefix)
 			continue;
 		did_checkout = 1;
 		if (checkout_entry(ce, &state,
-		    to_tempfile ? topath[ce_stage(ce)] : NULL) < 0)
+				   to_tempfile ? topath[ce_stage(ce)] : NULL,
+				   NULL) < 0)
 			errs++;
 	}
 
@@ -111,7 +112,8 @@ static void checkout_all(const char *prefix, int prefix_length)
 				write_tempfile_record(last_ce->name, prefix);
 		}
 		if (checkout_entry(ce, &state,
-		    to_tempfile ? topath[ce_stage(ce)] : NULL) < 0)
+				   to_tempfile ? topath[ce_stage(ce)] : NULL,
+				   NULL) < 0)
 			errs++;
 		last_ce = ce;
 	}
diff --git a/builtin/checkout.c b/builtin/checkout.c
index acdafc6e4c..3a0b86ec1c 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -44,6 +44,7 @@ struct checkout_opts {
 	int ignore_skipworktree;
 	int ignore_other_worktrees;
 	int show_progress;
+	int count_checkout_paths;
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
 	 * should be updated accordingly.
@@ -165,12 +166,13 @@ static int check_stages(unsigned stages, const struct cache_entry *ce, int pos)
 }
 
 static int checkout_stage(int stage, const struct cache_entry *ce, int pos,
-			  const struct checkout *state)
+			  const struct checkout *state, int *nr_checkouts)
 {
 	while (pos < active_nr &&
 	       !strcmp(active_cache[pos]->name, ce->name)) {
 		if (ce_stage(active_cache[pos]) == stage)
-			return checkout_entry(active_cache[pos], state, NULL);
+			return checkout_entry(active_cache[pos], state,
+					      NULL, nr_checkouts);
 		pos++;
 	}
 	if (stage == 2)
@@ -179,7 +181,7 @@ static int checkout_stage(int stage, const struct cache_entry *ce, int pos,
 		return error(_("path '%s' does not have their version"), ce->name);
 }
 
-static int checkout_merged(int pos, const struct checkout *state)
+static int checkout_merged(int pos, const struct checkout *state, int *nr_checkouts)
 {
 	struct cache_entry *ce = active_cache[pos];
 	const char *path = ce->name;
@@ -242,7 +244,7 @@ static int checkout_merged(int pos, const struct checkout *state)
 	ce = make_transient_cache_entry(mode, &oid, path, 2);
 	if (!ce)
 		die(_("make_cache_entry failed for path '%s'"), path);
-	status = checkout_entry(ce, state, NULL);
+	status = checkout_entry(ce, state, NULL, nr_checkouts);
 	discard_cache_entry(ce);
 	return status;
 }
@@ -257,6 +259,7 @@ static int checkout_paths(const struct checkout_opts *opts,
 	struct commit *head;
 	int errs = 0;
 	struct lock_file lock_file = LOCK_INIT;
+	int nr_checkouts = 0;
 
 	if (opts->track != BRANCH_TRACK_UNSPECIFIED)
 		die(_("'%s' cannot be used with updating paths"), "--track");
@@ -371,17 +374,36 @@ static int checkout_paths(const struct checkout_opts *opts,
 		struct cache_entry *ce = active_cache[pos];
 		if (ce->ce_flags & CE_MATCHED) {
 			if (!ce_stage(ce)) {
-				errs |= checkout_entry(ce, &state, NULL);
+				errs |= checkout_entry(ce, &state,
+						       NULL, &nr_checkouts);
 				continue;
 			}
 			if (opts->writeout_stage)
-				errs |= checkout_stage(opts->writeout_stage, ce, pos, &state);
+				errs |= checkout_stage(opts->writeout_stage,
+						       ce, pos,
+						       &state, &nr_checkouts);
 			else if (opts->merge)
-				errs |= checkout_merged(pos, &state);
+				errs |= checkout_merged(pos, &state,
+							&nr_checkouts);
 			pos = skip_same_name(ce, pos) - 1;
 		}
 	}
-	errs |= finish_delayed_checkout(&state);
+	errs |= finish_delayed_checkout(&state, &nr_checkouts);
+
+	if (opts->count_checkout_paths) {
+		if (opts->source_tree)
+			fprintf_ln(stderr, Q_("Checked out %d path out of %s",
+					      "Checked out %d paths out of %s",
+					      nr_checkouts),
+				   nr_checkouts,
+				   find_unique_abbrev(&opts->source_tree->object.oid,
+						      DEFAULT_ABBREV));
+		else
+			fprintf_ln(stderr, Q_("Checked out %d path out of the index",
+					      "Checked out %d paths out of the index",
+					      nr_checkouts),
+				   nr_checkouts);
+	}
 
 	if (write_locked_index(&the_index, &lock_file, COMMIT_LOCK))
 		die(_("unable to write new index file"));
@@ -1064,6 +1086,7 @@ static int parse_branchname_arg(int argc, const char **argv,
 		has_dash_dash = 1; /* case (3) or (1) */
 	else if (dash_dash_pos >= 2)
 		die(_("only one reference expected, %d given."), dash_dash_pos);
+	opts->count_checkout_paths = !opts->quiet && !has_dash_dash;
 
 	if (!strcmp(arg, "-"))
 		arg = "@{-1}";
diff --git a/builtin/difftool.c b/builtin/difftool.c
index 544b0e8639..71318c26e1 100644
--- a/builtin/difftool.c
+++ b/builtin/difftool.c
@@ -323,7 +323,7 @@ static int checkout_path(unsigned mode, struct object_id *oid,
 	int ret;
 
 	ce = make_transient_cache_entry(mode, oid, path, 0);
-	ret = checkout_entry(ce, state, NULL);
+	ret = checkout_entry(ce, state, NULL, NULL);
 
 	discard_cache_entry(ce);
 	return ret;
diff --git a/cache.h b/cache.h
index 8b1ee42ae9..52fb6ba148 100644
--- a/cache.h
+++ b/cache.h
@@ -1540,9 +1540,9 @@ struct checkout {
 #define CHECKOUT_INIT { NULL, "" }
 
 #define TEMPORARY_FILENAME_LENGTH 25
-extern int checkout_entry(struct cache_entry *ce, const struct checkout *state, char *topath);
+extern int checkout_entry(struct cache_entry *ce, const struct checkout *state, char *topath, int *nr_checkouts);
 extern void enable_delayed_checkout(struct checkout *state);
-extern int finish_delayed_checkout(struct checkout *state);
+extern int finish_delayed_checkout(struct checkout *state, int *nr_checkouts);
 
 struct cache_def {
 	struct strbuf path;
diff --git a/entry.c b/entry.c
index 5d136c5d55..5f213c30fe 100644
--- a/entry.c
+++ b/entry.c
@@ -161,7 +161,7 @@ static int remove_available_paths(struct string_list_item *item, void *cb_data)
 	return !available;
 }
 
-int finish_delayed_checkout(struct checkout *state)
+int finish_delayed_checkout(struct checkout *state, int *nr_checkouts)
 {
 	int errs = 0;
 	unsigned delayed_object_count;
@@ -226,7 +226,7 @@ int finish_delayed_checkout(struct checkout *state)
 				ce = index_file_exists(state->istate, path->string,
 						       strlen(path->string), 0);
 				if (ce) {
-					errs |= checkout_entry(ce, state, NULL);
+					errs |= checkout_entry(ce, state, NULL, nr_checkouts);
 					filtered_bytes += ce->ce_stat_data.sd_size;
 					display_throughput(progress, filtered_bytes);
 				} else
@@ -435,8 +435,8 @@ static void mark_colliding_entries(const struct checkout *state,
  * its name is returned in topath[], which must be able to hold at
  * least TEMPORARY_FILENAME_LENGTH bytes long.
  */
-int checkout_entry(struct cache_entry *ce,
-		   const struct checkout *state, char *topath)
+int checkout_entry(struct cache_entry *ce, const struct checkout *state,
+		   char *topath, int *nr_checkouts)
 {
 	static struct strbuf path = STRBUF_INIT;
 	struct stat st;
@@ -506,5 +506,7 @@ int checkout_entry(struct cache_entry *ce,
 		return 0;
 
 	create_directories(path.buf, path.len, state);
+	if (nr_checkouts)
+		(*nr_checkouts)++;
 	return write_entry(ce, path.buf, state, 0);
 }
diff --git a/unpack-trees.c b/unpack-trees.c
index 7570df481b..17f1e601da 100644
--- a/unpack-trees.c
+++ b/unpack-trees.c
@@ -294,7 +294,7 @@ static void load_gitmodules_file(struct index_state *index,
 			repo_read_gitmodules(the_repository);
 		} else if (state && (ce->ce_flags & CE_UPDATE)) {
 			submodule_free(the_repository);
-			checkout_entry(ce, state, NULL);
+			checkout_entry(ce, state, NULL, NULL);
 			repo_read_gitmodules(the_repository);
 		}
 	}
@@ -450,12 +450,12 @@ static int check_updates(struct unpack_trees_options *o)
 			display_progress(progress, ++cnt);
 			ce->ce_flags &= ~CE_UPDATE;
 			if (o->update && !o->dry_run) {
-				errs |= checkout_entry(ce, &state, NULL);
+				errs |= checkout_entry(ce, &state, NULL, NULL);
 			}
 		}
 	}
 	stop_progress(&progress);
-	errs |= finish_delayed_checkout(&state);
+	errs |= finish_delayed_checkout(&state, NULL);
 	if (o->update)
 		git_attr_set_direction(GIT_ATTR_CHECKIN);
 
-- 
2.19.1.1318.g5295c6727d

Re: [PATCH v2] checkout: print something when checking out paths

[Junio C Hamano]

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> One of the problems with "git checkout" is that it does so many
> different things and could confuse people specially when we fail to
> handle ambiguation correctly.

You would have realized that this is way too noisy if you ran "make
test", which may have spewed something like this on the screen.

[19:09:19] t4120-apply-popt.sh ................................ ok     1624 ms ( 0.26 usr  0.21 sys +  5.31 cusr  3.51 csys =  9.29 CPU)
[19:09:20] t9164-git-svn-dcommit-concurrent.sh ................ skipped: Perl SVN libraries not found or unusable
[19:09:20] t1310-config-default.sh ............................ ok      177 ms ( 0.07 usr  0.01 sys +  0.89 cusr  0.66 csys =  1.63 CPU)
===(   20175;154  1297/?  155/?  6/?  3/3  2/?  4/?  4/?  3/?  5... )===Checked out 1 path out of the index
Checked out 1 path out of the index
Checked out 1 path out of the index
Checked out 1 path out of the index
Checked out 1 path out of the index
[19:09:20] t1408-packed-refs.sh ............................... ok      310 ms ( 0.06 usr  0.00 sys +  0.69 cusr  0.52 csys =  1.27 CPU)
[19:09:20] t0025-crlf-renormalize.sh .......................... ok      246 ms ( 0.03 usr  0.01 sys +  0.34 cusr  0.22 csys =  0.60 CPU)

I am very tempted to suggest to treat this as a training wheel and
enable only when checkout.showpathcount is set to true, or something
like that.

Re: [PATCH v2] checkout: print something when checking out paths

[Duy Nguyen]

On Wed, Nov 14, 2018 at 11:12 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>
> > One of the problems with "git checkout" is that it does so many
> > different things and could confuse people specially when we fail to
> > handle ambiguation correctly.
>
> You would have realized that this is way too noisy if you ran "make
> test", which may have spewed something like this on the screen.

Oh I realize it because it's part of my git build and I often use "git
co <paths>". I'm just telling (or kidding?) myself that I'm just so
used to the old behavior and may need some time to feel comfortable
with the new one.

> [19:09:19] t4120-apply-popt.sh ................................ ok     1624 ms ( 0.26 usr  0.21 sys +  5.31 cusr  3.51 csys =  9.29 CPU)
> [19:09:20] t9164-git-svn-dcommit-concurrent.sh ................ skipped: Perl SVN libraries not found or unusable
> [19:09:20] t1310-config-default.sh ............................ ok      177 ms ( 0.07 usr  0.01 sys +  0.89 cusr  0.66 csys =  1.63 CPU)
> ===(   20175;154  1297/?  155/?  6/?  3/3  2/?  4/?  4/?  3/?  5... )===Checked out 1 path out of the index
> Checked out 1 path out of the index
> Checked out 1 path out of the index
> Checked out 1 path out of the index
> Checked out 1 path out of the index
> [19:09:20] t1408-packed-refs.sh ............................... ok      310 ms ( 0.06 usr  0.00 sys +  0.69 cusr  0.52 csys =  1.27 CPU)
> [19:09:20] t0025-crlf-renormalize.sh .......................... ok      246 ms ( 0.03 usr  0.01 sys +  0.34 cusr  0.22 csys =  0.60 CPU)
>
> I am very tempted to suggest to treat this as a training wheel and
> enable only when checkout.showpathcount is set to true, or something
> like that.

Maybe we just drop it then. I'm not adding a training wheel. I'm
trying to make this complex command safer somewhat. But maybe this is
a wrong direction. I'll give the idea "switch-branch / restore-path
alternative commands" a go some time. Then the new generation can just
stick to those and old timers stay with "git checkout".
-- 
Duy

Re: [PATCH/RFC] checkout: print something when checking out paths

[Ævar Arnfjörð Bjarmason]

On Mon, Nov 12 2018, Duy Nguyen wrote:

> On Mon, Nov 12, 2018 at 7:21 AM Junio C Hamano <gitster@pobox.com> wrote:
>>
>> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>>
>> > Since the purpose of printing this is to help disambiguate. Only do it
>> > when "--" is missing (the actual reason though is many tests check
>> > empty stderr to see that no error is raised and I'm too lazy to fix
>> > all the test cases).
>>
>> Heh, honesty is good but in this particular case the official reason
>> alone would make perfect sense, too ;-)
>>
>> As with progress output, shouldn't this automatically be turned off
>> when the standard error stream goes to non tty, as the real purpose
>> of printing is to help the user sitting in front of the terminal and
>> interacting with the command?
>
> I see this at the same level as "Switched to branch 'foo'" which is
> also protected by opts->quiet. If we start hiding messages based on
> tty it should be done for other messages in update_refs_for_switch()
> too, I guess?

I have no real opinion either way, but whatever we can do about
"checkout" being so confusing since it does so many things is most
welcome.

Just an alternative: Maybe we can start this out as advice() output
that's either opt-in via config (not on by default) to start with, or
have some advice_tty() that only emits it in the same circumstances we
emit the progress output?

Re: [PATCH/RFC] checkout: print something when checking out paths

[Duy Nguyen]

On Mon, Nov 19, 2018 at 2:08 PM Ævar Arnfjörð Bjarmason
<avarab@gmail.com> wrote:
>
>
> On Mon, Nov 12 2018, Duy Nguyen wrote:
>
> > On Mon, Nov 12, 2018 at 7:21 AM Junio C Hamano <gitster@pobox.com> wrote:
> >>
> >> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
> >>
> >> > Since the purpose of printing this is to help disambiguate. Only do it
> >> > when "--" is missing (the actual reason though is many tests check
> >> > empty stderr to see that no error is raised and I'm too lazy to fix
> >> > all the test cases).
> >>
> >> Heh, honesty is good but in this particular case the official reason
> >> alone would make perfect sense, too ;-)
> >>
> >> As with progress output, shouldn't this automatically be turned off
> >> when the standard error stream goes to non tty, as the real purpose
> >> of printing is to help the user sitting in front of the terminal and
> >> interacting with the command?
> >
> > I see this at the same level as "Switched to branch 'foo'" which is
> > also protected by opts->quiet. If we start hiding messages based on
> > tty it should be done for other messages in update_refs_for_switch()
> > too, I guess?
>
> I have no real opinion either way, but whatever we can do about
> "checkout" being so confusing since it does so many things is most
> welcome.
>
> Just an alternative: Maybe we can start this out as advice() output
> that's either opt-in via config (not on by default) to start with, or
> have some advice_tty() that only emits it in the same circumstances we
> emit the progress output?

No let's drop this for now. I promise to come back with something
better (at least it still sounds better in my mind). If that idea does
not work out, we can come back and see if we can improve this.
-- 
Duy

Re: [PATCH/RFC] checkout: print something when checking out paths

[Junio C Hamano]

Duy Nguyen <pclouds@gmail.com> writes:

>> > I see this at the same level as "Switched to branch 'foo'" which is
>> > also protected by opts->quiet. If we start hiding messages based on
>> > tty it should be done for other messages in update_refs_for_switch()
>> > too, I guess?
>
> No let's drop this for now. I promise to come back with something
> better (at least it still sounds better in my mind). If that idea does
> not work out, we can come back and see if we can improve this.

Let's leave it in 'pu'.

I do agree that this is similar to existing messages that talk about
checkout out a branch to work on etc., and I think giving feedback
when checkout paths out _is_ a good thing to do for interactive
users.

If we were to squelch such "notice" output for non-interactive use,
we should do so to the "notice" messages for checking out a branch,
as well, and also to other subcommands that report what they did
with these "notice" output.  And that is a separate topic.

The primary reason why I was annoyed was because "make test" (I
think I have DEFAULT_TEST_TARGET=prove, if it matters) output was
littered with these "checked out N paths", even though I am not
asking for verbose output.  

It could be that the real cause of that is perhaps because we have
too many "git checkout" that is outside test_expect_* block, in
which case we should fix these tests to perform the checkout inside
a test_expect_success block for test set-up, and my annoyance was
only shooting at the messenger.

For example, the attached patch illustrates the right problem but
addresses it in a wrong way.  This checkout_files() helper does too
many things outside (and before) the test_expect_success block it
has (other helpers like commit_chk_wrnNNO share the same problem),
and making "git checkout" noisy will reveal that as a problem by
leaking the noisy output directly to the tester.  But the real fix
is to enclose the set-up step inside a test_expect_success block,
which is not done by the following illustration patch, which instead
just squelches the message.

 t/t0027-auto-crlf.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/t/t0027-auto-crlf.sh b/t/t0027-auto-crlf.sh
index beb5927f77..3587e454f1 100755
--- a/t/t0027-auto-crlf.sh
+++ b/t/t0027-auto-crlf.sh
@@ -293,9 +293,9 @@ checkout_files () {
 	do
 		rm crlf_false_attr__$f.txt &&
 		if test -z "$ceol"; then
-			git checkout crlf_false_attr__$f.txt
+			git checkout -- crlf_false_attr__$f.txt
 		else
-			git -c core.eol=$ceol checkout crlf_false_attr__$f.txt
+			git -c core.eol=$ceol checkout -- crlf_false_attr__$f.txt
 		fi
 	done
 

[RFC] Introduce two new commands, switch-branch and restore-paths

[Duy Nguyen]

On Mon, Nov 19, 2018 at 04:19:53PM +0100, Duy Nguyen wrote:
> I promise to come back with something better (at least it still
> sounds better in my mind). If that idea does not work out, we can
> come back and see if we can improve this.

So this is it. The patch isn't pretty, mostly as a proof of
concept. Just look at the three functions at the bottom of checkout.c,
which is the main thing.

This patch tries to split "git checkout" command in two new ones:

- git switch-branch is all about switching branches
- git restore-paths (maybe restore-file is better) for checking out
  paths

The main idea is these two commands will co-exist with the good old
'git checkout', which will NOT be deprecated. Old timers will still
use "git checkout". But new people should be introduced to the new two
instead. And the new ones are just as capable as "git checkout".

Since the three commands will co-exist (with duplicate functionality),
maintenance cost must be kept to minimum. The way I did this is simply
split the command line options into three pieces: common,
switch-branch and checkout-paths. "git checkout" has all three, the
other two have common and another piece.

With this, a new option added to git checkout will be automatically
available in either switch-branch or checkout-paths. Bug fixes apply
to all relevant commands.

Later on, we could start to add a bit more stuff in, e.g. some form of
disambiguation is no longer needed when running as switch-branch, or
restore-paths.

So, what do you think?

-- 8< --
diff --git a/builtin.h b/builtin.h
index 6538932e99..6e321ec8a4 100644
--- a/builtin.h
+++ b/builtin.h
@@ -214,6 +214,7 @@ extern int cmd_remote_fd(int argc, const char **argv, const char *prefix);
 extern int cmd_repack(int argc, const char **argv, const char *prefix);
 extern int cmd_rerere(int argc, const char **argv, const char *prefix);
 extern int cmd_reset(int argc, const char **argv, const char *prefix);
+extern int cmd_restore_paths(int argc, const char **argv, const char *prefix);
 extern int cmd_rev_list(int argc, const char **argv, const char *prefix);
 extern int cmd_rev_parse(int argc, const char **argv, const char *prefix);
 extern int cmd_revert(int argc, const char **argv, const char *prefix);
@@ -227,6 +228,7 @@ extern int cmd_show_index(int argc, const char **argv, const char *prefix);
 extern int cmd_status(int argc, const char **argv, const char *prefix);
 extern int cmd_stripspace(int argc, const char **argv, const char *prefix);
 extern int cmd_submodule__helper(int argc, const char **argv, const char *prefix);
+extern int cmd_switch_branch(int argc, const char **argv, const char *prefix);
 extern int cmd_symbolic_ref(int argc, const char **argv, const char *prefix);
 extern int cmd_tag(int argc, const char **argv, const char *prefix);
 extern int cmd_tar_tree(int argc, const char **argv, const char *prefix);
diff --git a/builtin/checkout.c b/builtin/checkout.c
index acdafc6e4c..868ca3c223 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -33,6 +33,16 @@ static const char * const checkout_usage[] = {
 	NULL,
 };
 
+static const char * const switch_branch_usage[] = {
+	N_("git switch-branch [<options>] <branch>"),
+	NULL,
+};
+
+static const char * const restore_paths_usage[] = {
+	N_("git restore-paths [<options>] [<branch>] -- <file>..."),
+	NULL,
+};
+
 struct checkout_opts {
 	int patch_mode;
 	int quiet;
@@ -44,6 +54,7 @@ struct checkout_opts {
 	int ignore_skipworktree;
 	int ignore_other_worktrees;
 	int show_progress;
+	int dwim_new_local_branch;
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
 	 * should be updated accordingly.
@@ -55,6 +66,7 @@ struct checkout_opts {
 	int new_branch_log;
 	enum branch_track track;
 	struct diff_options diff_options;
+	char *conflict_style;
 
 	int branch_exists;
 	const char *prefix;
@@ -1223,78 +1235,105 @@ static int checkout_branch(struct checkout_opts *opts,
 	return switch_branches(opts, new_branch_info);
 }
 
-int cmd_checkout(int argc, const char **argv, const char *prefix)
+static struct option *add_common_options(struct checkout_opts *opts,
+					 struct option *prevopts)
 {
-	struct checkout_opts opts;
-	struct branch_info new_branch_info;
-	char *conflict_style = NULL;
-	int dwim_new_local_branch = 1;
-	int dwim_remotes_matched = 0;
 	struct option options[] = {
-		OPT__QUIET(&opts.quiet, N_("suppress progress reporting")),
-		OPT_STRING('b', NULL, &opts.new_branch, N_("branch"),
+		OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
+		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
+			 N_("do not limit pathspecs to sparse entries only")),
+		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
+			    "checkout", "control recursive updating of submodules",
+			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
+		OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
+		OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
+			   PARSE_OPT_NOCOMPLETE),
+		OPT_STRING(0, "conflict", &opts->conflict_style, N_("style"),
+			   N_("conflict style (merge or diff3)")),
+		OPT_END()
+	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
+
+static struct option *add_switch_branch_options(struct checkout_opts *opts,
+						struct option *prevopts)
+{
+	struct option options[] = {
+		OPT_STRING('b', NULL, &opts->new_branch, N_("branch"),
 			   N_("create and checkout a new branch")),
-		OPT_STRING('B', NULL, &opts.new_branch_force, N_("branch"),
+		OPT_STRING('B', NULL, &opts->new_branch_force, N_("branch"),
 			   N_("create/reset and checkout a branch")),
-		OPT_BOOL('l', NULL, &opts.new_branch_log, N_("create reflog for new branch")),
-		OPT_BOOL(0, "detach", &opts.force_detach, N_("detach HEAD at named commit")),
-		OPT_SET_INT('t', "track",  &opts.track, N_("set upstream info for new branch"),
+		OPT_BOOL('l', NULL, &opts->new_branch_log, N_("create reflog for new branch")),
+		OPT_BOOL(0, "detach", &opts->force_detach, N_("detach HEAD at named commit")),
+		OPT_SET_INT('t', "track",  &opts->track, N_("set upstream info for new branch"),
 			BRANCH_TRACK_EXPLICIT),
-		OPT_STRING(0, "orphan", &opts.new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
-		OPT_SET_INT_F('2', "ours", &opts.writeout_stage,
+		OPT_STRING(0, "orphan", &opts->new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
+		OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
+		OPT_HIDDEN_BOOL(0, "guess", &opts->dwim_new_local_branch,
+				N_("second guess 'git checkout <no-such-branch>'")),
+		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
+			 N_("do not check if another worktree is holding the given ref")),
+		OPT_END()
+	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
+
+static struct option *add_checkout_path_options(struct checkout_opts *opts,
+						struct option *prevopts)
+{
+	struct option options[] = {
+		OPT_SET_INT_F('2', "ours", &opts->writeout_stage,
 			      N_("checkout our version for unmerged files"),
 			      2, PARSE_OPT_NONEG),
-		OPT_SET_INT_F('3', "theirs", &opts.writeout_stage,
+		OPT_SET_INT_F('3', "theirs", &opts->writeout_stage,
 			      N_("checkout their version for unmerged files"),
 			      3, PARSE_OPT_NONEG),
-		OPT__FORCE(&opts.force, N_("force checkout (throw away local modifications)"),
-			   PARSE_OPT_NOCOMPLETE),
-		OPT_BOOL('m', "merge", &opts.merge, N_("perform a 3-way merge with the new branch")),
-		OPT_BOOL_F(0, "overwrite-ignore", &opts.overwrite_ignore,
-			   N_("update ignored files (default)"),
-			   PARSE_OPT_NOCOMPLETE),
-		OPT_STRING(0, "conflict", &conflict_style, N_("style"),
-			   N_("conflict style (merge or diff3)")),
-		OPT_BOOL('p', "patch", &opts.patch_mode, N_("select hunks interactively")),
-		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts.ignore_skipworktree,
-			 N_("do not limit pathspecs to sparse entries only")),
-		OPT_HIDDEN_BOOL(0, "guess", &dwim_new_local_branch,
-				N_("second guess 'git checkout <no-such-branch>'")),
-		OPT_BOOL(0, "ignore-other-worktrees", &opts.ignore_other_worktrees,
-			 N_("do not check if another worktree is holding the given ref")),
-		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
-			    "checkout", "control recursive updating of submodules",
-			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
-		OPT_BOOL(0, "progress", &opts.show_progress, N_("force progress reporting")),
-		OPT_END(),
+		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
+		OPT_END()
 	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
 
-	memset(&opts, 0, sizeof(opts));
+static int checkout_main(int argc, const char **argv, const char *prefix,
+			 struct checkout_opts *opts, struct option *options,
+			 const char * const usagestr[])
+{
+	struct branch_info new_branch_info;
+	int dwim_remotes_matched = 0;
+
+	memset(opts, 0, sizeof(*opts));
+	opts->dwim_new_local_branch = 1;
 	memset(&new_branch_info, 0, sizeof(new_branch_info));
-	opts.overwrite_ignore = 1;
-	opts.prefix = prefix;
-	opts.show_progress = -1;
+	opts->overwrite_ignore = 1;
+	opts->prefix = prefix;
+	opts->show_progress = -1;
 
-	git_config(git_checkout_config, &opts);
+	git_config(git_checkout_config, opts);
 
-	opts.track = BRANCH_TRACK_UNSPECIFIED;
+	opts->track = BRANCH_TRACK_UNSPECIFIED;
 
-	argc = parse_options(argc, argv, prefix, options, checkout_usage,
+	argc = parse_options(argc, argv, prefix, options, usagestr,
 			     PARSE_OPT_KEEP_DASHDASH);
 
-	if (opts.show_progress < 0) {
-		if (opts.quiet)
-			opts.show_progress = 0;
+	if (opts->show_progress < 0) {
+		if (opts->quiet)
+			opts->show_progress = 0;
 		else
-			opts.show_progress = isatty(2);
+			opts->show_progress = isatty(2);
 	}
 
-	if (conflict_style) {
-		opts.merge = 1; /* implied */
-		git_xmerge_config("merge.conflictstyle", conflict_style, NULL);
+	if (opts->conflict_style) {
+		opts->merge = 1; /* implied */
+		git_xmerge_config("merge.conflictstyle", opts->conflict_style, NULL);
 	}
 
-	if ((!!opts.new_branch + !!opts.new_branch_force + !!opts.new_orphan_branch) > 1)
+	if ((!!opts->new_branch + !!opts->new_branch_force + !!opts->new_orphan_branch) > 1)
 		die(_("-b, -B and --orphan are mutually exclusive"));
 
 	/*
@@ -1302,14 +1341,14 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	 * and new_branch_force and new_orphan_branch will tell us which one of
 	 * -b/-B/--orphan is being used.
 	 */
-	if (opts.new_branch_force)
-		opts.new_branch = opts.new_branch_force;
+	if (opts->new_branch_force)
+		opts->new_branch = opts->new_branch_force;
 
-	if (opts.new_orphan_branch)
-		opts.new_branch = opts.new_orphan_branch;
+	if (opts->new_orphan_branch)
+		opts->new_branch = opts->new_orphan_branch;
 
 	/* --track without -b/-B/--orphan should DWIM */
-	if (opts.track != BRANCH_TRACK_UNSPECIFIED && !opts.new_branch) {
+	if (opts->track != BRANCH_TRACK_UNSPECIFIED && !opts->new_branch) {
 		const char *argv0 = argv[0];
 		if (!argc || !strcmp(argv0, "--"))
 			die(_("--track needs a branch name"));
@@ -1318,7 +1357,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		argv0 = strchr(argv0, '/');
 		if (!argv0 || !argv0[1])
 			die(_("missing branch name; try -b"));
-		opts.new_branch = argv0 + 1;
+		opts->new_branch = argv0 + 1;
 	}
 
 	/*
@@ -1337,56 +1376,56 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	if (argc) {
 		struct object_id rev;
 		int dwim_ok =
-			!opts.patch_mode &&
-			dwim_new_local_branch &&
-			opts.track == BRANCH_TRACK_UNSPECIFIED &&
-			!opts.new_branch;
+			!opts->patch_mode &&
+			opts->dwim_new_local_branch &&
+			opts->track == BRANCH_TRACK_UNSPECIFIED &&
+			!opts->new_branch;
 		int n = parse_branchname_arg(argc, argv, dwim_ok,
-					     &new_branch_info, &opts, &rev,
+					     &new_branch_info, opts, &rev,
 					     &dwim_remotes_matched);
 		argv += n;
 		argc -= n;
 	}
 
 	if (argc) {
-		parse_pathspec(&opts.pathspec, 0,
-			       opts.patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
+		parse_pathspec(&opts->pathspec, 0,
+			       opts->patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
 			       prefix, argv);
 
-		if (!opts.pathspec.nr)
+		if (!opts->pathspec.nr)
 			die(_("invalid path specification"));
 
 		/*
 		 * Try to give more helpful suggestion.
 		 * new_branch && argc > 1 will be caught later.
 		 */
-		if (opts.new_branch && argc == 1)
+		if (opts->new_branch && argc == 1)
 			die(_("'%s' is not a commit and a branch '%s' cannot be created from it"),
-				argv[0], opts.new_branch);
+				argv[0], opts->new_branch);
 
-		if (opts.force_detach)
+		if (opts->force_detach)
 			die(_("git checkout: --detach does not take a path argument '%s'"),
 			    argv[0]);
 
-		if (1 < !!opts.writeout_stage + !!opts.force + !!opts.merge)
+		if (1 < !!opts->writeout_stage + !!opts->force + !!opts->merge)
 			die(_("git checkout: --ours/--theirs, --force and --merge are incompatible when\n"
 			      "checking out of the index."));
 	}
 
-	if (opts.new_branch) {
+	if (opts->new_branch) {
 		struct strbuf buf = STRBUF_INIT;
 
-		if (opts.new_branch_force)
-			opts.branch_exists = validate_branchname(opts.new_branch, &buf);
+		if (opts->new_branch_force)
+			opts->branch_exists = validate_branchname(opts->new_branch, &buf);
 		else
-			opts.branch_exists =
-				validate_new_branchname(opts.new_branch, &buf, 0);
+			opts->branch_exists =
+				validate_new_branchname(opts->new_branch, &buf, 0);
 		strbuf_release(&buf);
 	}
 
 	UNLEAK(opts);
-	if (opts.patch_mode || opts.pathspec.nr) {
-		int ret = checkout_paths(&opts, new_branch_info.name);
+	if (opts->patch_mode || opts->pathspec.nr) {
+		int ret = checkout_paths(opts, new_branch_info.name);
 		if (ret && dwim_remotes_matched > 1 &&
 		    advice_checkout_ambiguous_remote_branch_name)
 			advise(_("'%s' matched more than one remote tracking branch.\n"
@@ -1405,6 +1444,49 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 			       dwim_remotes_matched);
 		return ret;
 	} else {
-		return checkout_branch(&opts, &new_branch_info);
+		return checkout_branch(opts, &new_branch_info);
 	}
 }
+
+int cmd_checkout(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	options = add_common_options(&opts, options);
+	options = add_switch_branch_options(&opts, options);
+	options = add_checkout_path_options(&opts, options);
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, checkout_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
+
+int cmd_switch_branch(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	options = add_common_options(&opts, options);
+	options = add_switch_branch_options(&opts, options);
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, switch_branch_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
+
+int cmd_restore_paths(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	options = add_common_options(&opts, options);
+	options = add_checkout_path_options(&opts, options);
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, restore_paths_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
diff --git a/git.c b/git.c
index 2f604a41ea..e8a76a99da 100644
--- a/git.c
+++ b/git.c
@@ -542,6 +542,7 @@ static struct cmd_struct commands[] = {
 	{ "replace", cmd_replace, RUN_SETUP },
 	{ "rerere", cmd_rerere, RUN_SETUP },
 	{ "reset", cmd_reset, RUN_SETUP },
+	{ "restore-paths", cmd_restore_paths, RUN_SETUP | NEED_WORK_TREE },
 	{ "rev-list", cmd_rev_list, RUN_SETUP | NO_PARSEOPT },
 	{ "rev-parse", cmd_rev_parse, NO_PARSEOPT },
 	{ "revert", cmd_revert, RUN_SETUP | NEED_WORK_TREE },
@@ -557,6 +558,7 @@ static struct cmd_struct commands[] = {
 	{ "status", cmd_status, RUN_SETUP | NEED_WORK_TREE },
 	{ "stripspace", cmd_stripspace },
 	{ "submodule--helper", cmd_submodule__helper, RUN_SETUP | SUPPORT_SUPER_PREFIX | NO_PARSEOPT },
+	{ "switch-branch", cmd_switch_branch, RUN_SETUP | NEED_WORK_TREE },
 	{ "symbolic-ref", cmd_symbolic_ref, RUN_SETUP },
 	{ "tag", cmd_tag, RUN_SETUP | DELAY_PAGER_CONFIG },
 	{ "unpack-file", cmd_unpack_file, RUN_SETUP | NO_PARSEOPT },
diff --git a/parse-options-cb.c b/parse-options-cb.c
index 8c9edce52f..c609d52926 100644
--- a/parse-options-cb.c
+++ b/parse-options-cb.c
@@ -126,7 +126,7 @@ struct option *parse_options_concat(struct option *a, struct option *b)
 	struct option *ret;
 	size_t i, a_len = 0, b_len = 0;
 
-	for (i = 0; a[i].type != OPTION_END; i++)
+	for (i = 0; a && a[i].type != OPTION_END; i++)
 		a_len++;
 	for (i = 0; b[i].type != OPTION_END; i++)
 		b_len++;
-- 8< --

Re: [RFC] Introduce two new commands, switch-branch and restore-paths

[Thomas Gummerer]

On 11/20, Duy Nguyen wrote:
> On Mon, Nov 19, 2018 at 04:19:53PM +0100, Duy Nguyen wrote:
> > I promise to come back with something better (at least it still
> > sounds better in my mind). If that idea does not work out, we can
> > come back and see if we can improve this.
> 
> So this is it. The patch isn't pretty, mostly as a proof of
> concept. Just look at the three functions at the bottom of checkout.c,
> which is the main thing.
> 
> This patch tries to split "git checkout" command in two new ones:
> 
> - git switch-branch is all about switching branches
> - git restore-paths (maybe restore-file is better) for checking out
>   paths
> 
> The main idea is these two commands will co-exist with the good old
> 'git checkout', which will NOT be deprecated. Old timers will still
> use "git checkout". But new people should be introduced to the new two
> instead. And the new ones are just as capable as "git checkout".
> 
> Since the three commands will co-exist (with duplicate functionality),
> maintenance cost must be kept to minimum. The way I did this is simply
> split the command line options into three pieces: common,
> switch-branch and checkout-paths. "git checkout" has all three, the
> other two have common and another piece.
>
> With this, a new option added to git checkout will be automatically
> available in either switch-branch or checkout-paths. Bug fixes apply
> to all relevant commands.
> 
> Later on, we could start to add a bit more stuff in, e.g. some form of
> disambiguation is no longer needed when running as switch-branch, or
> restore-paths.
> 
> So, what do you think?

I like the idea of splitting those commands up, in fact it is
something I've been considering working on myself.  I do think we
should consider if we want to change the behaviour of those new
commands in any way compared to 'git checkout', since we're starting
with a clean slate.

One thing in particular that I have in mind is something I'm currently
working on, namely adding a --index flag to 'git checkout', which
would make 'git checkout' work in non-overlay mode (for more
discussion on that see also [*1*].  I got something working, that
needs to be polished a bit and am hoping to send that to the list
sometime soon.

I wonder if such the --index behaviour could be the default in
restore-paths command?

Most of the underlying machinery for 'checkout' could and should of
course still be shared between the commands.

*1*: <xmqq4loqplou.fsf@gitster.mtv.corp.google.com>

> -- 8< --
> diff --git a/builtin.h b/builtin.h
> index 6538932e99..6e321ec8a4 100644
> --- a/builtin.h
> +++ b/builtin.h
> @@ -214,6 +214,7 @@ extern int cmd_remote_fd(int argc, const char **argv, const char *prefix);
>  extern int cmd_repack(int argc, const char **argv, const char *prefix);
>  extern int cmd_rerere(int argc, const char **argv, const char *prefix);
>  extern int cmd_reset(int argc, const char **argv, const char *prefix);
> +extern int cmd_restore_paths(int argc, const char **argv, const char *prefix);
>  extern int cmd_rev_list(int argc, const char **argv, const char *prefix);
>  extern int cmd_rev_parse(int argc, const char **argv, const char *prefix);
>  extern int cmd_revert(int argc, const char **argv, const char *prefix);
> @@ -227,6 +228,7 @@ extern int cmd_show_index(int argc, const char **argv, const char *prefix);
>  extern int cmd_status(int argc, const char **argv, const char *prefix);
>  extern int cmd_stripspace(int argc, const char **argv, const char *prefix);
>  extern int cmd_submodule__helper(int argc, const char **argv, const char *prefix);
> +extern int cmd_switch_branch(int argc, const char **argv, const char *prefix);
>  extern int cmd_symbolic_ref(int argc, const char **argv, const char *prefix);
>  extern int cmd_tag(int argc, const char **argv, const char *prefix);
>  extern int cmd_tar_tree(int argc, const char **argv, const char *prefix);
> diff --git a/builtin/checkout.c b/builtin/checkout.c
> index acdafc6e4c..868ca3c223 100644
> --- a/builtin/checkout.c
> +++ b/builtin/checkout.c
> @@ -33,6 +33,16 @@ static const char * const checkout_usage[] = {
>  	NULL,
>  };
>  
> +static const char * const switch_branch_usage[] = {
> +	N_("git switch-branch [<options>] <branch>"),
> +	NULL,
> +};
> +
> +static const char * const restore_paths_usage[] = {
> +	N_("git restore-paths [<options>] [<branch>] -- <file>..."),
> +	NULL,
> +};
> +
>  struct checkout_opts {
>  	int patch_mode;
>  	int quiet;
> @@ -44,6 +54,7 @@ struct checkout_opts {
>  	int ignore_skipworktree;
>  	int ignore_other_worktrees;
>  	int show_progress;
> +	int dwim_new_local_branch;
>  	/*
>  	 * If new checkout options are added, skip_merge_working_tree
>  	 * should be updated accordingly.
> @@ -55,6 +66,7 @@ struct checkout_opts {
>  	int new_branch_log;
>  	enum branch_track track;
>  	struct diff_options diff_options;
> +	char *conflict_style;
>  
>  	int branch_exists;
>  	const char *prefix;
> @@ -1223,78 +1235,105 @@ static int checkout_branch(struct checkout_opts *opts,
>  	return switch_branches(opts, new_branch_info);
>  }
>  
> -int cmd_checkout(int argc, const char **argv, const char *prefix)
> +static struct option *add_common_options(struct checkout_opts *opts,
> +					 struct option *prevopts)
>  {
> -	struct checkout_opts opts;
> -	struct branch_info new_branch_info;
> -	char *conflict_style = NULL;
> -	int dwim_new_local_branch = 1;
> -	int dwim_remotes_matched = 0;
>  	struct option options[] = {
> -		OPT__QUIET(&opts.quiet, N_("suppress progress reporting")),
> -		OPT_STRING('b', NULL, &opts.new_branch, N_("branch"),
> +		OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
> +		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
> +			 N_("do not limit pathspecs to sparse entries only")),
> +		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
> +			    "checkout", "control recursive updating of submodules",
> +			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
> +		OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
> +		OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
> +			   PARSE_OPT_NOCOMPLETE),
> +		OPT_STRING(0, "conflict", &opts->conflict_style, N_("style"),
> +			   N_("conflict style (merge or diff3)")),
> +		OPT_END()
> +	};
> +	struct option *newopts = parse_options_concat(prevopts, options);
> +	free(prevopts);
> +	return newopts;
> +}
> +
> +static struct option *add_switch_branch_options(struct checkout_opts *opts,
> +						struct option *prevopts)
> +{
> +	struct option options[] = {
> +		OPT_STRING('b', NULL, &opts->new_branch, N_("branch"),
>  			   N_("create and checkout a new branch")),
> -		OPT_STRING('B', NULL, &opts.new_branch_force, N_("branch"),
> +		OPT_STRING('B', NULL, &opts->new_branch_force, N_("branch"),
>  			   N_("create/reset and checkout a branch")),
> -		OPT_BOOL('l', NULL, &opts.new_branch_log, N_("create reflog for new branch")),
> -		OPT_BOOL(0, "detach", &opts.force_detach, N_("detach HEAD at named commit")),
> -		OPT_SET_INT('t', "track",  &opts.track, N_("set upstream info for new branch"),
> +		OPT_BOOL('l', NULL, &opts->new_branch_log, N_("create reflog for new branch")),
> +		OPT_BOOL(0, "detach", &opts->force_detach, N_("detach HEAD at named commit")),
> +		OPT_SET_INT('t', "track",  &opts->track, N_("set upstream info for new branch"),
>  			BRANCH_TRACK_EXPLICIT),
> -		OPT_STRING(0, "orphan", &opts.new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
> -		OPT_SET_INT_F('2', "ours", &opts.writeout_stage,
> +		OPT_STRING(0, "orphan", &opts->new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
> +		OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
> +		OPT_HIDDEN_BOOL(0, "guess", &opts->dwim_new_local_branch,
> +				N_("second guess 'git checkout <no-such-branch>'")),
> +		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
> +			 N_("do not check if another worktree is holding the given ref")),
> +		OPT_END()
> +	};
> +	struct option *newopts = parse_options_concat(prevopts, options);
> +	free(prevopts);
> +	return newopts;
> +}
> +
> +static struct option *add_checkout_path_options(struct checkout_opts *opts,
> +						struct option *prevopts)
> +{
> +	struct option options[] = {
> +		OPT_SET_INT_F('2', "ours", &opts->writeout_stage,
>  			      N_("checkout our version for unmerged files"),
>  			      2, PARSE_OPT_NONEG),
> -		OPT_SET_INT_F('3', "theirs", &opts.writeout_stage,
> +		OPT_SET_INT_F('3', "theirs", &opts->writeout_stage,
>  			      N_("checkout their version for unmerged files"),
>  			      3, PARSE_OPT_NONEG),
> -		OPT__FORCE(&opts.force, N_("force checkout (throw away local modifications)"),
> -			   PARSE_OPT_NOCOMPLETE),
> -		OPT_BOOL('m', "merge", &opts.merge, N_("perform a 3-way merge with the new branch")),
> -		OPT_BOOL_F(0, "overwrite-ignore", &opts.overwrite_ignore,
> -			   N_("update ignored files (default)"),
> -			   PARSE_OPT_NOCOMPLETE),
> -		OPT_STRING(0, "conflict", &conflict_style, N_("style"),
> -			   N_("conflict style (merge or diff3)")),
> -		OPT_BOOL('p', "patch", &opts.patch_mode, N_("select hunks interactively")),
> -		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts.ignore_skipworktree,
> -			 N_("do not limit pathspecs to sparse entries only")),
> -		OPT_HIDDEN_BOOL(0, "guess", &dwim_new_local_branch,
> -				N_("second guess 'git checkout <no-such-branch>'")),
> -		OPT_BOOL(0, "ignore-other-worktrees", &opts.ignore_other_worktrees,
> -			 N_("do not check if another worktree is holding the given ref")),
> -		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
> -			    "checkout", "control recursive updating of submodules",
> -			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
> -		OPT_BOOL(0, "progress", &opts.show_progress, N_("force progress reporting")),
> -		OPT_END(),
> +		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
> +		OPT_END()
>  	};
> +	struct option *newopts = parse_options_concat(prevopts, options);
> +	free(prevopts);
> +	return newopts;
> +}
>  
> -	memset(&opts, 0, sizeof(opts));
> +static int checkout_main(int argc, const char **argv, const char *prefix,
> +			 struct checkout_opts *opts, struct option *options,
> +			 const char * const usagestr[])
> +{
> +	struct branch_info new_branch_info;
> +	int dwim_remotes_matched = 0;
> +
> +	memset(opts, 0, sizeof(*opts));
> +	opts->dwim_new_local_branch = 1;
>  	memset(&new_branch_info, 0, sizeof(new_branch_info));
> -	opts.overwrite_ignore = 1;
> -	opts.prefix = prefix;
> -	opts.show_progress = -1;
> +	opts->overwrite_ignore = 1;
> +	opts->prefix = prefix;
> +	opts->show_progress = -1;
>  
> -	git_config(git_checkout_config, &opts);
> +	git_config(git_checkout_config, opts);
>  
> -	opts.track = BRANCH_TRACK_UNSPECIFIED;
> +	opts->track = BRANCH_TRACK_UNSPECIFIED;
>  
> -	argc = parse_options(argc, argv, prefix, options, checkout_usage,
> +	argc = parse_options(argc, argv, prefix, options, usagestr,
>  			     PARSE_OPT_KEEP_DASHDASH);
>  
> -	if (opts.show_progress < 0) {
> -		if (opts.quiet)
> -			opts.show_progress = 0;
> +	if (opts->show_progress < 0) {
> +		if (opts->quiet)
> +			opts->show_progress = 0;
>  		else
> -			opts.show_progress = isatty(2);
> +			opts->show_progress = isatty(2);
>  	}
>  
> -	if (conflict_style) {
> -		opts.merge = 1; /* implied */
> -		git_xmerge_config("merge.conflictstyle", conflict_style, NULL);
> +	if (opts->conflict_style) {
> +		opts->merge = 1; /* implied */
> +		git_xmerge_config("merge.conflictstyle", opts->conflict_style, NULL);
>  	}
>  
> -	if ((!!opts.new_branch + !!opts.new_branch_force + !!opts.new_orphan_branch) > 1)
> +	if ((!!opts->new_branch + !!opts->new_branch_force + !!opts->new_orphan_branch) > 1)
>  		die(_("-b, -B and --orphan are mutually exclusive"));
>  
>  	/*
> @@ -1302,14 +1341,14 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
>  	 * and new_branch_force and new_orphan_branch will tell us which one of
>  	 * -b/-B/--orphan is being used.
>  	 */
> -	if (opts.new_branch_force)
> -		opts.new_branch = opts.new_branch_force;
> +	if (opts->new_branch_force)
> +		opts->new_branch = opts->new_branch_force;
>  
> -	if (opts.new_orphan_branch)
> -		opts.new_branch = opts.new_orphan_branch;
> +	if (opts->new_orphan_branch)
> +		opts->new_branch = opts->new_orphan_branch;
>  
>  	/* --track without -b/-B/--orphan should DWIM */
> -	if (opts.track != BRANCH_TRACK_UNSPECIFIED && !opts.new_branch) {
> +	if (opts->track != BRANCH_TRACK_UNSPECIFIED && !opts->new_branch) {
>  		const char *argv0 = argv[0];
>  		if (!argc || !strcmp(argv0, "--"))
>  			die(_("--track needs a branch name"));
> @@ -1318,7 +1357,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
>  		argv0 = strchr(argv0, '/');
>  		if (!argv0 || !argv0[1])
>  			die(_("missing branch name; try -b"));
> -		opts.new_branch = argv0 + 1;
> +		opts->new_branch = argv0 + 1;
>  	}
>  
>  	/*
> @@ -1337,56 +1376,56 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
>  	if (argc) {
>  		struct object_id rev;
>  		int dwim_ok =
> -			!opts.patch_mode &&
> -			dwim_new_local_branch &&
> -			opts.track == BRANCH_TRACK_UNSPECIFIED &&
> -			!opts.new_branch;
> +			!opts->patch_mode &&
> +			opts->dwim_new_local_branch &&
> +			opts->track == BRANCH_TRACK_UNSPECIFIED &&
> +			!opts->new_branch;
>  		int n = parse_branchname_arg(argc, argv, dwim_ok,
> -					     &new_branch_info, &opts, &rev,
> +					     &new_branch_info, opts, &rev,
>  					     &dwim_remotes_matched);
>  		argv += n;
>  		argc -= n;
>  	}
>  
>  	if (argc) {
> -		parse_pathspec(&opts.pathspec, 0,
> -			       opts.patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
> +		parse_pathspec(&opts->pathspec, 0,
> +			       opts->patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
>  			       prefix, argv);
>  
> -		if (!opts.pathspec.nr)
> +		if (!opts->pathspec.nr)
>  			die(_("invalid path specification"));
>  
>  		/*
>  		 * Try to give more helpful suggestion.
>  		 * new_branch && argc > 1 will be caught later.
>  		 */
> -		if (opts.new_branch && argc == 1)
> +		if (opts->new_branch && argc == 1)
>  			die(_("'%s' is not a commit and a branch '%s' cannot be created from it"),
> -				argv[0], opts.new_branch);
> +				argv[0], opts->new_branch);
>  
> -		if (opts.force_detach)
> +		if (opts->force_detach)
>  			die(_("git checkout: --detach does not take a path argument '%s'"),
>  			    argv[0]);
>  
> -		if (1 < !!opts.writeout_stage + !!opts.force + !!opts.merge)
> +		if (1 < !!opts->writeout_stage + !!opts->force + !!opts->merge)
>  			die(_("git checkout: --ours/--theirs, --force and --merge are incompatible when\n"
>  			      "checking out of the index."));
>  	}
>  
> -	if (opts.new_branch) {
> +	if (opts->new_branch) {
>  		struct strbuf buf = STRBUF_INIT;
>  
> -		if (opts.new_branch_force)
> -			opts.branch_exists = validate_branchname(opts.new_branch, &buf);
> +		if (opts->new_branch_force)
> +			opts->branch_exists = validate_branchname(opts->new_branch, &buf);
>  		else
> -			opts.branch_exists =
> -				validate_new_branchname(opts.new_branch, &buf, 0);
> +			opts->branch_exists =
> +				validate_new_branchname(opts->new_branch, &buf, 0);
>  		strbuf_release(&buf);
>  	}
>  
>  	UNLEAK(opts);
> -	if (opts.patch_mode || opts.pathspec.nr) {
> -		int ret = checkout_paths(&opts, new_branch_info.name);
> +	if (opts->patch_mode || opts->pathspec.nr) {
> +		int ret = checkout_paths(opts, new_branch_info.name);
>  		if (ret && dwim_remotes_matched > 1 &&
>  		    advice_checkout_ambiguous_remote_branch_name)
>  			advise(_("'%s' matched more than one remote tracking branch.\n"
> @@ -1405,6 +1444,49 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
>  			       dwim_remotes_matched);
>  		return ret;
>  	} else {
> -		return checkout_branch(&opts, &new_branch_info);
> +		return checkout_branch(opts, &new_branch_info);
>  	}
>  }
> +
> +int cmd_checkout(int argc, const char **argv, const char *prefix)
> +{
> +	struct checkout_opts opts;
> +	struct option *options = NULL;
> +	int ret;
> +
> +	options = add_common_options(&opts, options);
> +	options = add_switch_branch_options(&opts, options);
> +	options = add_checkout_path_options(&opts, options);
> +	ret = checkout_main(argc, argv, prefix, &opts,
> +			    options, checkout_usage);
> +	FREE_AND_NULL(options);
> +	return ret;
> +}
> +
> +int cmd_switch_branch(int argc, const char **argv, const char *prefix)
> +{
> +	struct checkout_opts opts;
> +	struct option *options = NULL;
> +	int ret;
> +
> +	options = add_common_options(&opts, options);
> +	options = add_switch_branch_options(&opts, options);
> +	ret = checkout_main(argc, argv, prefix, &opts,
> +			    options, switch_branch_usage);
> +	FREE_AND_NULL(options);
> +	return ret;
> +}
> +
> +int cmd_restore_paths(int argc, const char **argv, const char *prefix)
> +{
> +	struct checkout_opts opts;
> +	struct option *options = NULL;
> +	int ret;
> +
> +	options = add_common_options(&opts, options);
> +	options = add_checkout_path_options(&opts, options);
> +	ret = checkout_main(argc, argv, prefix, &opts,
> +			    options, restore_paths_usage);
> +	FREE_AND_NULL(options);
> +	return ret;
> +}
> diff --git a/git.c b/git.c
> index 2f604a41ea..e8a76a99da 100644
> --- a/git.c
> +++ b/git.c
> @@ -542,6 +542,7 @@ static struct cmd_struct commands[] = {
>  	{ "replace", cmd_replace, RUN_SETUP },
>  	{ "rerere", cmd_rerere, RUN_SETUP },
>  	{ "reset", cmd_reset, RUN_SETUP },
> +	{ "restore-paths", cmd_restore_paths, RUN_SETUP | NEED_WORK_TREE },
>  	{ "rev-list", cmd_rev_list, RUN_SETUP | NO_PARSEOPT },
>  	{ "rev-parse", cmd_rev_parse, NO_PARSEOPT },
>  	{ "revert", cmd_revert, RUN_SETUP | NEED_WORK_TREE },
> @@ -557,6 +558,7 @@ static struct cmd_struct commands[] = {
>  	{ "status", cmd_status, RUN_SETUP | NEED_WORK_TREE },
>  	{ "stripspace", cmd_stripspace },
>  	{ "submodule--helper", cmd_submodule__helper, RUN_SETUP | SUPPORT_SUPER_PREFIX | NO_PARSEOPT },
> +	{ "switch-branch", cmd_switch_branch, RUN_SETUP | NEED_WORK_TREE },
>  	{ "symbolic-ref", cmd_symbolic_ref, RUN_SETUP },
>  	{ "tag", cmd_tag, RUN_SETUP | DELAY_PAGER_CONFIG },
>  	{ "unpack-file", cmd_unpack_file, RUN_SETUP | NO_PARSEOPT },
> diff --git a/parse-options-cb.c b/parse-options-cb.c
> index 8c9edce52f..c609d52926 100644
> --- a/parse-options-cb.c
> +++ b/parse-options-cb.c
> @@ -126,7 +126,7 @@ struct option *parse_options_concat(struct option *a, struct option *b)
>  	struct option *ret;
>  	size_t i, a_len = 0, b_len = 0;
>  
> -	for (i = 0; a[i].type != OPTION_END; i++)
> +	for (i = 0; a && a[i].type != OPTION_END; i++)
>  		a_len++;
>  	for (i = 0; b[i].type != OPTION_END; i++)
>  		b_len++;
> -- 8< --

Re: [RFC] Introduce two new commands, switch-branch and restore-paths

[Junio C Hamano]

Thomas Gummerer <t.gummerer@gmail.com> writes:

> I like the idea of splitting those commands up, in fact it is
> something I've been considering working on myself.  I do think we
> should consider if we want to change the behaviour of those new
> commands in any way compared to 'git checkout', since we're starting
> with a clean slate.
>
> One thing in particular that I have in mind is something I'm currently
> working on, namely adding a --index flag to 'git checkout', which
> would make 'git checkout' work in non-overlay mode (for more
> discussion on that see also [*1*].

Ah, thanks for reminding me of that.  That explains why I felt
uneasy to see "restore" in the proposed command name.  In short, I
think "checkout --index <tree> <pathspec>", i.e. if the <pathspec>
matches a directory in <tree> and the current index and/or the
working tree has tracked paths in that directory that do not exist
in <tree>, the operation _removes_ these paths so that the result
matches <tree>, should become the default of "I want to check out
these paths from the named tree-ish".  The current one is not
exactly "checking out the paths" in that it ignores and does not
check out the absense of paths in <tree>.  That operation sounds
more like "restoring paths out of a given tree".  If the tree does
not have some paths, these paths won't be "restored" from that tree,
so "restore" matches the current "overlay what's taken out of the
given tree on top of what is already in the index and the working
tree, without checking out the absense of paths" better from that
point of view.

Re: [RFC] Introduce two new commands, switch-branch and restore-paths

[Duy Nguyen]

On Mon, Nov 26, 2018 at 4:03 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> Thomas Gummerer <t.gummerer@gmail.com> writes:
>
> > I like the idea of splitting those commands up, in fact it is
> > something I've been considering working on myself.  I do think we
> > should consider if we want to change the behaviour of those new
> > commands in any way compared to 'git checkout', since we're starting
> > with a clean slate.

Better defaults? Hell yes!

> > One thing in particular that I have in mind is something I'm currently
> > working on, namely adding a --index flag to 'git checkout', which
> > would make 'git checkout' work in non-overlay mode (for more
> > discussion on that see also [*1*].
>
> Ah, thanks for reminding me of that.  That explains why I felt
> uneasy to see "restore" in the proposed command name.

About that name. I didn't want to start the command name with checkout
to avoid completion conflict (the obvious choice was checkout-path,
the function name behind it). And I didn't find any other good name,
so I picked "restore" out of git-checkout.txt's one-line description.
If we end up with a better command name, perhaps reword that line too.
-- 
Duy

Re: [RFC] Introduce two new commands, switch-branch and restore-paths

[Ævar Arnfjörð Bjarmason]

On Tue, Nov 20 2018, Duy Nguyen wrote:

> On Mon, Nov 19, 2018 at 04:19:53PM +0100, Duy Nguyen wrote:
>> I promise to come back with something better (at least it still
>> sounds better in my mind). If that idea does not work out, we can
>> come back and see if we can improve this.
>
> So this is it. The patch isn't pretty, mostly as a proof of
> concept. Just look at the three functions at the bottom of checkout.c,
> which is the main thing.
>
> This patch tries to split "git checkout" command in two new ones:
>
> - git switch-branch is all about switching branches

Isn't this going to also take the other ref arguments 'git checkout'
takes now? I.e. tags, detached HEADs etc? I'm reminded of the discussion
about what "range-diff" should be called :)

> - git restore-paths (maybe restore-file is better) for checking out
>   paths

If it takes globs/dirs then a plural is probably better.

> The main idea is these two commands will co-exist with the good old
> 'git checkout', which will NOT be deprecated. Old timers will still
> use "git checkout". But new people should be introduced to the new two
> instead. And the new ones are just as capable as "git checkout".
>
> Since the three commands will co-exist (with duplicate functionality),
> maintenance cost must be kept to minimum. The way I did this is simply
> split the command line options into three pieces: common,
> switch-branch and checkout-paths. "git checkout" has all three, the
> other two have common and another piece.
>
> With this, a new option added to git checkout will be automatically
> available in either switch-branch or checkout-paths. Bug fixes apply
> to all relevant commands.
>
> Later on, we could start to add a bit more stuff in, e.g. some form of
> disambiguation is no longer needed when running as switch-branch, or
> restore-paths.
>
> So, what do you think?

That "git checkout" does too many things is something that keeps coming
up in online discussions about Git's UI. Two things:

a) It would really help to have some comparison of cases where these
   split commands are much clearer or less ambiguous than
   git-checkout. I can think of some (e.g. branch with the same name as
   a file) but having some overall picture of what the new UI looks like
   with solved / not solved cases would be nice. Also a comparison with
   other SCMs people find less confusing (svn, hg, bzr, ...)

b) I think we really need to have some end-game where we'd actually
   switch away from "checkout" (which we could still auto-route to new
   commands in perpetuity, but print a warning or error). Otherwise
   we'll just end up with https://xkcd.com/927/ and more UI confusion
   for all.

Re: [RFC] Introduce two new commands, switch-branch and restore-paths

[Duy Nguyen]

On Mon, Nov 26, 2018 at 5:01 PM Ævar Arnfjörð Bjarmason
<avarab@gmail.com> wrote:
> > So, what do you think?
>
> That "git checkout" does too many things is something that keeps coming
> up in online discussions about Git's UI. Two things:
>
> a) It would really help to have some comparison of cases where these
>    split commands are much clearer or less ambiguous than
>    git-checkout. I can think of some (e.g. branch with the same name as
>    a file) but having some overall picture of what the new UI looks like
>    with solved / not solved cases would be nice. Also a comparison with
>    other SCMs people find less confusing (svn, hg, bzr, ...)

Less ambiguous is indeed one of the reasons I wanted to do this.

> b) I think we really need to have some end-game where we'd actually
>    switch away from "checkout" (which we could still auto-route to new
>    commands in perpetuity, but print a warning or error). Otherwise
>    we'll just end up with https://xkcd.com/927/ and more UI confusion
>    for all.

I'm not going to remove "git checkout". Not until the majority of
users are really in favor of the new ones. So my end-game plan is to
just promote the two new commands in man pages, tutorial, advice...
Perhaps at some point "git checkout" is also removed from "git help".
But that's about it. If you want to stick with "git checkout", it's
there (and not nagging you to move to new ones).
-- 
Duy

Re: [RFC] Introduce two new commands, switch-branch and restore-paths

[Stefan Beller]

On Mon, Nov 26, 2018 at 8:01 AM Ævar Arnfjörð Bjarmason
<avarab@gmail.com> wrote:
>
>
> On Tue, Nov 20 2018, Duy Nguyen wrote:
>
> > On Mon, Nov 19, 2018 at 04:19:53PM +0100, Duy Nguyen wrote:
> >> I promise to come back with something better (at least it still
> >> sounds better in my mind). If that idea does not work out, we can
> >> come back and see if we can improve this.
> >
> > So this is it. The patch isn't pretty, mostly as a proof of
> > concept. Just look at the three functions at the bottom of checkout.c,
> > which is the main thing.
> >
> > This patch tries to split "git checkout" command in two new ones:
> >
> > - git switch-branch is all about switching branches
>
> Isn't this going to also take the other ref arguments 'git checkout'
> takes now? I.e. tags, detached HEADs etc? I'm reminded of the discussion
> about what "range-diff" should be called :)

Heh, good call. :-)
Note that the color of a bikeshed has fewer functional implications
than coming up with proper names in your API exposed to millions
of users, as cognitive associations playing mind tricks, can have a
huge impact on their productivity. ;-)

In a neighboring thread there is discussion of the concept of a
'change' (and evolving the change locally), which is yet another
thing in the refs-space.

'switch-branch' sounds like a good name for a beginner who is just
getting started, but as soon as they discover that there is more than
branches (detached HEAD via commits, tags,
remote tracking "branches"), this name may be confusing.
So it would not be a good choice for the intermediate Git user.
The old time power user would not care as they have 'checkout'
in their muscle memory, aliased to 'co', but maybe they'd find
it nice for explaining to new users.

So I'd be thrilled to find a name that serves users on all levels.

Maybe we need to step back and consider what the command does.
And from that I would name it "rewire-HEAD-and-update-index&worktree"
and then simplify from there. For the beginner user, the concept of
HEAD might be overwhelming, such that we don't want to have that
in there.

So I'd be tempted to suggest to call it "switch-to-ref", but that would
be wrong in the corner case as well: When using that with a remote
tracking branch, you don't "switch to it" by putting it into your HEAD,
but you merely checkout the commit that it's pointing at.



>
> > - git restore-paths (maybe restore-file is better) for checking out
> >   paths

"content-to-path", maybe(?) as it moves the content (as given by commit
or implicitly assuming the index when omitted) into that path(, again).
(I am not enthused about this, as you can similarly argue for
content-to-paths, content-to-worktree, which then could split up into
"index-to-worktree [pathspec]" as well as "tree-to-worktree <commit>".
also the notion of X-to-Y seems a novel concept in our naming, so maybe
verb-noun is better, hence restore-path or "fix-paths" may be better)

> > Later on, we could start to add a bit more stuff in, e.g. some form of
> > disambiguation is no longer needed when running as switch-branch, or
> > restore-paths.
> >
> > So, what do you think?

The patch looks interestingly small :-)

> That "git checkout" does too many things is something that keeps coming
> up in online discussions about Git's UI. Two things:
>
> a) It would really help to have some comparison of cases where these
>    split commands are much clearer or less ambiguous than
>    git-checkout. I can think of some (e.g. branch with the same name as
>    a file) but having some overall picture of what the new UI looks like
>    with solved / not solved cases would be nice. Also a comparison with
>    other SCMs people find less confusing (svn, hg, bzr, ...)

How do other SCMs solve this issue? (What is their design space?
How many commands do they have for what git-checkout does
all-in-one?)

> b) I think we really need to have some end-game where we'd actually
>    switch away from "checkout" (which we could still auto-route to new
>    commands in perpetuity, but print a warning or error). Otherwise
>    we'll just end up with https://xkcd.com/927/ and more UI confusion
>    for all.

Heh, that situation is only avoided when the new command has clear
advantages over the old, and ISTM that we can only compete on
UX and better defaults, so maybe I'd push for making it more logical,
maybe so:

  git tree-to-worktree # git checkout <commit> -- <path>
  git index-to-worktree # git checkout -- <path>
  git rev-to-ref # git checkout <commit>

Just food for thought, specifically the last one would be
hilarious if we'd end up with it.

Stefan

Re: [RFC] Introduce two new commands, switch-branch and restore-paths

[Junio C Hamano]

Stefan Beller <sbeller@google.com> writes:

> Maybe we need to step back and consider what the command does.
> And from that I would name it "rewire-HEAD-and-update-index&worktree"
> and then simplify from there. For the beginner user, the concept of
> HEAD might be overwhelming, such that we don't want to have that
> in there.

I'd have to say that it is totally backwards.

Use of HEAD, ref, etc. is merely a means to what the end users want
to achieve, which is to switch to a "branch" (in air quotes, as the
word in this sentence means a bit broader than "a ref that is in
refs/heads/"), to choose which lineage of commits to work on to grow
or reshape the history.  Most of the time, you would be working on a
branch (that is a ref that is in refs/heads/), sometimes you would
be working on an unnamed "branch" (i.e. detached HEAD state, only
difference between being on it and a normal branch is whether it is
named---the history manipulation can happen exactly the same way).

In other words, your initial motivation of stepping back and
thinking about what the command is about is very good.  But in the
context of checking out a branch, the concept the end user works
with are at the level of "branch", not HEAD, ref, index, working
tree (all of which are underlying implementation details that let
you work on manipulating the history, represented by the "branch").

> "content-to-path", maybe(?) ...

Path is not a place.  A path t/Makefile is a shared name of a place
in a tree, the index, or in the working tree.  Renaming "git add" to
"content-to-path" because it adds the content to the index at path
may be equally OK, but it misses the essense (which is that it is to
add to the index and not to anythng else like a tree or the working
tree).

I actually think the easiest-to-understand shorthand for the
operation "checking out the contents at the path to the working
tree" is "checkout".  

These...

>   git tree-to-worktree # git checkout <commit> -- <path>
>   git index-to-worktree # git checkout -- <path>

...are interesting and worth learning lessons from.  These would be
something people would suggest when users start making noises about
"restore-to-path" or "content-to-path" overloads three different
operations and is confusing.

I think "restore-to-path" that can take contents for individual
paths out of either a treeish or the index and update the index and
the working tree, depending on what the user tells it to do, is not
confusing to the end users.  At the conceptual level, the users need
to have a mental model that has three places (tree-ish, index and
working tree) that can hold contents to make use of these
operations, so it is only the matter of how to express it clearly
and concisely.

For that matter, "checkout <branch>", "checkout <treeish> -- <path>"
and "checkout -- <path>" already is a trio of clear and cocncise way
to spell three distinct operations, so with the right mental model,
it may not be so confusing to the users.  And "checkout-branch",
"checkout-contents-from-tree", and "checkout-contents-from-index"
longhands may be a good way to help new users form the right mental
model, as they are more explicit in their names; once they form the
right mental model (that is, there are three places the contents
live, and there are a few operations to move contents from the two
places to the working tree, i.e. what traditionally is called
"checking things out"), they may gradulate to the shorthand form.

[PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Nguyễn Thái Ngọc Duy]

v2 is just a bit better to look at than v1. This is by no means final.
If you think the command name is bad, the default behavior should
change, or something else, speak up. It's still very "RFC".

v2 breaks down the giant patch in v1 and starts adding some changes in
these new commands:

- restore-paths is renamed to checkout-paths. I wrote I didn't like
  "checkout" because of completion conflict. But who am I kidding,
  I'll use aliases anyway. "-files" instead of "-paths" because we
  already have ls-files.
- both commands will not accept no arguments. There is no "git
  checkout" equivalent.
- ambiguation rules are now aware that "switch-branch" for example
  can't take pathspec...
- the last patch tries to hide "git checkout" away. The command
  example updates show how these will be used. Which probably helps
  figure better names or defaults for them too

One thing I notice that we often use "git checkout -- <path>" and
rarely "git checkout <tree-ish> -- <path>". Which makes me think
perhaps "git checkout-files" should not use "--" to separate the two.
We'll have this instead

    git checkout-files [--from=<tree-ish>] <paths>


Oh and of course I'll be waiting for the new --index from Thomas
before submitting submitting any thing serious for 'next'. We still
have plenty of time.

Nguyễn Thái Ngọc Duy (7):
  parse-options: allow parse_options_concat(NULL, options)
  checkout: make "opts" in cmd_checkout() a pointer
  checkout: move 'confict_style' to checkout_opts
  checkout: move dwim_new_local_branch to checkout_opts
  checkout: split options[] array in three pieces
  checkout: split into switch-branch and checkout-files
  Suggest other commands instead of "git checkout"

 Documentation/git-branch.txt           |   8 +-
 Documentation/git-check-ref-format.txt |   2 +-
 Documentation/git-format-patch.txt     |   2 +-
 Documentation/git-merge-base.txt       |   2 +-
 Documentation/git-rebase.txt           |   2 +-
 Documentation/git-remote.txt           |   2 +-
 Documentation/git-rerere.txt           |  10 +-
 Documentation/git-reset.txt            |  18 +-
 Documentation/git-revert.txt           |   2 +-
 Documentation/git-stash.txt            |   6 +-
 Documentation/gitattributes.txt        |   2 +-
 Documentation/gitcli.txt               |   4 +-
 Documentation/gitcore-tutorial.txt     |  18 +-
 Documentation/giteveryday.txt          |  24 +--
 Documentation/githooks.txt             |   5 +-
 Documentation/gittutorial-2.txt        |   2 +-
 Documentation/gittutorial.txt          |   4 +-
 Documentation/revisions.txt            |   2 +-
 Documentation/user-manual.txt          |  54 +++---
 advice.c                               |   2 +-
 builtin.h                              |   2 +
 builtin/checkout.c                     | 256 +++++++++++++++++--------
 git.c                                  |   2 +
 parse-options-cb.c                     |   2 +-
 sha1-name.c                            |   2 +-
 wt-status.c                            |   2 +-
 26 files changed, 271 insertions(+), 166 deletions(-)

-- 
2.19.1.1327.g328c130451.dirty

[PATCH v2 1/7] parse-options: allow parse_options_concat(NULL, options)

[Nguyễn Thái Ngọc Duy]

There is currently no caller that calls this function with "a" being
NULL. But it will be introduced shortly. It is used to construct the
option array from scratch, e.g.

   struct parse_options opts = NULL;
   opts = parse_options_concat(opts, opts_1);
   opts = parse_options_concat(opts, opts_2);

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 parse-options-cb.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/parse-options-cb.c b/parse-options-cb.c
index 8c9edce52f..c609d52926 100644
--- a/parse-options-cb.c
+++ b/parse-options-cb.c
@@ -126,7 +126,7 @@ struct option *parse_options_concat(struct option *a, struct option *b)
 	struct option *ret;
 	size_t i, a_len = 0, b_len = 0;
 
-	for (i = 0; a[i].type != OPTION_END; i++)
+	for (i = 0; a && a[i].type != OPTION_END; i++)
 		a_len++;
 	for (i = 0; b[i].type != OPTION_END; i++)
 		b_len++;
-- 
2.19.1.1327.g328c130451.dirty

[PATCH v2 2/7] checkout: make "opts" in cmd_checkout() a pointer

[Nguyễn Thái Ngọc Duy]

"opts" will soon be moved out of cmd_checkout(). To keep changes in
that patch smaller, convert "opts" to a pointer and keep the real
thing behind "real_opts".

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 109 +++++++++++++++++++++++----------------------
 1 file changed, 55 insertions(+), 54 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index acdafc6e4c..31245c1eb4 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -1225,76 +1225,77 @@ static int checkout_branch(struct checkout_opts *opts,
 
 int cmd_checkout(int argc, const char **argv, const char *prefix)
 {
-	struct checkout_opts opts;
+	struct checkout_opts real_opts;
+	struct checkout_opts *opts = &real_opts;
 	struct branch_info new_branch_info;
 	char *conflict_style = NULL;
 	int dwim_new_local_branch = 1;
 	int dwim_remotes_matched = 0;
 	struct option options[] = {
-		OPT__QUIET(&opts.quiet, N_("suppress progress reporting")),
-		OPT_STRING('b', NULL, &opts.new_branch, N_("branch"),
+		OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
+		OPT_STRING('b', NULL, &opts->new_branch, N_("branch"),
 			   N_("create and checkout a new branch")),
-		OPT_STRING('B', NULL, &opts.new_branch_force, N_("branch"),
+		OPT_STRING('B', NULL, &opts->new_branch_force, N_("branch"),
 			   N_("create/reset and checkout a branch")),
-		OPT_BOOL('l', NULL, &opts.new_branch_log, N_("create reflog for new branch")),
-		OPT_BOOL(0, "detach", &opts.force_detach, N_("detach HEAD at named commit")),
-		OPT_SET_INT('t', "track",  &opts.track, N_("set upstream info for new branch"),
+		OPT_BOOL('l', NULL, &opts->new_branch_log, N_("create reflog for new branch")),
+		OPT_BOOL(0, "detach", &opts->force_detach, N_("detach HEAD at named commit")),
+		OPT_SET_INT('t', "track",  &opts->track, N_("set upstream info for new branch"),
 			BRANCH_TRACK_EXPLICIT),
-		OPT_STRING(0, "orphan", &opts.new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
-		OPT_SET_INT_F('2', "ours", &opts.writeout_stage,
+		OPT_STRING(0, "orphan", &opts->new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
+		OPT_SET_INT_F('2', "ours", &opts->writeout_stage,
 			      N_("checkout our version for unmerged files"),
 			      2, PARSE_OPT_NONEG),
-		OPT_SET_INT_F('3', "theirs", &opts.writeout_stage,
+		OPT_SET_INT_F('3', "theirs", &opts->writeout_stage,
 			      N_("checkout their version for unmerged files"),
 			      3, PARSE_OPT_NONEG),
-		OPT__FORCE(&opts.force, N_("force checkout (throw away local modifications)"),
+		OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
 			   PARSE_OPT_NOCOMPLETE),
-		OPT_BOOL('m', "merge", &opts.merge, N_("perform a 3-way merge with the new branch")),
-		OPT_BOOL_F(0, "overwrite-ignore", &opts.overwrite_ignore,
+		OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
+		OPT_BOOL_F(0, "overwrite-ignore", &opts->overwrite_ignore,
 			   N_("update ignored files (default)"),
 			   PARSE_OPT_NOCOMPLETE),
 		OPT_STRING(0, "conflict", &conflict_style, N_("style"),
 			   N_("conflict style (merge or diff3)")),
-		OPT_BOOL('p', "patch", &opts.patch_mode, N_("select hunks interactively")),
-		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts.ignore_skipworktree,
+		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
+		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
 			 N_("do not limit pathspecs to sparse entries only")),
 		OPT_HIDDEN_BOOL(0, "guess", &dwim_new_local_branch,
 				N_("second guess 'git checkout <no-such-branch>'")),
-		OPT_BOOL(0, "ignore-other-worktrees", &opts.ignore_other_worktrees,
+		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
 			 N_("do not check if another worktree is holding the given ref")),
 		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
 			    "checkout", "control recursive updating of submodules",
 			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
-		OPT_BOOL(0, "progress", &opts.show_progress, N_("force progress reporting")),
+		OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
 		OPT_END(),
 	};
 
-	memset(&opts, 0, sizeof(opts));
+	memset(opts, 0, sizeof(*opts));
 	memset(&new_branch_info, 0, sizeof(new_branch_info));
-	opts.overwrite_ignore = 1;
-	opts.prefix = prefix;
-	opts.show_progress = -1;
+	opts->overwrite_ignore = 1;
+	opts->prefix = prefix;
+	opts->show_progress = -1;
 
-	git_config(git_checkout_config, &opts);
+	git_config(git_checkout_config, opts);
 
-	opts.track = BRANCH_TRACK_UNSPECIFIED;
+	opts->track = BRANCH_TRACK_UNSPECIFIED;
 
 	argc = parse_options(argc, argv, prefix, options, checkout_usage,
 			     PARSE_OPT_KEEP_DASHDASH);
 
-	if (opts.show_progress < 0) {
-		if (opts.quiet)
-			opts.show_progress = 0;
+	if (opts->show_progress < 0) {
+		if (opts->quiet)
+			opts->show_progress = 0;
 		else
-			opts.show_progress = isatty(2);
+			opts->show_progress = isatty(2);
 	}
 
 	if (conflict_style) {
-		opts.merge = 1; /* implied */
+		opts->merge = 1; /* implied */
 		git_xmerge_config("merge.conflictstyle", conflict_style, NULL);
 	}
 
-	if ((!!opts.new_branch + !!opts.new_branch_force + !!opts.new_orphan_branch) > 1)
+	if ((!!opts->new_branch + !!opts->new_branch_force + !!opts->new_orphan_branch) > 1)
 		die(_("-b, -B and --orphan are mutually exclusive"));
 
 	/*
@@ -1302,14 +1303,14 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	 * and new_branch_force and new_orphan_branch will tell us which one of
 	 * -b/-B/--orphan is being used.
 	 */
-	if (opts.new_branch_force)
-		opts.new_branch = opts.new_branch_force;
+	if (opts->new_branch_force)
+		opts->new_branch = opts->new_branch_force;
 
-	if (opts.new_orphan_branch)
-		opts.new_branch = opts.new_orphan_branch;
+	if (opts->new_orphan_branch)
+		opts->new_branch = opts->new_orphan_branch;
 
 	/* --track without -b/-B/--orphan should DWIM */
-	if (opts.track != BRANCH_TRACK_UNSPECIFIED && !opts.new_branch) {
+	if (opts->track != BRANCH_TRACK_UNSPECIFIED && !opts->new_branch) {
 		const char *argv0 = argv[0];
 		if (!argc || !strcmp(argv0, "--"))
 			die(_("--track needs a branch name"));
@@ -1318,7 +1319,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		argv0 = strchr(argv0, '/');
 		if (!argv0 || !argv0[1])
 			die(_("missing branch name; try -b"));
-		opts.new_branch = argv0 + 1;
+		opts->new_branch = argv0 + 1;
 	}
 
 	/*
@@ -1337,56 +1338,56 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	if (argc) {
 		struct object_id rev;
 		int dwim_ok =
-			!opts.patch_mode &&
+			!opts->patch_mode &&
 			dwim_new_local_branch &&
-			opts.track == BRANCH_TRACK_UNSPECIFIED &&
-			!opts.new_branch;
+			opts->track == BRANCH_TRACK_UNSPECIFIED &&
+			!opts->new_branch;
 		int n = parse_branchname_arg(argc, argv, dwim_ok,
-					     &new_branch_info, &opts, &rev,
+					     &new_branch_info, opts, &rev,
 					     &dwim_remotes_matched);
 		argv += n;
 		argc -= n;
 	}
 
 	if (argc) {
-		parse_pathspec(&opts.pathspec, 0,
-			       opts.patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
+		parse_pathspec(&opts->pathspec, 0,
+			       opts->patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
 			       prefix, argv);
 
-		if (!opts.pathspec.nr)
+		if (!opts->pathspec.nr)
 			die(_("invalid path specification"));
 
 		/*
 		 * Try to give more helpful suggestion.
 		 * new_branch && argc > 1 will be caught later.
 		 */
-		if (opts.new_branch && argc == 1)
+		if (opts->new_branch && argc == 1)
 			die(_("'%s' is not a commit and a branch '%s' cannot be created from it"),
-				argv[0], opts.new_branch);
+				argv[0], opts->new_branch);
 
-		if (opts.force_detach)
+		if (opts->force_detach)
 			die(_("git checkout: --detach does not take a path argument '%s'"),
 			    argv[0]);
 
-		if (1 < !!opts.writeout_stage + !!opts.force + !!opts.merge)
+		if (1 < !!opts->writeout_stage + !!opts->force + !!opts->merge)
 			die(_("git checkout: --ours/--theirs, --force and --merge are incompatible when\n"
 			      "checking out of the index."));
 	}
 
-	if (opts.new_branch) {
+	if (opts->new_branch) {
 		struct strbuf buf = STRBUF_INIT;
 
-		if (opts.new_branch_force)
-			opts.branch_exists = validate_branchname(opts.new_branch, &buf);
+		if (opts->new_branch_force)
+			opts->branch_exists = validate_branchname(opts->new_branch, &buf);
 		else
-			opts.branch_exists =
-				validate_new_branchname(opts.new_branch, &buf, 0);
+			opts->branch_exists =
+				validate_new_branchname(opts->new_branch, &buf, 0);
 		strbuf_release(&buf);
 	}
 
 	UNLEAK(opts);
-	if (opts.patch_mode || opts.pathspec.nr) {
-		int ret = checkout_paths(&opts, new_branch_info.name);
+	if (opts->patch_mode || opts->pathspec.nr) {
+		int ret = checkout_paths(opts, new_branch_info.name);
 		if (ret && dwim_remotes_matched > 1 &&
 		    advice_checkout_ambiguous_remote_branch_name)
 			advise(_("'%s' matched more than one remote tracking branch.\n"
@@ -1405,6 +1406,6 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 			       dwim_remotes_matched);
 		return ret;
 	} else {
-		return checkout_branch(&opts, &new_branch_info);
+		return checkout_branch(opts, &new_branch_info);
 	}
 }
-- 
2.19.1.1327.g328c130451.dirty

[PATCH v2 3/7] checkout: move 'confict_style' to checkout_opts

[Nguyễn Thái Ngọc Duy]

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 31245c1eb4..211a347a0c 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -55,6 +55,7 @@ struct checkout_opts {
 	int new_branch_log;
 	enum branch_track track;
 	struct diff_options diff_options;
+	char *conflict_style;
 
 	int branch_exists;
 	const char *prefix;
@@ -1228,7 +1229,6 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	struct checkout_opts real_opts;
 	struct checkout_opts *opts = &real_opts;
 	struct branch_info new_branch_info;
-	char *conflict_style = NULL;
 	int dwim_new_local_branch = 1;
 	int dwim_remotes_matched = 0;
 	struct option options[] = {
@@ -1254,7 +1254,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		OPT_BOOL_F(0, "overwrite-ignore", &opts->overwrite_ignore,
 			   N_("update ignored files (default)"),
 			   PARSE_OPT_NOCOMPLETE),
-		OPT_STRING(0, "conflict", &conflict_style, N_("style"),
+		OPT_STRING(0, "conflict", &opts->conflict_style, N_("style"),
 			   N_("conflict style (merge or diff3)")),
 		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
 		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
@@ -1290,9 +1290,9 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 			opts->show_progress = isatty(2);
 	}
 
-	if (conflict_style) {
+	if (opts->conflict_style) {
 		opts->merge = 1; /* implied */
-		git_xmerge_config("merge.conflictstyle", conflict_style, NULL);
+		git_xmerge_config("merge.conflictstyle", opts->conflict_style, NULL);
 	}
 
 	if ((!!opts->new_branch + !!opts->new_branch_force + !!opts->new_orphan_branch) > 1)
-- 
2.19.1.1327.g328c130451.dirty

[PATCH v2 4/7] checkout: move dwim_new_local_branch to checkout_opts

[Nguyễn Thái Ngọc Duy]

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 211a347a0c..a50c51f287 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -44,6 +44,8 @@ struct checkout_opts {
 	int ignore_skipworktree;
 	int ignore_other_worktrees;
 	int show_progress;
+	int dwim_new_local_branch;
+
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
 	 * should be updated accordingly.
@@ -1229,7 +1231,6 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	struct checkout_opts real_opts;
 	struct checkout_opts *opts = &real_opts;
 	struct branch_info new_branch_info;
-	int dwim_new_local_branch = 1;
 	int dwim_remotes_matched = 0;
 	struct option options[] = {
 		OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
@@ -1259,7 +1260,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
 		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
 			 N_("do not limit pathspecs to sparse entries only")),
-		OPT_HIDDEN_BOOL(0, "guess", &dwim_new_local_branch,
+		OPT_HIDDEN_BOOL(0, "guess", &opts->dwim_new_local_branch,
 				N_("second guess 'git checkout <no-such-branch>'")),
 		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
 			 N_("do not check if another worktree is holding the given ref")),
@@ -1275,6 +1276,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	opts->overwrite_ignore = 1;
 	opts->prefix = prefix;
 	opts->show_progress = -1;
+	opts->dwim_new_local_branch = 1;
 
 	git_config(git_checkout_config, opts);
 
@@ -1339,7 +1341,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		struct object_id rev;
 		int dwim_ok =
 			!opts->patch_mode &&
-			dwim_new_local_branch &&
+			opts->dwim_new_local_branch &&
 			opts->track == BRANCH_TRACK_UNSPECIFIED &&
 			!opts->new_branch;
 		int n = parse_branchname_arg(argc, argv, dwim_ok,
-- 
2.19.1.1327.g328c130451.dirty

[PATCH v2 5/7] checkout: split options[] array in three pieces

[Nguyễn Thái Ngọc Duy]

This is a preparation step for introducing new commands that do parts
of what checkout does. There will be two new commands, one is about
switching branches, detaching HEAD... one about checking out
paths. These share the a subset of command line options. The rest of
command line options are separate.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 80 ++++++++++++++++++++++++++++++++--------------
 1 file changed, 56 insertions(+), 24 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index a50c51f287..d9dbd2d40d 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -1226,14 +1226,32 @@ static int checkout_branch(struct checkout_opts *opts,
 	return switch_branches(opts, new_branch_info);
 }
 
-int cmd_checkout(int argc, const char **argv, const char *prefix)
+static struct option *add_common_options(struct checkout_opts *opts,
+					 struct option *prevopts)
 {
-	struct checkout_opts real_opts;
-	struct checkout_opts *opts = &real_opts;
-	struct branch_info new_branch_info;
-	int dwim_remotes_matched = 0;
 	struct option options[] = {
 		OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
+		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
+			 N_("do not limit pathspecs to sparse entries only")),
+		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
+			    "checkout", "control recursive updating of submodules",
+			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
+		OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
+		OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
+			   PARSE_OPT_NOCOMPLETE),
+		OPT_STRING(0, "conflict", &opts->conflict_style, N_("style"),
+			   N_("conflict style (merge or diff3)")),
+		OPT_END()
+	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
+
+static struct option *add_switch_branch_options(struct checkout_opts *opts,
+						struct option *prevopts)
+{
+	struct option options[] = {
 		OPT_STRING('b', NULL, &opts->new_branch, N_("branch"),
 			   N_("create and checkout a new branch")),
 		OPT_STRING('B', NULL, &opts->new_branch_force, N_("branch"),
@@ -1243,33 +1261,43 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		OPT_SET_INT('t', "track",  &opts->track, N_("set upstream info for new branch"),
 			BRANCH_TRACK_EXPLICIT),
 		OPT_STRING(0, "orphan", &opts->new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
+		OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
+		OPT_HIDDEN_BOOL(0, "guess", &opts->dwim_new_local_branch,
+				N_("second guess 'git checkout <no-such-branch>'")),
+		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
+			 N_("do not check if another worktree is holding the given ref")),
+		OPT_END()
+	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
+
+static struct option *add_checkout_path_options(struct checkout_opts *opts,
+						struct option *prevopts)
+{
+	struct option options[] = {
 		OPT_SET_INT_F('2', "ours", &opts->writeout_stage,
 			      N_("checkout our version for unmerged files"),
 			      2, PARSE_OPT_NONEG),
 		OPT_SET_INT_F('3', "theirs", &opts->writeout_stage,
 			      N_("checkout their version for unmerged files"),
 			      3, PARSE_OPT_NONEG),
-		OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
-			   PARSE_OPT_NOCOMPLETE),
-		OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
-		OPT_BOOL_F(0, "overwrite-ignore", &opts->overwrite_ignore,
-			   N_("update ignored files (default)"),
-			   PARSE_OPT_NOCOMPLETE),
-		OPT_STRING(0, "conflict", &opts->conflict_style, N_("style"),
-			   N_("conflict style (merge or diff3)")),
 		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
-		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
-			 N_("do not limit pathspecs to sparse entries only")),
-		OPT_HIDDEN_BOOL(0, "guess", &opts->dwim_new_local_branch,
-				N_("second guess 'git checkout <no-such-branch>'")),
-		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
-			 N_("do not check if another worktree is holding the given ref")),
-		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
-			    "checkout", "control recursive updating of submodules",
-			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
-		OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
-		OPT_END(),
+		OPT_END()
 	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
+
+int cmd_checkout(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts real_opts;
+	struct checkout_opts *opts = &real_opts;
+	struct branch_info new_branch_info;
+	int dwim_remotes_matched = 0;
+	struct option *options = NULL;
 
 	memset(opts, 0, sizeof(*opts));
 	memset(&new_branch_info, 0, sizeof(new_branch_info));
@@ -1282,6 +1310,10 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 
 	opts->track = BRANCH_TRACK_UNSPECIFIED;
 
+	options = add_common_options(opts, options);
+	options = add_switch_branch_options(opts, options);
+	options = add_checkout_path_options(opts, options);
+
 	argc = parse_options(argc, argv, prefix, options, checkout_usage,
 			     PARSE_OPT_KEEP_DASHDASH);
 
-- 
2.19.1.1327.g328c130451.dirty

[PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Nguyễn Thái Ngọc Duy]

"git checkout" doing too many things is a source of confusion for many
users (and it even bites old timers sometimes). To rememdy that, the
command is now split in two: switch-branch and checkout-files.

The switch-branch command is all about switching branches, detaching,
DWIM-ing new branch... It does not accept pathspecs and it always
requires a ref (in contrast, "git checkout" without arguments works)

The checkout-files command on the other hand is all about resetting
certain files in worktree, either from the index or from a specific
tree. It could accept a tree-ish, but it will never touch HEAD or the
ref it points to.

The good old "git checkout" command is still here and will be until
all (or most of users) are sick of it.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin.h          |  2 +
 builtin/checkout.c | 91 +++++++++++++++++++++++++++++++++++++++-------
 git.c              |  2 +
 3 files changed, 82 insertions(+), 13 deletions(-)

diff --git a/builtin.h b/builtin.h
index 6538932e99..d4a66e5f79 100644
--- a/builtin.h
+++ b/builtin.h
@@ -138,6 +138,7 @@ extern int cmd_branch(int argc, const char **argv, const char *prefix);
 extern int cmd_bundle(int argc, const char **argv, const char *prefix);
 extern int cmd_cat_file(int argc, const char **argv, const char *prefix);
 extern int cmd_checkout(int argc, const char **argv, const char *prefix);
+extern int cmd_checkout_files(int argc, const char **argv, const char *prefix);
 extern int cmd_checkout_index(int argc, const char **argv, const char *prefix);
 extern int cmd_check_attr(int argc, const char **argv, const char *prefix);
 extern int cmd_check_ignore(int argc, const char **argv, const char *prefix);
@@ -227,6 +228,7 @@ extern int cmd_show_index(int argc, const char **argv, const char *prefix);
 extern int cmd_status(int argc, const char **argv, const char *prefix);
 extern int cmd_stripspace(int argc, const char **argv, const char *prefix);
 extern int cmd_submodule__helper(int argc, const char **argv, const char *prefix);
+extern int cmd_switch_branch(int argc, const char **argv, const char *prefix);
 extern int cmd_symbolic_ref(int argc, const char **argv, const char *prefix);
 extern int cmd_tag(int argc, const char **argv, const char *prefix);
 extern int cmd_tar_tree(int argc, const char **argv, const char *prefix);
diff --git a/builtin/checkout.c b/builtin/checkout.c
index d9dbd2d40d..c09d2da47a 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -33,6 +33,16 @@ static const char * const checkout_usage[] = {
 	NULL,
 };
 
+static const char * const switch_branch_usage[] = {
+	N_("git switch-branch [<options>] <branch>"),
+	NULL,
+};
+
+static const char * const checkout_files_usage[] = {
+	N_("git checkout-files [<options>] [<branch>] -- <file>..."),
+	NULL,
+};
+
 struct checkout_opts {
 	int patch_mode;
 	int quiet;
@@ -45,6 +55,8 @@ struct checkout_opts {
 	int ignore_other_worktrees;
 	int show_progress;
 	int dwim_new_local_branch;
+	int accept_pathspec;
+	int empty_arg_ok;
 
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
@@ -1056,7 +1068,7 @@ static int parse_branchname_arg(int argc, const char **argv,
 	arg = argv[0];
 	dash_dash_pos = -1;
 	for (i = 0; i < argc; i++) {
-		if (!strcmp(argv[i], "--")) {
+		if (opts->accept_pathspec && !strcmp(argv[i], "--")) {
 			dash_dash_pos = i;
 			break;
 		}
@@ -1067,6 +1079,8 @@ static int parse_branchname_arg(int argc, const char **argv,
 		has_dash_dash = 1; /* case (3) or (1) */
 	else if (dash_dash_pos >= 2)
 		die(_("only one reference expected, %d given."), dash_dash_pos);
+	else if (!opts->accept_pathspec)
+		has_dash_dash = 1;
 
 	if (!strcmp(arg, "-"))
 		arg = "@{-1}";
@@ -1291,30 +1305,23 @@ static struct option *add_checkout_path_options(struct checkout_opts *opts,
 	return newopts;
 }
 
-int cmd_checkout(int argc, const char **argv, const char *prefix)
+static int checkout_main(int argc, const char **argv, const char *prefix,
+			 struct checkout_opts *opts, struct option *options,
+			 const char * const usagestr[])
 {
-	struct checkout_opts real_opts;
-	struct checkout_opts *opts = &real_opts;
 	struct branch_info new_branch_info;
 	int dwim_remotes_matched = 0;
-	struct option *options = NULL;
 
-	memset(opts, 0, sizeof(*opts));
 	memset(&new_branch_info, 0, sizeof(new_branch_info));
 	opts->overwrite_ignore = 1;
 	opts->prefix = prefix;
 	opts->show_progress = -1;
-	opts->dwim_new_local_branch = 1;
 
 	git_config(git_checkout_config, opts);
 
 	opts->track = BRANCH_TRACK_UNSPECIFIED;
 
-	options = add_common_options(opts, options);
-	options = add_switch_branch_options(opts, options);
-	options = add_checkout_path_options(opts, options);
-
-	argc = parse_options(argc, argv, prefix, options, checkout_usage,
+	argc = parse_options(argc, argv, prefix, options, usagestr,
 			     PARSE_OPT_KEEP_DASHDASH);
 
 	if (opts->show_progress < 0) {
@@ -1381,7 +1388,8 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 					     &dwim_remotes_matched);
 		argv += n;
 		argc -= n;
-	}
+	} else if (!opts->empty_arg_ok)
+		usage_with_options(usagestr, options);
 
 	if (argc) {
 		parse_pathspec(&opts->pathspec, 0,
@@ -1443,3 +1451,60 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		return checkout_branch(opts, &new_branch_info);
 	}
 }
+
+int cmd_checkout(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	memset(&opts, 0, sizeof(opts));
+	opts.dwim_new_local_branch = 1;
+	opts.accept_pathspec = 1;
+	opts.empty_arg_ok = 1;
+
+	options = add_common_options(&opts, options);
+	options = add_switch_branch_options(&opts, options);
+	options = add_checkout_path_options(&opts, options);
+
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, checkout_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
+
+int cmd_switch_branch(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	memset(&opts, 0, sizeof(opts));
+	opts.dwim_new_local_branch = 1;
+
+	options = add_common_options(&opts, options);
+	options = add_switch_branch_options(&opts, options);
+
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, switch_branch_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
+
+int cmd_checkout_files(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	memset(&opts, 0, sizeof(opts));
+	opts.accept_pathspec = 1;
+
+	options = add_common_options(&opts, options);
+	options = add_checkout_path_options(&opts, options);
+
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, checkout_files_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
diff --git a/git.c b/git.c
index 2f604a41ea..3b86ba765c 100644
--- a/git.c
+++ b/git.c
@@ -457,6 +457,7 @@ static struct cmd_struct commands[] = {
 	{ "check-mailmap", cmd_check_mailmap, RUN_SETUP },
 	{ "check-ref-format", cmd_check_ref_format, NO_PARSEOPT  },
 	{ "checkout", cmd_checkout, RUN_SETUP | NEED_WORK_TREE },
+	{ "checkout-files", cmd_checkout_files, RUN_SETUP | NEED_WORK_TREE },
 	{ "checkout-index", cmd_checkout_index,
 		RUN_SETUP | NEED_WORK_TREE},
 	{ "cherry", cmd_cherry, RUN_SETUP },
@@ -557,6 +558,7 @@ static struct cmd_struct commands[] = {
 	{ "status", cmd_status, RUN_SETUP | NEED_WORK_TREE },
 	{ "stripspace", cmd_stripspace },
 	{ "submodule--helper", cmd_submodule__helper, RUN_SETUP | SUPPORT_SUPER_PREFIX | NO_PARSEOPT },
+	{ "switch-branch", cmd_switch_branch, RUN_SETUP | NEED_WORK_TREE },
 	{ "symbolic-ref", cmd_symbolic_ref, RUN_SETUP },
 	{ "tag", cmd_tag, RUN_SETUP | DELAY_PAGER_CONFIG },
 	{ "unpack-file", cmd_unpack_file, RUN_SETUP | NO_PARSEOPT },
-- 
2.19.1.1327.g328c130451.dirty

[PATCH v2 7/7] Suggest other commands instead of "git checkout"

[Nguyễn Thái Ngọc Duy]

The assumption made is here

- "git checkout" is a horrible monster that should only be touched
  with a two-meter pole

- there are other commands that can achieve the same thing

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 Documentation/git-branch.txt           |  8 ++--
 Documentation/git-check-ref-format.txt |  2 +-
 Documentation/git-format-patch.txt     |  2 +-
 Documentation/git-merge-base.txt       |  2 +-
 Documentation/git-rebase.txt           |  2 +-
 Documentation/git-remote.txt           |  2 +-
 Documentation/git-rerere.txt           | 10 ++---
 Documentation/git-reset.txt            | 18 ++++-----
 Documentation/git-revert.txt           |  2 +-
 Documentation/git-stash.txt            |  6 +--
 Documentation/gitattributes.txt        |  2 +-
 Documentation/gitcli.txt               |  4 +-
 Documentation/gitcore-tutorial.txt     | 18 ++++-----
 Documentation/giteveryday.txt          | 24 ++++++------
 Documentation/githooks.txt             |  5 ++-
 Documentation/gittutorial-2.txt        |  2 +-
 Documentation/gittutorial.txt          |  4 +-
 Documentation/revisions.txt            |  2 +-
 Documentation/user-manual.txt          | 54 +++++++++++++-------------
 advice.c                               |  2 +-
 sha1-name.c                            |  2 +-
 wt-status.c                            |  2 +-
 22 files changed, 88 insertions(+), 87 deletions(-)

diff --git a/Documentation/git-branch.txt b/Documentation/git-branch.txt
index bf5316ffa9..1564df47d2 100644
--- a/Documentation/git-branch.txt
+++ b/Documentation/git-branch.txt
@@ -48,7 +48,7 @@ The command's second form creates a new branch head named <branchname>
 which points to the current `HEAD`, or <start-point> if given.
 
 Note that this will create the new branch, but it will not switch the
-working tree to it; use "git checkout <newbranch>" to switch to the
+working tree to it; use "git switch-branch <newbranch>" to switch to the
 new branch.
 
 When a local branch is started off a remote-tracking branch, Git sets up the
@@ -194,7 +194,7 @@ This option is only applicable in non-verbose mode.
 +
 This behavior is the default when the start point is a remote-tracking branch.
 Set the branch.autoSetupMerge configuration variable to `false` if you
-want `git checkout` and `git branch` to always behave as if `--no-track`
+want `git switch-branch` and `git branch` to always behave as if `--no-track`
 were given. Set it to `always` if you want this behavior when the
 start-point is either a local or remote-tracking branch.
 
@@ -293,7 +293,7 @@ Start development from a known tag::
 $ git clone git://git.kernel.org/pub/scm/.../linux-2.6 my2.6
 $ cd my2.6
 $ git branch my2.6.14 v2.6.14   <1>
-$ git checkout my2.6.14
+$ git switch-branch my2.6.14
 ------------
 +
 <1> This step and the next one could be combined into a single step with
@@ -319,7 +319,7 @@ NOTES
 -----
 
 If you are creating a branch that you want to checkout immediately, it is
-easier to use the git checkout command with its `-b` option to create
+easier to use the "git switch-branch" command with its `-b` option to create
 a branch and check it out with a single command.
 
 The options `--contains`, `--no-contains`, `--merged` and `--no-merged`
diff --git a/Documentation/git-check-ref-format.txt b/Documentation/git-check-ref-format.txt
index d9de992585..38c2169d7a 100644
--- a/Documentation/git-check-ref-format.txt
+++ b/Documentation/git-check-ref-format.txt
@@ -88,7 +88,7 @@ but it is explicitly forbidden at the beginning of a branch name).
 When run with `--branch` option in a repository, the input is first
 expanded for the ``previous checkout syntax''
 `@{-n}`.  For example, `@{-1}` is a way to refer the last thing that
-was checked out using "git checkout" operation. This option should be
+was checked out using "git switch-branch" operation. This option should be
 used by porcelains to accept this syntax anywhere a branch name is
 expected, so they can act as if you typed the branch name. As an
 exception note that, the ``previous checkout operation'' might result
diff --git a/Documentation/git-format-patch.txt b/Documentation/git-format-patch.txt
index aba4c5febe..0ceaa1173c 100644
--- a/Documentation/git-format-patch.txt
+++ b/Documentation/git-format-patch.txt
@@ -416,7 +416,7 @@ One way to test if your MUA is set up correctly is:
 * Apply it:
 
     $ git fetch <project> master:test-apply
-    $ git checkout test-apply
+    $ git switch-branch test-apply
     $ git reset --hard
     $ git am a.patch
 
diff --git a/Documentation/git-merge-base.txt b/Documentation/git-merge-base.txt
index 9f07f4f6ed..1b25e5d530 100644
--- a/Documentation/git-merge-base.txt
+++ b/Documentation/git-merge-base.txt
@@ -149,7 +149,7 @@ instead.
 Discussion on fork-point mode
 -----------------------------
 
-After working on the `topic` branch created with `git checkout -b
+After working on the `topic` branch created with `git switch-branch -b
 topic origin/master`, the history of remote-tracking branch
 `origin/master` may have been rewound and rebuilt, leading to a
 history of this shape:
diff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt
index 80793bad8d..fe10880633 100644
--- a/Documentation/git-rebase.txt
+++ b/Documentation/git-rebase.txt
@@ -17,7 +17,7 @@ SYNOPSIS
 DESCRIPTION
 -----------
 If <branch> is specified, 'git rebase' will perform an automatic
-`git checkout <branch>` before doing anything else.  Otherwise
+`git switch-branch <branch>` before doing anything else.  Otherwise
 it remains on the current branch.
 
 If <upstream> is not specified, the upstream configured in
diff --git a/Documentation/git-remote.txt b/Documentation/git-remote.txt
index 0cad37fb81..044bbdb27c 100644
--- a/Documentation/git-remote.txt
+++ b/Documentation/git-remote.txt
@@ -230,7 +230,7 @@ $ git branch -r
   staging/master
   staging/staging-linus
   staging/staging-next
-$ git checkout -b staging staging/master
+$ git switch-branch -b staging staging/master
 ...
 ------------
 
diff --git a/Documentation/git-rerere.txt b/Documentation/git-rerere.txt
index df310d2a58..fe9d21b395 100644
--- a/Documentation/git-rerere.txt
+++ b/Documentation/git-rerere.txt
@@ -91,7 +91,7 @@ For such a test, you need to merge master and topic somehow.
 One way to do it is to pull master into the topic branch:
 
 ------------
-	$ git checkout topic
+	$ git switch-branch topic
 	$ git merge master
 
               o---*---o---+ topic
@@ -113,10 +113,10 @@ the upstream might have been advanced since the test merge `+`,
 in which case the final commit graph would look like this:
 
 ------------
-	$ git checkout topic
+	$ git switch-branch topic
 	$ git merge master
 	$ ... work on both topic and master branches
-	$ git checkout master
+	$ git switch-branch master
 	$ git merge topic
 
               o---*---o---+---o---o topic
@@ -136,11 +136,11 @@ merges, you could blow away the test merge, and keep building on
 top of the tip before the test merge:
 
 ------------
-	$ git checkout topic
+	$ git switch-branch topic
 	$ git merge master
 	$ git reset --hard HEAD^ ;# rewind the test merge
 	$ ... work on both topic and master branches
-	$ git checkout master
+	$ git switch-branch master
 	$ git merge topic
 
               o---*---o-------o---o topic
diff --git a/Documentation/git-reset.txt b/Documentation/git-reset.txt
index 2dac95c71a..ca46b4c967 100644
--- a/Documentation/git-reset.txt
+++ b/Documentation/git-reset.txt
@@ -149,9 +149,9 @@ See also the --amend option to linkgit:git-commit[1].
 Undo a commit, making it a topic branch::
 +
 ------------
-$ git branch topic/wip     <1>
-$ git reset --hard HEAD~3  <2>
-$ git checkout topic/wip   <3>
+$ git branch topic/wip          <1>
+$ git reset --hard HEAD~3       <2>
+$ git switch-branch topic/wip   <3>
 ------------
 +
 <1> You have made some commits, but realize they were premature
@@ -232,13 +232,13 @@ working tree are not in any shape to be committed yet, but you
 need to get to the other branch for a quick bugfix.
 +
 ------------
-$ git checkout feature ;# you were working in "feature" branch and
+$ git switch-branch feature ;# you were working in "feature" branch and
 $ work work work       ;# got interrupted
 $ git commit -a -m "snapshot WIP"                 <1>
-$ git checkout master
+$ git switch-branch master
 $ fix fix fix
 $ git commit ;# commit with real log
-$ git checkout feature
+$ git switch-branch feature
 $ git reset --soft HEAD^ ;# go back to WIP state  <2>
 $ git reset                                       <3>
 ------------
@@ -279,18 +279,18 @@ reset it while keeping the changes in your working tree.
 +
 ------------
 $ git tag start
-$ git checkout -b branch1
+$ git switch-branch -b branch1
 $ edit
 $ git commit ...                            <1>
 $ edit
-$ git checkout -b branch2                   <2>
+$ git switch-branch -b branch2              <2>
 $ git reset --keep start                    <3>
 ------------
 +
 <1> This commits your first edits in branch1.
 <2> In the ideal world, you could have realized that the earlier
     commit did not belong to the new topic when you created and switched
-    to branch2 (i.e. "git checkout -b branch2 start"), but nobody is
+    to branch2 (i.e. "git switch-branch -b branch2 start"), but nobody is
     perfect.
 <3> But you can use "reset --keep" to remove the unwanted commit after
     you switched to "branch2".
diff --git a/Documentation/git-revert.txt b/Documentation/git-revert.txt
index 837707a8fd..e49dbbec83 100644
--- a/Documentation/git-revert.txt
+++ b/Documentation/git-revert.txt
@@ -26,7 +26,7 @@ effect of some earlier commits (often only a faulty one).  If you want to
 throw away all uncommitted changes in your working directory, you
 should see linkgit:git-reset[1], particularly the `--hard` option.  If
 you want to extract specific files as they were in another commit, you
-should see linkgit:git-checkout[1], specifically the `git checkout
+should see linkgit:git-checkout[1], specifically the `git checkout-files
 <commit> -- <filename>` syntax.  Take care with these alternatives as
 both will discard uncommitted changes in your working directory.
 
diff --git a/Documentation/git-stash.txt b/Documentation/git-stash.txt
index 7ef8c47911..ea226979b1 100644
--- a/Documentation/git-stash.txt
+++ b/Documentation/git-stash.txt
@@ -235,12 +235,12 @@ return to your original branch to make the emergency fix, like this:
 +
 ----------------------------------------------------------------
 # ... hack hack hack ...
-$ git checkout -b my_wip
+$ git switch-branch -b my_wip
 $ git commit -a -m "WIP"
-$ git checkout master
+$ git switch-branch master
 $ edit emergency fix
 $ git commit -a -m "Fix in a hurry"
-$ git checkout my_wip
+$ git switch-branch my_wip
 $ git reset --soft HEAD^
 # ... continue hacking ...
 ----------------------------------------------------------------
diff --git a/Documentation/gitattributes.txt b/Documentation/gitattributes.txt
index b8392fc330..df62bd8019 100644
--- a/Documentation/gitattributes.txt
+++ b/Documentation/gitattributes.txt
@@ -112,7 +112,7 @@ Checking-out and checking-in
 
 These attributes affect how the contents stored in the
 repository are copied to the working tree files when commands
-such as 'git checkout' and 'git merge' run.  They also affect how
+such as 'git switch-branch' and 'git merge' run.  They also affect how
 Git stores the contents you prepare in the working tree in the
 repository upon 'git add' and 'git commit'.
 
diff --git a/Documentation/gitcli.txt b/Documentation/gitcli.txt
index 592e06d839..0ad4869f2c 100644
--- a/Documentation/gitcli.txt
+++ b/Documentation/gitcli.txt
@@ -47,8 +47,8 @@ disambiguating `--` at appropriate places.
    things:
 +
 --------------------------------
-$ git checkout -- *.c
-$ git checkout -- \*.c
+$ git checkout-files -- *.c
+$ git checkout-files -- \*.c
 --------------------------------
 +
 The former lets your shell expand the fileglob, and you are asking
diff --git a/Documentation/gitcore-tutorial.txt b/Documentation/gitcore-tutorial.txt
index e29a9effcc..49a8b5aa52 100644
--- a/Documentation/gitcore-tutorial.txt
+++ b/Documentation/gitcore-tutorial.txt
@@ -741,7 +741,7 @@ used earlier, and create a branch in it. You do that by simply just
 saying that you want to check out a new branch:
 
 ------------
-$ git checkout -b mybranch
+$ git branch mybranch
 ------------
 
 will create a new branch based at the current `HEAD` position, and switch
@@ -755,7 +755,7 @@ just telling 'git checkout' what the base of the checkout would be.
 In other words, if you have an earlier tag or branch, you'd just do
 
 ------------
-$ git checkout -b mybranch earlier-commit
+$ git switch-branch -b mybranch earlier-commit
 ------------
 
 and it would create the new branch `mybranch` at the earlier commit,
@@ -765,7 +765,7 @@ and check out the state at that time.
 You can always just jump back to your original `master` branch by doing
 
 ------------
-$ git checkout master
+$ git switch-branch master
 ------------
 
 (or any other branch-name, for that matter) and if you forget which
@@ -794,7 +794,7 @@ $ git branch <branchname> [startingpoint]
 
 which will simply _create_ the branch, but will not do anything further.
 You can then later -- once you decide that you want to actually develop
-on that branch -- switch to that branch with a regular 'git checkout'
+on that branch -- switch to that branch with a regular 'git switch-branch
 with the branchname as the argument.
 
 
@@ -808,7 +808,7 @@ being the same as the original `master` branch, let's make sure we're in
 that branch, and do some work there.
 
 ------------------------------------------------
-$ git checkout mybranch
+$ git switch-branch mybranch
 $ echo "Work, work, work" >>hello
 $ git commit -m "Some work." -i hello
 ------------------------------------------------
@@ -825,7 +825,7 @@ does some work in the original branch, and simulate that by going back
 to the master branch, and editing the same file differently there:
 
 ------------
-$ git checkout master
+$ git switch-branch master
 ------------
 
 Here, take a moment to look at the contents of `hello`, and notice how they
@@ -958,7 +958,7 @@ to the `master` branch. Let's go back to `mybranch`, and run
 'git merge' to get the "upstream changes" back to your branch.
 
 ------------
-$ git checkout mybranch
+$ git switch-branch mybranch
 $ git merge -m "Merge upstream changes." master
 ------------
 
@@ -1133,9 +1133,9 @@ Remember, before running 'git merge', our `master` head was at
 work." commit.
 
 ------------
-$ git checkout mybranch
+$ git switch-branch mybranch
 $ git reset --hard master^2
-$ git checkout master
+$ git switch-branch master
 $ git reset --hard master^
 ------------
 
diff --git a/Documentation/giteveryday.txt b/Documentation/giteveryday.txt
index 9f2528fc8c..861b2bb616 100644
--- a/Documentation/giteveryday.txt
+++ b/Documentation/giteveryday.txt
@@ -80,9 +80,9 @@ $ git tag v2.43 <2>
 Create a topic branch and develop.::
 +
 ------------
-$ git checkout -b alsa-audio <1>
+$ git branch alsa-audio <1>
 $ edit/compile/test
-$ git checkout -- curses/ux_audio_oss.c <2>
+$ git checkout-files -- curses/ux_audio_oss.c <2>
 $ git add curses/ux_audio_alsa.c <3>
 $ edit/compile/test
 $ git diff HEAD <4>
@@ -90,7 +90,7 @@ $ git commit -a -s <5>
 $ edit/compile/test
 $ git diff HEAD^ <6>
 $ git commit -a --amend <7>
-$ git checkout master <8>
+$ git switch-branch master <8>
 $ git merge alsa-audio <9>
 $ git log --since='3 days ago' <10>
 $ git log v2.43.. curses/ <11>
@@ -148,11 +148,11 @@ Clone the upstream and work on it.  Feed changes to upstream.::
 ------------
 $ git clone git://git.kernel.org/pub/scm/.../torvalds/linux-2.6 my2.6
 $ cd my2.6
-$ git checkout -b mine master <1>
+$ git switch-branch -b mine master <1>
 $ edit/compile/test; git commit -a -s <2>
 $ git format-patch master <3>
 $ git send-email --to="person <email@example.com>" 00*.patch <4>
-$ git checkout master <5>
+$ git switch-branch master <5>
 $ git pull <6>
 $ git log -p ORIG_HEAD.. arch/i386 include/asm-i386 <7>
 $ git ls-remote --heads http://git.kernel.org/.../jgarzik/libata-dev.git <8>
@@ -194,7 +194,7 @@ satellite$ edit/compile/test/commit
 satellite$ git push origin <4>
 
 mothership$ cd frotz
-mothership$ git checkout master
+mothership$ git switch-branch master
 mothership$ git merge satellite/master <5>
 ------------
 +
@@ -216,7 +216,7 @@ machine into the master branch.
 Branch off of a specific tag.::
 +
 ------------
-$ git checkout -b private2.6.14 v2.6.14 <1>
+$ git switch-branch -b private2.6.14 v2.6.14 <1>
 $ edit/compile/test; git commit -a
 $ git checkout master
 $ git cherry-pick v2.6.14..private2.6.14 <2>
@@ -274,14 +274,14 @@ $ mailx <3>
 & s 2 3 4 5 ./+to-apply
 & s 7 8 ./+hold-linus
 & q
-$ git checkout -b topic/one master
+$ git switch-branch -b topic/one master
 $ git am -3 -i -s ./+to-apply <4>
 $ compile/test
-$ git checkout -b hold/linus && git am -3 -i -s ./+hold-linus <5>
-$ git checkout topic/one && git rebase master <6>
-$ git checkout pu && git reset --hard next <7>
+$ git switch-branch -b hold/linus && git am -3 -i -s ./+hold-linus <5>
+$ git switch-branch topic/one && git rebase master <6>
+$ git switch-branch pu && git reset --hard next <7>
 $ git merge topic/one topic/two && git merge hold/linus <8>
-$ git checkout maint
+$ git switch-branch maint
 $ git cherry-pick master~4 <9>
 $ compile/test
 $ git tag -s -m "GIT 0.99.9x" v0.99.9x <10>
diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt
index 959044347e..3939ec774a 100644
--- a/Documentation/githooks.txt
+++ b/Documentation/githooks.txt
@@ -166,7 +166,7 @@ worktree.  The hook is given three parameters: the ref of the previous HEAD,
 the ref of the new HEAD (which may or may not have changed), and a flag
 indicating whether the checkout was a branch checkout (changing branches,
 flag=1) or a file checkout (retrieving a file from the index, flag=0).
-This hook cannot affect the outcome of `git checkout`.
+This hook cannot affect the outcome of `git switch-branch` or `git checkout`.
 
 It is also run after linkgit:git-clone[1], unless the `--no-checkout` (`-n`) option is
 used. The first parameter given to the hook is the null-ref, the second the
@@ -402,7 +402,8 @@ exit with a zero status.
 For example, the hook can simply run `git read-tree -u -m HEAD "$1"`
 in order to emulate `git fetch` that is run in the reverse direction
 with `git push`, as the two-tree form of `git read-tree -u -m` is
-essentially the same as `git checkout` that switches branches while
+essentially the same as `git switch-branch` or `git checkout`
+that switches branches while
 keeping the local changes in the working tree that do not interfere
 with the difference between the branches.
 
diff --git a/Documentation/gittutorial-2.txt b/Documentation/gittutorial-2.txt
index e0976f6017..125213d951 100644
--- a/Documentation/gittutorial-2.txt
+++ b/Documentation/gittutorial-2.txt
@@ -376,7 +376,7 @@ Changes to be committed:
 
 Changes not staged for commit:
   (use "git add <file>..." to update what will be committed)
-  (use "git checkout -- <file>..." to discard changes in working directory)
+  (use "git checkout-files -- <file>..." to discard changes in working directory)
 
 	modified:   file.txt
 
diff --git a/Documentation/gittutorial.txt b/Documentation/gittutorial.txt
index 242de31cb6..396e55c191 100644
--- a/Documentation/gittutorial.txt
+++ b/Documentation/gittutorial.txt
@@ -207,7 +207,7 @@ automatically.  The asterisk marks the branch you are currently on;
 type
 
 ------------------------------------------------
-$ git checkout experimental
+$ git switch-branch experimental
 ------------------------------------------------
 
 to switch to the experimental branch.  Now edit a file, commit the
@@ -216,7 +216,7 @@ change, and switch back to the master branch:
 ------------------------------------------------
 (edit file)
 $ git commit -a
-$ git checkout master
+$ git switch-branch master
 ------------------------------------------------
 
 Check that the change you made is no longer visible, since it was
diff --git a/Documentation/revisions.txt b/Documentation/revisions.txt
index 72daa20e76..f55502cd50 100644
--- a/Documentation/revisions.txt
+++ b/Documentation/revisions.txt
@@ -115,7 +115,7 @@ Here's an example to make it more clear:
 ------------------------------
 $ git config push.default current
 $ git config remote.pushdefault myfork
-$ git checkout -b mybranch origin/master
+$ git switch-branch -b mybranch origin/master
 
 $ git rev-parse --symbolic-full-name @{upstream}
 refs/remotes/origin/master
diff --git a/Documentation/user-manual.txt b/Documentation/user-manual.txt
index eff7890274..56397e93d4 100644
--- a/Documentation/user-manual.txt
+++ b/Documentation/user-manual.txt
@@ -125,7 +125,7 @@ Create a new branch head pointing to one of these versions and check it
 out using linkgit:git-checkout[1]:
 
 ------------------------------------------------
-$ git checkout -b new v2.6.13
+$ git switch-branch -b new v2.6.13
 ------------------------------------------------
 
 The working directory then reflects the contents that the project had
@@ -282,10 +282,10 @@ a summary of the commands:
 	this command will fail with a warning.
 `git branch -D <branch>`::
 	delete the branch `<branch>` irrespective of its merged status.
-`git checkout <branch>`::
+`git switch-branch <branch>`::
 	make the current branch `<branch>`, updating the working
 	directory to reflect the version referenced by `<branch>`.
-`git checkout -b <new> <start-point>`::
+`git switch-branch -b <new> <start-point>`::
 	create a new branch `<new>` referencing `<start-point>`, and
 	check it out.
 
@@ -302,12 +302,12 @@ ref: refs/heads/master
 Examining an old version without creating a new branch
 ------------------------------------------------------
 
-The `git checkout` command normally expects a branch head, but will also
+The `git switch-branch` command normally expects a branch head, but will also
 accept an arbitrary commit; for example, you can check out the commit
 referenced by a tag:
 
 ------------------------------------------------
-$ git checkout v2.6.17
+$ git switch-branch v2.6.17
 Note: checking out 'v2.6.17'.
 
 You are in 'detached HEAD' state. You can look around, make experimental
@@ -317,7 +317,7 @@ state without impacting any branches by performing another checkout.
 If you want to create a new branch to retain commits you create, you may
 do so (now or later) by using -b with the checkout command again. Example:
 
-  git checkout -b new_branch_name
+  git switch-branch -b new_branch_name
 
 HEAD is now at 427abfa Linux v2.6.17
 ------------------------------------------------
@@ -373,7 +373,7 @@ You might want to build on one of these remote-tracking branches
 on a branch of your own, just as you would for a tag:
 
 ------------------------------------------------
-$ git checkout -b my-todo-copy origin/todo
+$ git switch-branch -b my-todo-copy origin/todo
 ------------------------------------------------
 
 You can also check out `origin/todo` directly to examine it or
@@ -1523,12 +1523,12 @@ Checking out an old version of a file
 
 In the process of undoing a previous bad change, you may find it
 useful to check out an older version of a particular file using
-linkgit:git-checkout[1].  We've used `git checkout` before to switch
+linkgit:git-checkout[1].  We've used `git switch-branch` before to switch
 branches, but it has quite different behavior if it is given a path
 name: the command
 
 -------------------------------------------------
-$ git checkout HEAD^ path/to/file
+$ git checkout-files HEAD^ path/to/file
 -------------------------------------------------
 
 replaces path/to/file by the contents it had in the commit HEAD^, and
@@ -2211,8 +2211,8 @@ $ git branch --track release origin/master
 These can be easily kept up to date using linkgit:git-pull[1].
 
 -------------------------------------------------
-$ git checkout test && git pull
-$ git checkout release && git pull
+$ git switch-branch test && git pull
+$ git switch-branch release && git pull
 -------------------------------------------------
 
 Important note!  If you have any local changes in these branches, then
@@ -2264,7 +2264,7 @@ tested changes
 2) help future bug hunters that use `git bisect` to find problems
 
 -------------------------------------------------
-$ git checkout -b speed-up-spinlocks v2.6.35
+$ git switch-branch -b speed-up-spinlocks v2.6.35
 -------------------------------------------------
 
 Now you apply the patch(es), run some tests, and commit the change(s).  If
@@ -2279,7 +2279,7 @@ When you are happy with the state of this change, you can merge it into the
 "test" branch in preparation to make it public:
 
 -------------------------------------------------
-$ git checkout test && git merge speed-up-spinlocks
+$ git switch-branch test && git merge speed-up-spinlocks
 -------------------------------------------------
 
 It is unlikely that you would have any conflicts here ... but you might if you
@@ -2291,7 +2291,7 @@ see the value of keeping each patch (or patch series) in its own branch.  It
 means that the patches can be moved into the `release` tree in any order.
 
 -------------------------------------------------
-$ git checkout release && git merge speed-up-spinlocks
+$ git switch-branch release && git merge speed-up-spinlocks
 -------------------------------------------------
 
 After a while, you will have a number of branches, and despite the
@@ -2358,7 +2358,7 @@ Here are some of the scripts that simplify all this even further.
 
 case "$1" in
 test|release)
-	git checkout $1 && git pull . origin
+	git switch-branch $1 && git pull . origin
 	;;
 origin)
 	before=$(git rev-parse refs/remotes/origin/master)
@@ -2400,7 +2400,7 @@ test|release)
 		echo $1 already merged into $2 1>&2
 		exit 1
 	fi
-	git checkout $2 && git pull . $1
+	git switch-branch $2 && git pull . $1
 	;;
 *)
 	usage
@@ -2512,7 +2512,7 @@ Suppose that you create a branch `mywork` on a remote-tracking branch
 `origin`, and create some commits on top of it:
 
 -------------------------------------------------
-$ git checkout -b mywork origin
+$ git switch-branch -b mywork origin
 $ vi file.txt
 $ git commit
 $ vi otherfile.txt
@@ -2552,7 +2552,7 @@ commits without any merges, you may instead choose to use
 linkgit:git-rebase[1]:
 
 -------------------------------------------------
-$ git checkout mywork
+$ git switch-branch mywork
 $ git rebase origin
 -------------------------------------------------
 
@@ -3668,13 +3668,13 @@ change within the submodule, and then update the superproject to reference the
 new commit:
 
 -------------------------------------------------
-$ git checkout master
+$ git switch-branch master
 -------------------------------------------------
 
 or
 
 -------------------------------------------------
-$ git checkout -b fix-up
+$ git switch-branch -b fix-up
 -------------------------------------------------
 
 then
@@ -4194,7 +4194,7 @@ start.
 A good place to start is with the contents of the initial commit, with:
 
 ----------------------------------------------------
-$ git checkout e83c5163
+$ git switch-branch e83c5163
 ----------------------------------------------------
 
 The initial revision lays the foundation for almost everything Git has
@@ -4437,10 +4437,10 @@ Managing branches
 -----------------
 
 -----------------------------------------------
-$ git branch	     # list all local branches in this repo
-$ git checkout test  # switch working directory to branch "test"
-$ git branch new     # create branch "new" starting at current HEAD
-$ git branch -d new  # delete branch "new"
+$ git branch			# list all local branches in this repo
+$ git switch-branch test	# switch working directory to branch "test"
+$ git branch new		# create branch "new" starting at current HEAD
+$ git branch -d new		# delete branch "new"
 -----------------------------------------------
 
 Instead of basing a new branch on current HEAD (the default), use:
@@ -4456,7 +4456,7 @@ $ git branch new test~10 # ten commits before tip of branch "test"
 Create and switch to a new branch at the same time:
 
 -----------------------------------------------
-$ git checkout -b new v2.6.15
+$ git switch-branch -b new v2.6.15
 -----------------------------------------------
 
 Update and examine branches from the repository you cloned from:
@@ -4467,7 +4467,7 @@ $ git branch -r		# list
   origin/master
   origin/next
   ...
-$ git checkout -b masterwork origin/master
+$ git switch-branch -b masterwork origin/master
 -----------------------------------------------
 
 Fetch a branch from a different repository, and give it a new
diff --git a/advice.c b/advice.c
index 5f35656409..1befdb2163 100644
--- a/advice.c
+++ b/advice.c
@@ -195,7 +195,7 @@ void detach_advice(const char *new_name)
 	"state without impacting any branches by performing another checkout.\n\n"
 	"If you want to create a new branch to retain commits you create, you may\n"
 	"do so (now or later) by using -b with the checkout command again. Example:\n\n"
-	"  git checkout -b <new-branch-name>\n\n");
+	"  git branch <new-branch-name>\n\n");
 
 	fprintf(stderr, fmt, new_name);
 }
diff --git a/sha1-name.c b/sha1-name.c
index faa60f69e3..4e4e14a45c 100644
--- a/sha1-name.c
+++ b/sha1-name.c
@@ -771,7 +771,7 @@ static int get_oid_basic(const char *str, int len, struct object_id *oid,
 	"because it will be ignored when you just specify 40-hex. These refs\n"
 	"may be created by mistake. For example,\n"
 	"\n"
-	"  git checkout -b $br $(git rev-parse ...)\n"
+	"  git switch-branch -b $br $(git rev-parse ...)\n"
 	"\n"
 	"where \"$br\" is somehow empty and a 40-hex ref is created. Please\n"
 	"examine these refs and maybe delete them. Turn this message off by\n"
diff --git a/wt-status.c b/wt-status.c
index a24711374c..6266683926 100644
--- a/wt-status.c
+++ b/wt-status.c
@@ -224,7 +224,7 @@ static void wt_longstatus_print_dirty_header(struct wt_status *s,
 		status_printf_ln(s, c, _("  (use \"git add <file>...\" to update what will be committed)"));
 	else
 		status_printf_ln(s, c, _("  (use \"git add/rm <file>...\" to update what will be committed)"));
-	status_printf_ln(s, c, _("  (use \"git checkout -- <file>...\" to discard changes in working directory)"));
+	status_printf_ln(s, c, _("  (use \"git checkout-files <file>...\" to discard changes in working directory)"));
 	if (has_dirty_submodules)
 		status_printf_ln(s, c, _("  (commit or discard the untracked or modified content in submodules)"));
 	status_printf_ln(s, c, "%s", "");
-- 
2.19.1.1327.g328c130451.dirty

Re: [PATCH v2 1/7] parse-options: allow parse_options_concat(NULL, options)

[Stefan Beller]

On Tue, Nov 27, 2018 at 8:53 AM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
>
> There is currently no caller that calls this function with "a" being
> NULL. But it will be introduced shortly. It is used to construct the
> option array from scratch, e.g.
>
>    struct parse_options opts = NULL;
>    opts = parse_options_concat(opts, opts_1);
>    opts = parse_options_concat(opts, opts_2);

While this addresses the immediate needs, I'd prefer to think
about the API exposure of parse_options_concat,
(related: do we want to have docs in its header file?)
and I'd recommend to make it symmetrical, i.e.
allow the second argument to also be NULL?

In the example given here, you'd just short it to

    struct parse_options opts = opts_1;
    opts = parse_options_concat(opts, opts_2);

if not for this patch. Are opts_{1,2} ever be NULL?

Re: [PATCH v2 3/7] checkout: move 'confict_style' to checkout_opts

[Stefan Beller]

On Tue, Nov 27, 2018 at 8:53 AM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
>

The last patches seemed self explanatory after the first RFC
and their commit message. This one is harder to reason about,
as --conflict is documented as "The same as --merge option
above, but ..." and --merge is "When switching branches, ..."
so ... ah my mind wandered off, expecting this to be a preparation
for the separate commands already, but this is just about ensuring
everything is in opts such that the split can be done later?

Re: [PATCH v2 4/7] checkout: move dwim_new_local_branch to checkout_opts

[Stefan Beller]

On Tue, Nov 27, 2018 at 8:53 AM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
>
> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>

I would not mind to have this squashed into the previous patch
but keeping it separated is fine, too.
(Reason for squashing: it makes it clearer that we do not
care about one specific option, but have to treat all the loose
options the same way.)

Re: [PATCH v2 1/7] parse-options: allow parse_options_concat(NULL, options)

[Junio C Hamano]

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> There is currently no caller that calls this function with "a" being
> NULL. But it will be introduced shortly. It is used to construct the
> option array from scratch, e.g.
>
>    struct parse_options opts = NULL;

Missing asterisk somewhere?

>    opts = parse_options_concat(opts, opts_1);
>    opts = parse_options_concat(opts, opts_2);
>
> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> ---
>  parse-options-cb.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/parse-options-cb.c b/parse-options-cb.c
> index 8c9edce52f..c609d52926 100644
> --- a/parse-options-cb.c
> +++ b/parse-options-cb.c
> @@ -126,7 +126,7 @@ struct option *parse_options_concat(struct option *a, struct option *b)
>  	struct option *ret;
>  	size_t i, a_len = 0, b_len = 0;
>  
> -	for (i = 0; a[i].type != OPTION_END; i++)
> +	for (i = 0; a && a[i].type != OPTION_END; i++)
>  		a_len++;
>  	for (i = 0; b[i].type != OPTION_END; i++)
>  		b_len++;

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Junio C Hamano]

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> The good old "git checkout" command is still here and will be until
> all (or most of users) are sick of it.

Two comments on the goal (the implementation looked reasonable
assuming the reader agrees with the gaol).

At least to me, the verb "switch" needs two things to switch
between, i.e. "switch A and B", unless it is "switch to X".
Either "switch-to-branch" or simply "switch-to", perhaps?

As I already hinted in my response to Stefan (?) about
checkout-from-tree vs checkout-from-index, a command with multiple
modes of operation is not confusing to people with the right mental
model, and I suspect that having two separate commands for "checking
out a branch" and "checking out paths" that is done by this step
would help users to form the right mental model.  So I tend to think
these two are "training wheels", and suspect that once they got it,
nobody will become "sick of" the single "checkout" command that can
be used to do either.  It's just the matter of being aware what can
be done (which requires the right mental model) and how to tell Git
what the user wants it do (two separate commands, operating mode
option, or just the implied command line syntax---once the user
knows what s/he is doing, these do not make that much a difference).

As to the implementation,

>  	int dwim_new_local_branch;
> +	int accept_pathspec;
> +	int empty_arg_ok;

I think this is a reasonable way to completely avoid the codepath
that considers an unknown arg being parsed could be a pathspec
element (e.g. switch-to-that-branch mode will never want to see
pathspec, so disambiguation logic and/or hint should not kick in).

At the beginning of cmd_switch_branches(), please "explicitly"
assign 0 to accept_pathspec field, after memset(0) clears the whole
structure.  When a reader sees memset(0), it is read as "giving a
clean and bland slate to futz with with a reasonable default", but
for this particular field, i.e. accept_pathspec, there is NO
reasoanble default.  It's not like switch-to-branch mode is the heir
to checkout and the other sibling checkout-files is doing a
non-default behaviour.

Same comment probably would apply to the other one, although I
didn't think too deeply about that one.


>  	/*
>  	 * If new checkout options are added, skip_merge_working_tree
> @@ -1056,7 +1068,7 @@ static int parse_branchname_arg(int argc, const char **argv,
>  	arg = argv[0];
>  	dash_dash_pos = -1;
>  	for (i = 0; i < argc; i++) {
> -		if (!strcmp(argv[i], "--")) {
> +		if (opts->accept_pathspec && !strcmp(argv[i], "--")) {
>  			dash_dash_pos = i;
>  			break;
>  		}
> @@ -1067,6 +1079,8 @@ static int parse_branchname_arg(int argc, const char **argv,
>  		has_dash_dash = 1; /* case (3) or (1) */
>  	else if (dash_dash_pos >= 2)
>  		die(_("only one reference expected, %d given."), dash_dash_pos);
> +	else if (!opts->accept_pathspec)
> +		has_dash_dash = 1;
>  
>  	if (!strcmp(arg, "-"))
>  		arg = "@{-1}";
> @@ -1291,30 +1305,23 @@ static struct option *add_checkout_path_options(struct checkout_opts *opts,
>  	return newopts;
>  }
>  
> -int cmd_checkout(int argc, const char **argv, const char *prefix)
> +static int checkout_main(int argc, const char **argv, const char *prefix,
> +			 struct checkout_opts *opts, struct option *options,
> +			 const char * const usagestr[])
>  {
> -	struct checkout_opts real_opts;
> -	struct checkout_opts *opts = &real_opts;
>  	struct branch_info new_branch_info;
>  	int dwim_remotes_matched = 0;
> -	struct option *options = NULL;
>  
> -	memset(opts, 0, sizeof(*opts));
>  	memset(&new_branch_info, 0, sizeof(new_branch_info));
>  	opts->overwrite_ignore = 1;
>  	opts->prefix = prefix;
>  	opts->show_progress = -1;
> -	opts->dwim_new_local_branch = 1;
>  
>  	git_config(git_checkout_config, opts);
>  
>  	opts->track = BRANCH_TRACK_UNSPECIFIED;
>  
> -	options = add_common_options(opts, options);
> -	options = add_switch_branch_options(opts, options);
> -	options = add_checkout_path_options(opts, options);
> -
> -	argc = parse_options(argc, argv, prefix, options, checkout_usage,
> +	argc = parse_options(argc, argv, prefix, options, usagestr,
>  			     PARSE_OPT_KEEP_DASHDASH);
>  
>  	if (opts->show_progress < 0) {
> @@ -1381,7 +1388,8 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
>  					     &dwim_remotes_matched);
>  		argv += n;
>  		argc -= n;
> -	}
> +	} else if (!opts->empty_arg_ok)
> +		usage_with_options(usagestr, options);
>  
>  	if (argc) {
>  		parse_pathspec(&opts->pathspec, 0,
> @@ -1443,3 +1451,60 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
>  		return checkout_branch(opts, &new_branch_info);
>  	}
>  }
> +
> +int cmd_checkout(int argc, const char **argv, const char *prefix)
> +{
> +	struct checkout_opts opts;
> +	struct option *options = NULL;
> +	int ret;
> +
> +	memset(&opts, 0, sizeof(opts));
> +	opts.dwim_new_local_branch = 1;
> +	opts.accept_pathspec = 1;
> +	opts.empty_arg_ok = 1;
> +
> +	options = add_common_options(&opts, options);
> +	options = add_switch_branch_options(&opts, options);
> +	options = add_checkout_path_options(&opts, options);
> +
> +	ret = checkout_main(argc, argv, prefix, &opts,
> +			    options, checkout_usage);
> +	FREE_AND_NULL(options);
> +	return ret;
> +}
> +
> +int cmd_switch_branch(int argc, const char **argv, const char *prefix)
> +{
> +	struct checkout_opts opts;
> +	struct option *options = NULL;
> +	int ret;
> +
> +	memset(&opts, 0, sizeof(opts));
> +	opts.dwim_new_local_branch = 1;
> +
> +	options = add_common_options(&opts, options);
> +	options = add_switch_branch_options(&opts, options);
> +
> +	ret = checkout_main(argc, argv, prefix, &opts,
> +			    options, switch_branch_usage);
> +	FREE_AND_NULL(options);
> +	return ret;
> +}
> +
> +int cmd_checkout_files(int argc, const char **argv, const char *prefix)
> +{
> +	struct checkout_opts opts;
> +	struct option *options = NULL;
> +	int ret;
> +
> +	memset(&opts, 0, sizeof(opts));
> +	opts.accept_pathspec = 1;
> +
> +	options = add_common_options(&opts, options);
> +	options = add_checkout_path_options(&opts, options);
> +
> +	ret = checkout_main(argc, argv, prefix, &opts,
> +			    options, checkout_files_usage);
> +	FREE_AND_NULL(options);
> +	return ret;
> +}
> diff --git a/git.c b/git.c
> index 2f604a41ea..3b86ba765c 100644
> --- a/git.c
> +++ b/git.c
> @@ -457,6 +457,7 @@ static struct cmd_struct commands[] = {
>  	{ "check-mailmap", cmd_check_mailmap, RUN_SETUP },
>  	{ "check-ref-format", cmd_check_ref_format, NO_PARSEOPT  },
>  	{ "checkout", cmd_checkout, RUN_SETUP | NEED_WORK_TREE },
> +	{ "checkout-files", cmd_checkout_files, RUN_SETUP | NEED_WORK_TREE },
>  	{ "checkout-index", cmd_checkout_index,
>  		RUN_SETUP | NEED_WORK_TREE},
>  	{ "cherry", cmd_cherry, RUN_SETUP },
> @@ -557,6 +558,7 @@ static struct cmd_struct commands[] = {
>  	{ "status", cmd_status, RUN_SETUP | NEED_WORK_TREE },
>  	{ "stripspace", cmd_stripspace },
>  	{ "submodule--helper", cmd_submodule__helper, RUN_SETUP | SUPPORT_SUPER_PREFIX | NO_PARSEOPT },
> +	{ "switch-branch", cmd_switch_branch, RUN_SETUP | NEED_WORK_TREE },
>  	{ "symbolic-ref", cmd_symbolic_ref, RUN_SETUP },
>  	{ "tag", cmd_tag, RUN_SETUP | DELAY_PAGER_CONFIG },
>  	{ "unpack-file", cmd_unpack_file, RUN_SETUP | NO_PARSEOPT },

Re: [PATCH v2 7/7] Suggest other commands instead of "git checkout"

[Junio C Hamano]

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> The assumption made is here
>
> - "git checkout" is a horrible monster that should only be touched
>   with a two-meter pole
>
> - there are other commands that can achieve the same thing

Thanks for clearly spelling out the assumptions.  It is good that
this step cames at the end, as the earlier 6 steps looked reasonable
to me.

Thanks.


>
> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> ---
>  Documentation/git-branch.txt           |  8 ++--
>  Documentation/git-check-ref-format.txt |  2 +-
>  Documentation/git-format-patch.txt     |  2 +-
>  Documentation/git-merge-base.txt       |  2 +-
>  Documentation/git-rebase.txt           |  2 +-
>  Documentation/git-remote.txt           |  2 +-
>  Documentation/git-rerere.txt           | 10 ++---
>  Documentation/git-reset.txt            | 18 ++++-----
>  Documentation/git-revert.txt           |  2 +-
>  Documentation/git-stash.txt            |  6 +--
>  Documentation/gitattributes.txt        |  2 +-
>  Documentation/gitcli.txt               |  4 +-
>  Documentation/gitcore-tutorial.txt     | 18 ++++-----
>  Documentation/giteveryday.txt          | 24 ++++++------
>  Documentation/githooks.txt             |  5 ++-
>  Documentation/gittutorial-2.txt        |  2 +-
>  Documentation/gittutorial.txt          |  4 +-
>  Documentation/revisions.txt            |  2 +-
>  Documentation/user-manual.txt          | 54 +++++++++++++-------------
>  advice.c                               |  2 +-
>  sha1-name.c                            |  2 +-
>  wt-status.c                            |  2 +-
>  22 files changed, 88 insertions(+), 87 deletions(-)
>
> diff --git a/Documentation/git-branch.txt b/Documentation/git-branch.txt
> index bf5316ffa9..1564df47d2 100644
> --- a/Documentation/git-branch.txt
> +++ b/Documentation/git-branch.txt
> @@ -48,7 +48,7 @@ The command's second form creates a new branch head named <branchname>
>  which points to the current `HEAD`, or <start-point> if given.
>  
>  Note that this will create the new branch, but it will not switch the
> -working tree to it; use "git checkout <newbranch>" to switch to the
> +working tree to it; use "git switch-branch <newbranch>" to switch to the
>  new branch.
>  
>  When a local branch is started off a remote-tracking branch, Git sets up the
> @@ -194,7 +194,7 @@ This option is only applicable in non-verbose mode.
>  +
>  This behavior is the default when the start point is a remote-tracking branch.
>  Set the branch.autoSetupMerge configuration variable to `false` if you
> -want `git checkout` and `git branch` to always behave as if `--no-track`
> +want `git switch-branch` and `git branch` to always behave as if `--no-track`
>  were given. Set it to `always` if you want this behavior when the
>  start-point is either a local or remote-tracking branch.
>  
> @@ -293,7 +293,7 @@ Start development from a known tag::
>  $ git clone git://git.kernel.org/pub/scm/.../linux-2.6 my2.6
>  $ cd my2.6
>  $ git branch my2.6.14 v2.6.14   <1>
> -$ git checkout my2.6.14
> +$ git switch-branch my2.6.14
>  ------------
>  +
>  <1> This step and the next one could be combined into a single step with
> @@ -319,7 +319,7 @@ NOTES
>  -----
>  
>  If you are creating a branch that you want to checkout immediately, it is
> -easier to use the git checkout command with its `-b` option to create
> +easier to use the "git switch-branch" command with its `-b` option to create
>  a branch and check it out with a single command.
>  
>  The options `--contains`, `--no-contains`, `--merged` and `--no-merged`
> diff --git a/Documentation/git-check-ref-format.txt b/Documentation/git-check-ref-format.txt
> index d9de992585..38c2169d7a 100644
> --- a/Documentation/git-check-ref-format.txt
> +++ b/Documentation/git-check-ref-format.txt
> @@ -88,7 +88,7 @@ but it is explicitly forbidden at the beginning of a branch name).
>  When run with `--branch` option in a repository, the input is first
>  expanded for the ``previous checkout syntax''
>  `@{-n}`.  For example, `@{-1}` is a way to refer the last thing that
> -was checked out using "git checkout" operation. This option should be
> +was checked out using "git switch-branch" operation. This option should be
>  used by porcelains to accept this syntax anywhere a branch name is
>  expected, so they can act as if you typed the branch name. As an
>  exception note that, the ``previous checkout operation'' might result
> diff --git a/Documentation/git-format-patch.txt b/Documentation/git-format-patch.txt
> index aba4c5febe..0ceaa1173c 100644
> --- a/Documentation/git-format-patch.txt
> +++ b/Documentation/git-format-patch.txt
> @@ -416,7 +416,7 @@ One way to test if your MUA is set up correctly is:
>  * Apply it:
>  
>      $ git fetch <project> master:test-apply
> -    $ git checkout test-apply
> +    $ git switch-branch test-apply
>      $ git reset --hard
>      $ git am a.patch
>  
> diff --git a/Documentation/git-merge-base.txt b/Documentation/git-merge-base.txt
> index 9f07f4f6ed..1b25e5d530 100644
> --- a/Documentation/git-merge-base.txt
> +++ b/Documentation/git-merge-base.txt
> @@ -149,7 +149,7 @@ instead.
>  Discussion on fork-point mode
>  -----------------------------
>  
> -After working on the `topic` branch created with `git checkout -b
> +After working on the `topic` branch created with `git switch-branch -b
>  topic origin/master`, the history of remote-tracking branch
>  `origin/master` may have been rewound and rebuilt, leading to a
>  history of this shape:
> diff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt
> index 80793bad8d..fe10880633 100644
> --- a/Documentation/git-rebase.txt
> +++ b/Documentation/git-rebase.txt
> @@ -17,7 +17,7 @@ SYNOPSIS
>  DESCRIPTION
>  -----------
>  If <branch> is specified, 'git rebase' will perform an automatic
> -`git checkout <branch>` before doing anything else.  Otherwise
> +`git switch-branch <branch>` before doing anything else.  Otherwise
>  it remains on the current branch.
>  
>  If <upstream> is not specified, the upstream configured in
> diff --git a/Documentation/git-remote.txt b/Documentation/git-remote.txt
> index 0cad37fb81..044bbdb27c 100644
> --- a/Documentation/git-remote.txt
> +++ b/Documentation/git-remote.txt
> @@ -230,7 +230,7 @@ $ git branch -r
>    staging/master
>    staging/staging-linus
>    staging/staging-next
> -$ git checkout -b staging staging/master
> +$ git switch-branch -b staging staging/master
>  ...
>  ------------
>  
> diff --git a/Documentation/git-rerere.txt b/Documentation/git-rerere.txt
> index df310d2a58..fe9d21b395 100644
> --- a/Documentation/git-rerere.txt
> +++ b/Documentation/git-rerere.txt
> @@ -91,7 +91,7 @@ For such a test, you need to merge master and topic somehow.
>  One way to do it is to pull master into the topic branch:
>  
>  ------------
> -	$ git checkout topic
> +	$ git switch-branch topic
>  	$ git merge master
>  
>                o---*---o---+ topic
> @@ -113,10 +113,10 @@ the upstream might have been advanced since the test merge `+`,
>  in which case the final commit graph would look like this:
>  
>  ------------
> -	$ git checkout topic
> +	$ git switch-branch topic
>  	$ git merge master
>  	$ ... work on both topic and master branches
> -	$ git checkout master
> +	$ git switch-branch master
>  	$ git merge topic
>  
>                o---*---o---+---o---o topic
> @@ -136,11 +136,11 @@ merges, you could blow away the test merge, and keep building on
>  top of the tip before the test merge:
>  
>  ------------
> -	$ git checkout topic
> +	$ git switch-branch topic
>  	$ git merge master
>  	$ git reset --hard HEAD^ ;# rewind the test merge
>  	$ ... work on both topic and master branches
> -	$ git checkout master
> +	$ git switch-branch master
>  	$ git merge topic
>  
>                o---*---o-------o---o topic
> diff --git a/Documentation/git-reset.txt b/Documentation/git-reset.txt
> index 2dac95c71a..ca46b4c967 100644
> --- a/Documentation/git-reset.txt
> +++ b/Documentation/git-reset.txt
> @@ -149,9 +149,9 @@ See also the --amend option to linkgit:git-commit[1].
>  Undo a commit, making it a topic branch::
>  +
>  ------------
> -$ git branch topic/wip     <1>
> -$ git reset --hard HEAD~3  <2>
> -$ git checkout topic/wip   <3>
> +$ git branch topic/wip          <1>
> +$ git reset --hard HEAD~3       <2>
> +$ git switch-branch topic/wip   <3>
>  ------------
>  +
>  <1> You have made some commits, but realize they were premature
> @@ -232,13 +232,13 @@ working tree are not in any shape to be committed yet, but you
>  need to get to the other branch for a quick bugfix.
>  +
>  ------------
> -$ git checkout feature ;# you were working in "feature" branch and
> +$ git switch-branch feature ;# you were working in "feature" branch and
>  $ work work work       ;# got interrupted
>  $ git commit -a -m "snapshot WIP"                 <1>
> -$ git checkout master
> +$ git switch-branch master
>  $ fix fix fix
>  $ git commit ;# commit with real log
> -$ git checkout feature
> +$ git switch-branch feature
>  $ git reset --soft HEAD^ ;# go back to WIP state  <2>
>  $ git reset                                       <3>
>  ------------
> @@ -279,18 +279,18 @@ reset it while keeping the changes in your working tree.
>  +
>  ------------
>  $ git tag start
> -$ git checkout -b branch1
> +$ git switch-branch -b branch1
>  $ edit
>  $ git commit ...                            <1>
>  $ edit
> -$ git checkout -b branch2                   <2>
> +$ git switch-branch -b branch2              <2>
>  $ git reset --keep start                    <3>
>  ------------
>  +
>  <1> This commits your first edits in branch1.
>  <2> In the ideal world, you could have realized that the earlier
>      commit did not belong to the new topic when you created and switched
> -    to branch2 (i.e. "git checkout -b branch2 start"), but nobody is
> +    to branch2 (i.e. "git switch-branch -b branch2 start"), but nobody is
>      perfect.
>  <3> But you can use "reset --keep" to remove the unwanted commit after
>      you switched to "branch2".
> diff --git a/Documentation/git-revert.txt b/Documentation/git-revert.txt
> index 837707a8fd..e49dbbec83 100644
> --- a/Documentation/git-revert.txt
> +++ b/Documentation/git-revert.txt
> @@ -26,7 +26,7 @@ effect of some earlier commits (often only a faulty one).  If you want to
>  throw away all uncommitted changes in your working directory, you
>  should see linkgit:git-reset[1], particularly the `--hard` option.  If
>  you want to extract specific files as they were in another commit, you
> -should see linkgit:git-checkout[1], specifically the `git checkout
> +should see linkgit:git-checkout[1], specifically the `git checkout-files
>  <commit> -- <filename>` syntax.  Take care with these alternatives as
>  both will discard uncommitted changes in your working directory.
>  
> diff --git a/Documentation/git-stash.txt b/Documentation/git-stash.txt
> index 7ef8c47911..ea226979b1 100644
> --- a/Documentation/git-stash.txt
> +++ b/Documentation/git-stash.txt
> @@ -235,12 +235,12 @@ return to your original branch to make the emergency fix, like this:
>  +
>  ----------------------------------------------------------------
>  # ... hack hack hack ...
> -$ git checkout -b my_wip
> +$ git switch-branch -b my_wip
>  $ git commit -a -m "WIP"
> -$ git checkout master
> +$ git switch-branch master
>  $ edit emergency fix
>  $ git commit -a -m "Fix in a hurry"
> -$ git checkout my_wip
> +$ git switch-branch my_wip
>  $ git reset --soft HEAD^
>  # ... continue hacking ...
>  ----------------------------------------------------------------
> diff --git a/Documentation/gitattributes.txt b/Documentation/gitattributes.txt
> index b8392fc330..df62bd8019 100644
> --- a/Documentation/gitattributes.txt
> +++ b/Documentation/gitattributes.txt
> @@ -112,7 +112,7 @@ Checking-out and checking-in
>  
>  These attributes affect how the contents stored in the
>  repository are copied to the working tree files when commands
> -such as 'git checkout' and 'git merge' run.  They also affect how
> +such as 'git switch-branch' and 'git merge' run.  They also affect how
>  Git stores the contents you prepare in the working tree in the
>  repository upon 'git add' and 'git commit'.
>  
> diff --git a/Documentation/gitcli.txt b/Documentation/gitcli.txt
> index 592e06d839..0ad4869f2c 100644
> --- a/Documentation/gitcli.txt
> +++ b/Documentation/gitcli.txt
> @@ -47,8 +47,8 @@ disambiguating `--` at appropriate places.
>     things:
>  +
>  --------------------------------
> -$ git checkout -- *.c
> -$ git checkout -- \*.c
> +$ git checkout-files -- *.c
> +$ git checkout-files -- \*.c
>  --------------------------------
>  +
>  The former lets your shell expand the fileglob, and you are asking
> diff --git a/Documentation/gitcore-tutorial.txt b/Documentation/gitcore-tutorial.txt
> index e29a9effcc..49a8b5aa52 100644
> --- a/Documentation/gitcore-tutorial.txt
> +++ b/Documentation/gitcore-tutorial.txt
> @@ -741,7 +741,7 @@ used earlier, and create a branch in it. You do that by simply just
>  saying that you want to check out a new branch:
>  
>  ------------
> -$ git checkout -b mybranch
> +$ git branch mybranch
>  ------------
>  
>  will create a new branch based at the current `HEAD` position, and switch
> @@ -755,7 +755,7 @@ just telling 'git checkout' what the base of the checkout would be.
>  In other words, if you have an earlier tag or branch, you'd just do
>  
>  ------------
> -$ git checkout -b mybranch earlier-commit
> +$ git switch-branch -b mybranch earlier-commit
>  ------------
>  
>  and it would create the new branch `mybranch` at the earlier commit,
> @@ -765,7 +765,7 @@ and check out the state at that time.
>  You can always just jump back to your original `master` branch by doing
>  
>  ------------
> -$ git checkout master
> +$ git switch-branch master
>  ------------
>  
>  (or any other branch-name, for that matter) and if you forget which
> @@ -794,7 +794,7 @@ $ git branch <branchname> [startingpoint]
>  
>  which will simply _create_ the branch, but will not do anything further.
>  You can then later -- once you decide that you want to actually develop
> -on that branch -- switch to that branch with a regular 'git checkout'
> +on that branch -- switch to that branch with a regular 'git switch-branch
>  with the branchname as the argument.
>  
>  
> @@ -808,7 +808,7 @@ being the same as the original `master` branch, let's make sure we're in
>  that branch, and do some work there.
>  
>  ------------------------------------------------
> -$ git checkout mybranch
> +$ git switch-branch mybranch
>  $ echo "Work, work, work" >>hello
>  $ git commit -m "Some work." -i hello
>  ------------------------------------------------
> @@ -825,7 +825,7 @@ does some work in the original branch, and simulate that by going back
>  to the master branch, and editing the same file differently there:
>  
>  ------------
> -$ git checkout master
> +$ git switch-branch master
>  ------------
>  
>  Here, take a moment to look at the contents of `hello`, and notice how they
> @@ -958,7 +958,7 @@ to the `master` branch. Let's go back to `mybranch`, and run
>  'git merge' to get the "upstream changes" back to your branch.
>  
>  ------------
> -$ git checkout mybranch
> +$ git switch-branch mybranch
>  $ git merge -m "Merge upstream changes." master
>  ------------
>  
> @@ -1133,9 +1133,9 @@ Remember, before running 'git merge', our `master` head was at
>  work." commit.
>  
>  ------------
> -$ git checkout mybranch
> +$ git switch-branch mybranch
>  $ git reset --hard master^2
> -$ git checkout master
> +$ git switch-branch master
>  $ git reset --hard master^
>  ------------
>  
> diff --git a/Documentation/giteveryday.txt b/Documentation/giteveryday.txt
> index 9f2528fc8c..861b2bb616 100644
> --- a/Documentation/giteveryday.txt
> +++ b/Documentation/giteveryday.txt
> @@ -80,9 +80,9 @@ $ git tag v2.43 <2>
>  Create a topic branch and develop.::
>  +
>  ------------
> -$ git checkout -b alsa-audio <1>
> +$ git branch alsa-audio <1>
>  $ edit/compile/test
> -$ git checkout -- curses/ux_audio_oss.c <2>
> +$ git checkout-files -- curses/ux_audio_oss.c <2>
>  $ git add curses/ux_audio_alsa.c <3>
>  $ edit/compile/test
>  $ git diff HEAD <4>
> @@ -90,7 +90,7 @@ $ git commit -a -s <5>
>  $ edit/compile/test
>  $ git diff HEAD^ <6>
>  $ git commit -a --amend <7>
> -$ git checkout master <8>
> +$ git switch-branch master <8>
>  $ git merge alsa-audio <9>
>  $ git log --since='3 days ago' <10>
>  $ git log v2.43.. curses/ <11>
> @@ -148,11 +148,11 @@ Clone the upstream and work on it.  Feed changes to upstream.::
>  ------------
>  $ git clone git://git.kernel.org/pub/scm/.../torvalds/linux-2.6 my2.6
>  $ cd my2.6
> -$ git checkout -b mine master <1>
> +$ git switch-branch -b mine master <1>
>  $ edit/compile/test; git commit -a -s <2>
>  $ git format-patch master <3>
>  $ git send-email --to="person <email@example.com>" 00*.patch <4>
> -$ git checkout master <5>
> +$ git switch-branch master <5>
>  $ git pull <6>
>  $ git log -p ORIG_HEAD.. arch/i386 include/asm-i386 <7>
>  $ git ls-remote --heads http://git.kernel.org/.../jgarzik/libata-dev.git <8>
> @@ -194,7 +194,7 @@ satellite$ edit/compile/test/commit
>  satellite$ git push origin <4>
>  
>  mothership$ cd frotz
> -mothership$ git checkout master
> +mothership$ git switch-branch master
>  mothership$ git merge satellite/master <5>
>  ------------
>  +
> @@ -216,7 +216,7 @@ machine into the master branch.
>  Branch off of a specific tag.::
>  +
>  ------------
> -$ git checkout -b private2.6.14 v2.6.14 <1>
> +$ git switch-branch -b private2.6.14 v2.6.14 <1>
>  $ edit/compile/test; git commit -a
>  $ git checkout master
>  $ git cherry-pick v2.6.14..private2.6.14 <2>
> @@ -274,14 +274,14 @@ $ mailx <3>
>  & s 2 3 4 5 ./+to-apply
>  & s 7 8 ./+hold-linus
>  & q
> -$ git checkout -b topic/one master
> +$ git switch-branch -b topic/one master
>  $ git am -3 -i -s ./+to-apply <4>
>  $ compile/test
> -$ git checkout -b hold/linus && git am -3 -i -s ./+hold-linus <5>
> -$ git checkout topic/one && git rebase master <6>
> -$ git checkout pu && git reset --hard next <7>
> +$ git switch-branch -b hold/linus && git am -3 -i -s ./+hold-linus <5>
> +$ git switch-branch topic/one && git rebase master <6>
> +$ git switch-branch pu && git reset --hard next <7>
>  $ git merge topic/one topic/two && git merge hold/linus <8>
> -$ git checkout maint
> +$ git switch-branch maint
>  $ git cherry-pick master~4 <9>
>  $ compile/test
>  $ git tag -s -m "GIT 0.99.9x" v0.99.9x <10>
> diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt
> index 959044347e..3939ec774a 100644
> --- a/Documentation/githooks.txt
> +++ b/Documentation/githooks.txt
> @@ -166,7 +166,7 @@ worktree.  The hook is given three parameters: the ref of the previous HEAD,
>  the ref of the new HEAD (which may or may not have changed), and a flag
>  indicating whether the checkout was a branch checkout (changing branches,
>  flag=1) or a file checkout (retrieving a file from the index, flag=0).
> -This hook cannot affect the outcome of `git checkout`.
> +This hook cannot affect the outcome of `git switch-branch` or `git checkout`.
>  
>  It is also run after linkgit:git-clone[1], unless the `--no-checkout` (`-n`) option is
>  used. The first parameter given to the hook is the null-ref, the second the
> @@ -402,7 +402,8 @@ exit with a zero status.
>  For example, the hook can simply run `git read-tree -u -m HEAD "$1"`
>  in order to emulate `git fetch` that is run in the reverse direction
>  with `git push`, as the two-tree form of `git read-tree -u -m` is
> -essentially the same as `git checkout` that switches branches while
> +essentially the same as `git switch-branch` or `git checkout`
> +that switches branches while
>  keeping the local changes in the working tree that do not interfere
>  with the difference between the branches.
>  
> diff --git a/Documentation/gittutorial-2.txt b/Documentation/gittutorial-2.txt
> index e0976f6017..125213d951 100644
> --- a/Documentation/gittutorial-2.txt
> +++ b/Documentation/gittutorial-2.txt
> @@ -376,7 +376,7 @@ Changes to be committed:
>  
>  Changes not staged for commit:
>    (use "git add <file>..." to update what will be committed)
> -  (use "git checkout -- <file>..." to discard changes in working directory)
> +  (use "git checkout-files -- <file>..." to discard changes in working directory)
>  
>  	modified:   file.txt
>  
> diff --git a/Documentation/gittutorial.txt b/Documentation/gittutorial.txt
> index 242de31cb6..396e55c191 100644
> --- a/Documentation/gittutorial.txt
> +++ b/Documentation/gittutorial.txt
> @@ -207,7 +207,7 @@ automatically.  The asterisk marks the branch you are currently on;
>  type
>  
>  ------------------------------------------------
> -$ git checkout experimental
> +$ git switch-branch experimental
>  ------------------------------------------------
>  
>  to switch to the experimental branch.  Now edit a file, commit the
> @@ -216,7 +216,7 @@ change, and switch back to the master branch:
>  ------------------------------------------------
>  (edit file)
>  $ git commit -a
> -$ git checkout master
> +$ git switch-branch master
>  ------------------------------------------------
>  
>  Check that the change you made is no longer visible, since it was
> diff --git a/Documentation/revisions.txt b/Documentation/revisions.txt
> index 72daa20e76..f55502cd50 100644
> --- a/Documentation/revisions.txt
> +++ b/Documentation/revisions.txt
> @@ -115,7 +115,7 @@ Here's an example to make it more clear:
>  ------------------------------
>  $ git config push.default current
>  $ git config remote.pushdefault myfork
> -$ git checkout -b mybranch origin/master
> +$ git switch-branch -b mybranch origin/master
>  
>  $ git rev-parse --symbolic-full-name @{upstream}
>  refs/remotes/origin/master
> diff --git a/Documentation/user-manual.txt b/Documentation/user-manual.txt
> index eff7890274..56397e93d4 100644
> --- a/Documentation/user-manual.txt
> +++ b/Documentation/user-manual.txt
> @@ -125,7 +125,7 @@ Create a new branch head pointing to one of these versions and check it
>  out using linkgit:git-checkout[1]:
>  
>  ------------------------------------------------
> -$ git checkout -b new v2.6.13
> +$ git switch-branch -b new v2.6.13
>  ------------------------------------------------
>  
>  The working directory then reflects the contents that the project had
> @@ -282,10 +282,10 @@ a summary of the commands:
>  	this command will fail with a warning.
>  `git branch -D <branch>`::
>  	delete the branch `<branch>` irrespective of its merged status.
> -`git checkout <branch>`::
> +`git switch-branch <branch>`::
>  	make the current branch `<branch>`, updating the working
>  	directory to reflect the version referenced by `<branch>`.
> -`git checkout -b <new> <start-point>`::
> +`git switch-branch -b <new> <start-point>`::
>  	create a new branch `<new>` referencing `<start-point>`, and
>  	check it out.
>  
> @@ -302,12 +302,12 @@ ref: refs/heads/master
>  Examining an old version without creating a new branch
>  ------------------------------------------------------
>  
> -The `git checkout` command normally expects a branch head, but will also
> +The `git switch-branch` command normally expects a branch head, but will also
>  accept an arbitrary commit; for example, you can check out the commit
>  referenced by a tag:
>  
>  ------------------------------------------------
> -$ git checkout v2.6.17
> +$ git switch-branch v2.6.17
>  Note: checking out 'v2.6.17'.
>  
>  You are in 'detached HEAD' state. You can look around, make experimental
> @@ -317,7 +317,7 @@ state without impacting any branches by performing another checkout.
>  If you want to create a new branch to retain commits you create, you may
>  do so (now or later) by using -b with the checkout command again. Example:
>  
> -  git checkout -b new_branch_name
> +  git switch-branch -b new_branch_name
>  
>  HEAD is now at 427abfa Linux v2.6.17
>  ------------------------------------------------
> @@ -373,7 +373,7 @@ You might want to build on one of these remote-tracking branches
>  on a branch of your own, just as you would for a tag:
>  
>  ------------------------------------------------
> -$ git checkout -b my-todo-copy origin/todo
> +$ git switch-branch -b my-todo-copy origin/todo
>  ------------------------------------------------
>  
>  You can also check out `origin/todo` directly to examine it or
> @@ -1523,12 +1523,12 @@ Checking out an old version of a file
>  
>  In the process of undoing a previous bad change, you may find it
>  useful to check out an older version of a particular file using
> -linkgit:git-checkout[1].  We've used `git checkout` before to switch
> +linkgit:git-checkout[1].  We've used `git switch-branch` before to switch
>  branches, but it has quite different behavior if it is given a path
>  name: the command
>  
>  -------------------------------------------------
> -$ git checkout HEAD^ path/to/file
> +$ git checkout-files HEAD^ path/to/file
>  -------------------------------------------------
>  
>  replaces path/to/file by the contents it had in the commit HEAD^, and
> @@ -2211,8 +2211,8 @@ $ git branch --track release origin/master
>  These can be easily kept up to date using linkgit:git-pull[1].
>  
>  -------------------------------------------------
> -$ git checkout test && git pull
> -$ git checkout release && git pull
> +$ git switch-branch test && git pull
> +$ git switch-branch release && git pull
>  -------------------------------------------------
>  
>  Important note!  If you have any local changes in these branches, then
> @@ -2264,7 +2264,7 @@ tested changes
>  2) help future bug hunters that use `git bisect` to find problems
>  
>  -------------------------------------------------
> -$ git checkout -b speed-up-spinlocks v2.6.35
> +$ git switch-branch -b speed-up-spinlocks v2.6.35
>  -------------------------------------------------
>  
>  Now you apply the patch(es), run some tests, and commit the change(s).  If
> @@ -2279,7 +2279,7 @@ When you are happy with the state of this change, you can merge it into the
>  "test" branch in preparation to make it public:
>  
>  -------------------------------------------------
> -$ git checkout test && git merge speed-up-spinlocks
> +$ git switch-branch test && git merge speed-up-spinlocks
>  -------------------------------------------------
>  
>  It is unlikely that you would have any conflicts here ... but you might if you
> @@ -2291,7 +2291,7 @@ see the value of keeping each patch (or patch series) in its own branch.  It
>  means that the patches can be moved into the `release` tree in any order.
>  
>  -------------------------------------------------
> -$ git checkout release && git merge speed-up-spinlocks
> +$ git switch-branch release && git merge speed-up-spinlocks
>  -------------------------------------------------
>  
>  After a while, you will have a number of branches, and despite the
> @@ -2358,7 +2358,7 @@ Here are some of the scripts that simplify all this even further.
>  
>  case "$1" in
>  test|release)
> -	git checkout $1 && git pull . origin
> +	git switch-branch $1 && git pull . origin
>  	;;
>  origin)
>  	before=$(git rev-parse refs/remotes/origin/master)
> @@ -2400,7 +2400,7 @@ test|release)
>  		echo $1 already merged into $2 1>&2
>  		exit 1
>  	fi
> -	git checkout $2 && git pull . $1
> +	git switch-branch $2 && git pull . $1
>  	;;
>  *)
>  	usage
> @@ -2512,7 +2512,7 @@ Suppose that you create a branch `mywork` on a remote-tracking branch
>  `origin`, and create some commits on top of it:
>  
>  -------------------------------------------------
> -$ git checkout -b mywork origin
> +$ git switch-branch -b mywork origin
>  $ vi file.txt
>  $ git commit
>  $ vi otherfile.txt
> @@ -2552,7 +2552,7 @@ commits without any merges, you may instead choose to use
>  linkgit:git-rebase[1]:
>  
>  -------------------------------------------------
> -$ git checkout mywork
> +$ git switch-branch mywork
>  $ git rebase origin
>  -------------------------------------------------
>  
> @@ -3668,13 +3668,13 @@ change within the submodule, and then update the superproject to reference the
>  new commit:
>  
>  -------------------------------------------------
> -$ git checkout master
> +$ git switch-branch master
>  -------------------------------------------------
>  
>  or
>  
>  -------------------------------------------------
> -$ git checkout -b fix-up
> +$ git switch-branch -b fix-up
>  -------------------------------------------------
>  
>  then
> @@ -4194,7 +4194,7 @@ start.
>  A good place to start is with the contents of the initial commit, with:
>  
>  ----------------------------------------------------
> -$ git checkout e83c5163
> +$ git switch-branch e83c5163
>  ----------------------------------------------------
>  
>  The initial revision lays the foundation for almost everything Git has
> @@ -4437,10 +4437,10 @@ Managing branches
>  -----------------
>  
>  -----------------------------------------------
> -$ git branch	     # list all local branches in this repo
> -$ git checkout test  # switch working directory to branch "test"
> -$ git branch new     # create branch "new" starting at current HEAD
> -$ git branch -d new  # delete branch "new"
> +$ git branch			# list all local branches in this repo
> +$ git switch-branch test	# switch working directory to branch "test"
> +$ git branch new		# create branch "new" starting at current HEAD
> +$ git branch -d new		# delete branch "new"
>  -----------------------------------------------
>  
>  Instead of basing a new branch on current HEAD (the default), use:
> @@ -4456,7 +4456,7 @@ $ git branch new test~10 # ten commits before tip of branch "test"
>  Create and switch to a new branch at the same time:
>  
>  -----------------------------------------------
> -$ git checkout -b new v2.6.15
> +$ git switch-branch -b new v2.6.15
>  -----------------------------------------------
>  
>  Update and examine branches from the repository you cloned from:
> @@ -4467,7 +4467,7 @@ $ git branch -r		# list
>    origin/master
>    origin/next
>    ...
> -$ git checkout -b masterwork origin/master
> +$ git switch-branch -b masterwork origin/master
>  -----------------------------------------------
>  
>  Fetch a branch from a different repository, and give it a new
> diff --git a/advice.c b/advice.c
> index 5f35656409..1befdb2163 100644
> --- a/advice.c
> +++ b/advice.c
> @@ -195,7 +195,7 @@ void detach_advice(const char *new_name)
>  	"state without impacting any branches by performing another checkout.\n\n"
>  	"If you want to create a new branch to retain commits you create, you may\n"
>  	"do so (now or later) by using -b with the checkout command again. Example:\n\n"
> -	"  git checkout -b <new-branch-name>\n\n");
> +	"  git branch <new-branch-name>\n\n");
>  
>  	fprintf(stderr, fmt, new_name);
>  }
> diff --git a/sha1-name.c b/sha1-name.c
> index faa60f69e3..4e4e14a45c 100644
> --- a/sha1-name.c
> +++ b/sha1-name.c
> @@ -771,7 +771,7 @@ static int get_oid_basic(const char *str, int len, struct object_id *oid,
>  	"because it will be ignored when you just specify 40-hex. These refs\n"
>  	"may be created by mistake. For example,\n"
>  	"\n"
> -	"  git checkout -b $br $(git rev-parse ...)\n"
> +	"  git switch-branch -b $br $(git rev-parse ...)\n"
>  	"\n"
>  	"where \"$br\" is somehow empty and a 40-hex ref is created. Please\n"
>  	"examine these refs and maybe delete them. Turn this message off by\n"
> diff --git a/wt-status.c b/wt-status.c
> index a24711374c..6266683926 100644
> --- a/wt-status.c
> +++ b/wt-status.c
> @@ -224,7 +224,7 @@ static void wt_longstatus_print_dirty_header(struct wt_status *s,
>  		status_printf_ln(s, c, _("  (use \"git add <file>...\" to update what will be committed)"));
>  	else
>  		status_printf_ln(s, c, _("  (use \"git add/rm <file>...\" to update what will be committed)"));
> -	status_printf_ln(s, c, _("  (use \"git checkout -- <file>...\" to discard changes in working directory)"));
> +	status_printf_ln(s, c, _("  (use \"git checkout-files <file>...\" to discard changes in working directory)"));
>  	if (has_dirty_submodules)
>  		status_printf_ln(s, c, _("  (commit or discard the untracked or modified content in submodules)"));
>  	status_printf_ln(s, c, "%s", "");

Re: [PATCH v2 1/7] parse-options: allow parse_options_concat(NULL, options)

[Duy Nguyen]

On Tue, Nov 27, 2018 at 8:44 PM Stefan Beller <sbeller@google.com> wrote:
>
> On Tue, Nov 27, 2018 at 8:53 AM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
> >
> > There is currently no caller that calls this function with "a" being
> > NULL. But it will be introduced shortly. It is used to construct the
> > option array from scratch, e.g.
> >
> >    struct parse_options opts = NULL;
> >    opts = parse_options_concat(opts, opts_1);
> >    opts = parse_options_concat(opts, opts_2);
>
> While this addresses the immediate needs, I'd prefer to think
> about the API exposure of parse_options_concat,
> (related: do we want to have docs in its header file?)
> and I'd recommend to make it symmetrical, i.e.
> allow the second argument to also be NULL?

I'll just drop this patch. There's a better way to do the same without
adding special handling like this.
-- 
Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Duy Nguyen]

On Wed, Nov 28, 2018 at 7:03 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>
> > The good old "git checkout" command is still here and will be until
> > all (or most of users) are sick of it.
>
> Two comments on the goal (the implementation looked reasonable
> assuming the reader agrees with the gaol).
>
> At least to me, the verb "switch" needs two things to switch
> between, i.e. "switch A and B", unless it is "switch to X".
> Either "switch-to-branch" or simply "switch-to", perhaps?
>
> As I already hinted in my response to Stefan (?) about
> checkout-from-tree vs checkout-from-index, a command with multiple
> modes of operation is not confusing to people with the right mental
> model, and I suspect that having two separate commands for "checking
> out a branch" and "checking out paths" that is done by this step
> would help users to form the right mental model.

Since the other one is already "checkout-files", maybe this one could
just be "checkout-branch".

> So I tend to think
> these two are "training wheels", and suspect that once they got it,
> nobody will become "sick of" the single "checkout" command that can
> be used to do either.  It's just the matter of being aware what can
> be done (which requires the right mental model) and how to tell Git
> what the user wants it do (two separate commands, operating mode
> option, or just the implied command line syntax---once the user
> knows what s/he is doing, these do not make that much a difference).

I would hope this becomes better defaults and being used 90% of time.
Even though I know "git checkout" quite well, it still bites me from
time to time. Having the right mental model is one thing. Having to
think a bit every time to write "git checkout" with the right syntax,
and whether you need "--" (that ambiguation problem can still bite you
from time to time), is frankly something I'd rather avoid.
-- 
Duy

Re: [PATCH v2 7/7] Suggest other commands instead of "git checkout"

[Duy Nguyen]

On Wed, Nov 28, 2018 at 7:04 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>
> > The assumption made is here
> >
> > - "git checkout" is a horrible monster that should only be touched
> >   with a two-meter pole
> >
> > - there are other commands that can achieve the same thing
>
> Thanks for clearly spelling out the assumptions.  It is good that
> this step cames at the end, as the earlier 6 steps looked reasonable
> to me.

I see my deliberate attempt to provoke has failed :D Giving your view
of the new commands as "training wheels", I take it we still should
make them visible as much as possible, but we just not try to hide
"git checkout" as much (e.g. we mention both new and old commands,
instead of just mentioning the new one, when suggesting something)?
-- 
Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Stefan Beller]

On Wed, Nov 28, 2018 at 7:31 AM Duy Nguyen <pclouds@gmail.com> wrote:
>
> On Wed, Nov 28, 2018 at 7:03 AM Junio C Hamano <gitster@pobox.com> wrote:
> >
> > Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
> >
> > > The good old "git checkout" command is still here and will be until
> > > all (or most of users) are sick of it.
> >
> > Two comments on the goal (the implementation looked reasonable
> > assuming the reader agrees with the gaol).
> >
> > At least to me, the verb "switch" needs two things to switch
> > between, i.e. "switch A and B", unless it is "switch to X".
> > Either "switch-to-branch" or simply "switch-to", perhaps?
> >
> > As I already hinted in my response to Stefan (?) about
> > checkout-from-tree vs checkout-from-index, a command with multiple
> > modes of operation is not confusing to people with the right mental
> > model, and I suspect that having two separate commands for "checking
> > out a branch" and "checking out paths" that is done by this step
> > would help users to form the right mental model.
>
> Since the other one is already "checkout-files", maybe this one could
> just be "checkout-branch".

I dislike the checkout-* names, as we already have checkout-index
as plumbing, so it would be confusing as to which checkout-* command
should be used when and why as it seems the co-index moves
content *from index* to the working tree, but the co-files moves content
*to files*, whereas checkout-branch is neither 'moving' to or from a branch
but rather 'switching' to that branch.

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Duy Nguyen]

On Wed, Nov 28, 2018 at 8:08 PM Stefan Beller <sbeller@google.com> wrote:
>
> On Wed, Nov 28, 2018 at 7:31 AM Duy Nguyen <pclouds@gmail.com> wrote:
> >
> > On Wed, Nov 28, 2018 at 7:03 AM Junio C Hamano <gitster@pobox.com> wrote:
> > >
> > > Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
> > >
> > > > The good old "git checkout" command is still here and will be until
> > > > all (or most of users) are sick of it.
> > >
> > > Two comments on the goal (the implementation looked reasonable
> > > assuming the reader agrees with the gaol).
> > >
> > > At least to me, the verb "switch" needs two things to switch
> > > between, i.e. "switch A and B", unless it is "switch to X".
> > > Either "switch-to-branch" or simply "switch-to", perhaps?
> > >
> > > As I already hinted in my response to Stefan (?) about
> > > checkout-from-tree vs checkout-from-index, a command with multiple
> > > modes of operation is not confusing to people with the right mental
> > > model, and I suspect that having two separate commands for "checking
> > > out a branch" and "checking out paths" that is done by this step
> > > would help users to form the right mental model.
> >
> > Since the other one is already "checkout-files", maybe this one could
> > just be "checkout-branch".
>
> I dislike the checkout-* names, as we already have checkout-index
> as plumbing, so it would be confusing as to which checkout-* command
> should be used when and why as it seems the co-index moves
> content *from index* to the working tree, but the co-files moves content
> *to files*, whereas checkout-branch is neither 'moving' to or from a branch
> but rather 'switching' to that branch.

OK back to square one. Another thing I noticed, not sure if it
matters, is that these two commands will be the only ones with a
hyphen in them in "git help". But I guess it's even harder to find
one-word command names for these.
-- 
Duy

Re: [PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Duy Nguyen]

On Tue, Nov 27, 2018 at 5:53 PM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
>
> v2 is just a bit better to look at than v1. This is by no means final.
> If you think the command name is bad, the default behavior should
> change, or something else, speak up. It's still very "RFC".
>
> v2 breaks down the giant patch in v1 and starts adding some changes in
> these new commands:
>
> - restore-paths is renamed to checkout-paths. I wrote I didn't like
>   "checkout" because of completion conflict. But who am I kidding,
>   I'll use aliases anyway. "-files" instead of "-paths" because we
>   already have ls-files.
> - both commands will not accept no arguments. There is no "git
>   checkout" equivalent.
> - ambiguation rules are now aware that "switch-branch" for example
>   can't take pathspec...

Another thing. Since we start with a clean slate, should we do
something about detached HEAD in this switch-branch command (or
whatever its name will be)?

This is usually a confusing concept to new users and I've seen some
users have a hard time getting out of it. I'm too comfortable with
detached HEAD to see this from new user's perspective and a better way
to deal with it.
-- 
Duy

Re: [PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Duy Nguyen]

On Wed, Nov 28, 2018 at 9:01 PM Duy Nguyen <pclouds@gmail.com> wrote:
> should we do
> something about detached HEAD in this switch-branch command (or
> whatever its name will be)?
>
> This is usually a confusing concept to new users

And it just occurred to me that perhaps we should call this "unnamed
branch" (at least at high UI level) instead of detached HEAD. It is
technically not as accurate, but much better to understand.
-- 
Duy

Re: [PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Stefan Beller]

On Wed, Nov 28, 2018 at 12:09 PM Duy Nguyen <pclouds@gmail.com> wrote:
>
> On Wed, Nov 28, 2018 at 9:01 PM Duy Nguyen <pclouds@gmail.com> wrote:
> > should we do
> > something about detached HEAD in this switch-branch command (or
> > whatever its name will be)?
> >
> > This is usually a confusing concept to new users
>
> And it just occurred to me that perhaps we should call this "unnamed
> branch" (at least at high UI level) instead of detached HEAD. It is
> technically not as accurate, but much better to understand.

or 'direct' branch? I mean 'detached HEAD' itself is also not correct
as the HEAD points to a valid commit/tag usually, so it is attached to
that content. The detachment comes from the implicit "from a branch".

Re: [PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Stefan Xenos]

I think users have problems with detached heads for several reasons:

1. Users often enter the detached head state unexpectedly (for
example, by mistyping a "checkout" command or not understanding its
multipurpose nature, or as a side-effect of running a submodule
command). The change described here will go in the right direction
towards addressing this, since you won't be able to accidentally
detach your head by mistyping checkout. If detaching was always an
intentional action by the user, it would be much less intimidating.

2. Git shows a lot of scary error messages when running in a detached
head state. They tell you you're doing something dangerous, which
dissuades users from experimenting with the feature. IMO, it would be
better to replace the scary warning messages with instructions on
where to get more information about detached heads (and those
instructions should frontload info on how to reattach to a branch and
how to recover commits from the reflog).

3. When in a detached head state, a lot of commands add extra
confirmation prompts, which are unnecessary and intimidating.

Basically, the current user interface implies to the user that a
detached head is an error condition they'll need to recover from
rather than the normal and useful mode of working that it is.

(I'm resending this email without html - sorry if you received it twice).

So - IMO - detaching should always be an explicit action. Some options
that occur to me:

git switch-branch --detach
git detach
git switch-branch HEAD

  - Stefan

On Wed, Nov 28, 2018 at 2:48 PM Stefan Xenos <sxenos@google.com> wrote:
>
> I think users have problems with detached heads for several reasons:
>
> 1. Users often enter the detached head state unexpectedly (for example, by mistyping a "checkout" command or not understanding its multipurpose nature, or as a side-effect of running a submodule command). The change described here will go in the right direction towards addressing this, since you won't be able to accidentally detach your head by mistyping checkout. If detaching was always an intentional action by the user, it would be much less intimidating.
>
> 2. Git shows a lot of scary error messages when running in a detached head state. They tell you you're doing something dangerous, which dissuades users from experimenting with the feature. IMO, it would be better to replace the scary warning messages with instructions on where to get more information about detached heads (and those instructions should frontload info on how to reattach to a branch and how to recover commits from the reflog).
>
> 3. When in a detached head state, a lot of commands add extra confirmation prompts, which are unnecessary and intimidating.
>
> Basically, the current user interface implies to the user that a detached head is an error condition they'll need to recover from rather than the normal and useful mode of working that it is.
>
>   - Stefan
>
> On Wed, Nov 28, 2018 at 12:01 PM Duy Nguyen <pclouds@gmail.com> wrote:
>>
>> On Tue, Nov 27, 2018 at 5:53 PM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
>> >
>> > v2 is just a bit better to look at than v1. This is by no means final.
>> > If you think the command name is bad, the default behavior should
>> > change, or something else, speak up. It's still very "RFC".
>> >
>> > v2 breaks down the giant patch in v1 and starts adding some changes in
>> > these new commands:
>> >
>> > - restore-paths is renamed to checkout-paths. I wrote I didn't like
>> >   "checkout" because of completion conflict. But who am I kidding,
>> >   I'll use aliases anyway. "-files" instead of "-paths" because we
>> >   already have ls-files.
>> > - both commands will not accept no arguments. There is no "git
>> >   checkout" equivalent.
>> > - ambiguation rules are now aware that "switch-branch" for example
>> >   can't take pathspec...
>>
>> Another thing. Since we start with a clean slate, should we do
>> something about detached HEAD in this switch-branch command (or
>> whatever its name will be)?
>>
>> This is usually a confusing concept to new users and I've seen some
>> users have a hard time getting out of it. I'm too comfortable with
>> detached HEAD to see this from new user's perspective and a better way
>> to deal with it.
>> --
>> Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Stefan Xenos]

> Since the other one is already "checkout-files", maybe this one could just be "checkout-branch".

I rather like switch-branch and dislike the word "checkout" since it
has been overloaded in git for so long (does it mean moving HEAD or
copying files to my working tree?)

> nobody will become "sick of" the single "checkout" command that can

I have to admit I'm already sick of the checkout command. :-p I can
see myself using these two new commands 100% of the time and never
missing the old one.

Some behaviors I'd expect to see from these commands (I haven't yet
checked to see if you've already done this):

git checkout-files <tree-ish>
should reset all the files in the repository regardless of the current
directory - it should produce the same effect as "git reset --hard
<tree-ish> && git reset HEAD@{1}". It should also delete
locally-created files that aren't present in <tree-ish>, such that the
final working tree is exactly identical to what was committed in that
tree-ish.

git checkout-files foo -- myfile.txt
should delete myfile.txt if it is present locally but not present in foo.

git checkout-files foo -- .
should recursively checkout all files in the current folder and all
subfolders, and delete any locally-created files if they're not
present in foo.

git checkout-files should never move HEAD in any circumstance.

Suggestion:
If git checkout-files overwrites or deletes any locally-modified files
from the workspace or index, those files could be auto-stashed. That
would make it easy to restore them in the event of a mistyped command.
Auto-stashing could be suppressed with a command-line argument (with
alternate behaviors being fail-if-modified or always-overwrite).

Idea:
If git checkout-files modifies the submodules file, it could also
auto-update the submodules. (For example, with something like "git
submodule update --init --recursive --progress").

  - Stefan
On Wed, Nov 28, 2018 at 7:31 AM Duy Nguyen <pclouds@gmail.com> wrote:
>
> On Wed, Nov 28, 2018 at 7:03 AM Junio C Hamano <gitster@pobox.com> wrote:
> >
> > Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
> >
> > > The good old "git checkout" command is still here and will be until
> > > all (or most of users) are sick of it.
> >
> > Two comments on the goal (the implementation looked reasonable
> > assuming the reader agrees with the gaol).
> >
> > At least to me, the verb "switch" needs two things to switch
> > between, i.e. "switch A and B", unless it is "switch to X".
> > Either "switch-to-branch" or simply "switch-to", perhaps?
> >
> > As I already hinted in my response to Stefan (?) about
> > checkout-from-tree vs checkout-from-index, a command with multiple
> > modes of operation is not confusing to people with the right mental
> > model, and I suspect that having two separate commands for "checking
> > out a branch" and "checking out paths" that is done by this step
> > would help users to form the right mental model.
>
> Since the other one is already "checkout-files", maybe this one could
> just be "checkout-branch".
>
> > So I tend to think
> > these two are "training wheels", and suspect that once they got it,
> > nobody will become "sick of" the single "checkout" command that can
> > be used to do either.  It's just the matter of being aware what can
> > be done (which requires the right mental model) and how to tell Git
> > what the user wants it do (two separate commands, operating mode
> > option, or just the implied command line syntax---once the user
> > knows what s/he is doing, these do not make that much a difference).
>
> I would hope this becomes better defaults and being used 90% of time.
> Even though I know "git checkout" quite well, it still bites me from
> time to time. Having the right mental model is one thing. Having to
> think a bit every time to write "git checkout" with the right syntax,
> and whether you need "--" (that ambiguation problem can still bite you
> from time to time), is frankly something I'd rather avoid.
> --
> Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Stefan Xenos]

Although I have no problem with "switch-branch" as a command name,
some alternative names we might consider for switch-branch might be:

chbranch
swbranch
switch
branch change (as a subcommand for the "branch" command)

I've personally been using "chbranch" as an alias for this
functionality for some time.

  - Stefan
On Wed, Nov 28, 2018 at 3:22 PM Stefan Xenos <sxenos@google.com> wrote:
>
> > Since the other one is already "checkout-files", maybe this one could just be "checkout-branch".
>
> I rather like switch-branch and dislike the word "checkout" since it
> has been overloaded in git for so long (does it mean moving HEAD or
> copying files to my working tree?)
>
> > nobody will become "sick of" the single "checkout" command that can
>
> I have to admit I'm already sick of the checkout command. :-p I can
> see myself using these two new commands 100% of the time and never
> missing the old one.
>
> Some behaviors I'd expect to see from these commands (I haven't yet
> checked to see if you've already done this):
>
> git checkout-files <tree-ish>
> should reset all the files in the repository regardless of the current
> directory - it should produce the same effect as "git reset --hard
> <tree-ish> && git reset HEAD@{1}". It should also delete
> locally-created files that aren't present in <tree-ish>, such that the
> final working tree is exactly identical to what was committed in that
> tree-ish.
>
> git checkout-files foo -- myfile.txt
> should delete myfile.txt if it is present locally but not present in foo.
>
> git checkout-files foo -- .
> should recursively checkout all files in the current folder and all
> subfolders, and delete any locally-created files if they're not
> present in foo.
>
> git checkout-files should never move HEAD in any circumstance.
>
> Suggestion:
> If git checkout-files overwrites or deletes any locally-modified files
> from the workspace or index, those files could be auto-stashed. That
> would make it easy to restore them in the event of a mistyped command.
> Auto-stashing could be suppressed with a command-line argument (with
> alternate behaviors being fail-if-modified or always-overwrite).
>
> Idea:
> If git checkout-files modifies the submodules file, it could also
> auto-update the submodules. (For example, with something like "git
> submodule update --init --recursive --progress").
>
>   - Stefan
> On Wed, Nov 28, 2018 at 7:31 AM Duy Nguyen <pclouds@gmail.com> wrote:
> >
> > On Wed, Nov 28, 2018 at 7:03 AM Junio C Hamano <gitster@pobox.com> wrote:
> > >
> > > Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
> > >
> > > > The good old "git checkout" command is still here and will be until
> > > > all (or most of users) are sick of it.
> > >
> > > Two comments on the goal (the implementation looked reasonable
> > > assuming the reader agrees with the gaol).
> > >
> > > At least to me, the verb "switch" needs two things to switch
> > > between, i.e. "switch A and B", unless it is "switch to X".
> > > Either "switch-to-branch" or simply "switch-to", perhaps?
> > >
> > > As I already hinted in my response to Stefan (?) about
> > > checkout-from-tree vs checkout-from-index, a command with multiple
> > > modes of operation is not confusing to people with the right mental
> > > model, and I suspect that having two separate commands for "checking
> > > out a branch" and "checking out paths" that is done by this step
> > > would help users to form the right mental model.
> >
> > Since the other one is already "checkout-files", maybe this one could
> > just be "checkout-branch".
> >
> > > So I tend to think
> > > these two are "training wheels", and suspect that once they got it,
> > > nobody will become "sick of" the single "checkout" command that can
> > > be used to do either.  It's just the matter of being aware what can
> > > be done (which requires the right mental model) and how to tell Git
> > > what the user wants it do (two separate commands, operating mode
> > > option, or just the implied command line syntax---once the user
> > > knows what s/he is doing, these do not make that much a difference).
> >
> > I would hope this becomes better defaults and being used 90% of time.
> > Even though I know "git checkout" quite well, it still bites me from
> > time to time. Having the right mental model is one thing. Having to
> > think a bit every time to write "git checkout" with the right syntax,
> > and whether you need "--" (that ambiguation problem can still bite you
> > from time to time), is frankly something I'd rather avoid.
> > --
> > Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Stefan Xenos]

More thoughts:

git switch-branch should never detach HEAD unless asked to do so
explicitly. That also means that "git switch-branch" shouldn't accept
any of the non-branch tree-ish arguments that would have caused "git
checkout" to do so.
On Wed, Nov 28, 2018 at 3:26 PM Stefan Xenos <sxenos@google.com> wrote:
>
> Although I have no problem with "switch-branch" as a command name,
> some alternative names we might consider for switch-branch might be:
>
> chbranch
> swbranch
> switch
> branch change (as a subcommand for the "branch" command)
>
> I've personally been using "chbranch" as an alias for this
> functionality for some time.
>
>   - Stefan
> On Wed, Nov 28, 2018 at 3:22 PM Stefan Xenos <sxenos@google.com> wrote:
> >
> > > Since the other one is already "checkout-files", maybe this one could just be "checkout-branch".
> >
> > I rather like switch-branch and dislike the word "checkout" since it
> > has been overloaded in git for so long (does it mean moving HEAD or
> > copying files to my working tree?)
> >
> > > nobody will become "sick of" the single "checkout" command that can
> >
> > I have to admit I'm already sick of the checkout command. :-p I can
> > see myself using these two new commands 100% of the time and never
> > missing the old one.
> >
> > Some behaviors I'd expect to see from these commands (I haven't yet
> > checked to see if you've already done this):
> >
> > git checkout-files <tree-ish>
> > should reset all the files in the repository regardless of the current
> > directory - it should produce the same effect as "git reset --hard
> > <tree-ish> && git reset HEAD@{1}". It should also delete
> > locally-created files that aren't present in <tree-ish>, such that the
> > final working tree is exactly identical to what was committed in that
> > tree-ish.
> >
> > git checkout-files foo -- myfile.txt
> > should delete myfile.txt if it is present locally but not present in foo.
> >
> > git checkout-files foo -- .
> > should recursively checkout all files in the current folder and all
> > subfolders, and delete any locally-created files if they're not
> > present in foo.
> >
> > git checkout-files should never move HEAD in any circumstance.
> >
> > Suggestion:
> > If git checkout-files overwrites or deletes any locally-modified files
> > from the workspace or index, those files could be auto-stashed. That
> > would make it easy to restore them in the event of a mistyped command.
> > Auto-stashing could be suppressed with a command-line argument (with
> > alternate behaviors being fail-if-modified or always-overwrite).
> >
> > Idea:
> > If git checkout-files modifies the submodules file, it could also
> > auto-update the submodules. (For example, with something like "git
> > submodule update --init --recursive --progress").
> >
> >   - Stefan
> > On Wed, Nov 28, 2018 at 7:31 AM Duy Nguyen <pclouds@gmail.com> wrote:
> > >
> > > On Wed, Nov 28, 2018 at 7:03 AM Junio C Hamano <gitster@pobox.com> wrote:
> > > >
> > > > Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
> > > >
> > > > > The good old "git checkout" command is still here and will be until
> > > > > all (or most of users) are sick of it.
> > > >
> > > > Two comments on the goal (the implementation looked reasonable
> > > > assuming the reader agrees with the gaol).
> > > >
> > > > At least to me, the verb "switch" needs two things to switch
> > > > between, i.e. "switch A and B", unless it is "switch to X".
> > > > Either "switch-to-branch" or simply "switch-to", perhaps?
> > > >
> > > > As I already hinted in my response to Stefan (?) about
> > > > checkout-from-tree vs checkout-from-index, a command with multiple
> > > > modes of operation is not confusing to people with the right mental
> > > > model, and I suspect that having two separate commands for "checking
> > > > out a branch" and "checking out paths" that is done by this step
> > > > would help users to form the right mental model.
> > >
> > > Since the other one is already "checkout-files", maybe this one could
> > > just be "checkout-branch".
> > >
> > > > So I tend to think
> > > > these two are "training wheels", and suspect that once they got it,
> > > > nobody will become "sick of" the single "checkout" command that can
> > > > be used to do either.  It's just the matter of being aware what can
> > > > be done (which requires the right mental model) and how to tell Git
> > > > what the user wants it do (two separate commands, operating mode
> > > > option, or just the implied command line syntax---once the user
> > > > knows what s/he is doing, these do not make that much a difference).
> > >
> > > I would hope this becomes better defaults and being used 90% of time.
> > > Even though I know "git checkout" quite well, it still bites me from
> > > time to time. Having the right mental model is one thing. Having to
> > > think a bit every time to write "git checkout" with the right syntax,
> > > and whether you need "--" (that ambiguation problem can still bite you
> > > from time to time), is frankly something I'd rather avoid.
> > > --
> > > Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Junio C Hamano]

Stefan Beller <sbeller@google.com> writes:

> I dislike the checkout-* names, as we already have checkout-index
> as plumbing, so it would be confusing as to which checkout-* command
> should be used when and why as it seems the co-index moves
> content *from index* to the working tree, but the co-files moves content
> *to files*, whereas checkout-branch is neither 'moving' to or from a branch
> but rather 'switching' to that branch.

To me, "switching to work on the branch", is like "checking the book
out from the library to read".  IOW, "check the branch out to work
on" does not have to involve *any* moving of contents.

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Junio C Hamano]

Stefan Xenos <sxenos@google.com> writes:

> Although I have no problem with "switch-branch" as a command name,
> some alternative names we might consider for switch-branch might be:
>
> chbranch
> swbranch

Please never go in that direction.  So far, we made a conscious
effort to keep the names of most frequently used subcommand to
proper words that can be understood by coders (IOW, I expect they
know what 'grep' is, even though that may not be a 'proper word').

> switch
> branch change (as a subcommand for the "branch" command)

It is more like moving to the branch to work on.  I think 'switch'
is something SVN veterans may find familiar.  Both are much better
than ch/swbranch that are overly cryptic.

Re: [PATCH v2 7/7] Suggest other commands instead of "git checkout"

[Junio C Hamano]

Duy Nguyen <pclouds@gmail.com> writes:

> I see my deliberate attempt to provoke has failed :D Giving your view
> of the new commands as "training wheels", I take it we still should
> make them visible as much as possible, but we just not try to hide
> "git checkout" as much (e.g. we mention both new and old commands,
> instead of just mentioning the new one, when suggesting something)?

Yes, I do support the overall idea of learning two (or possibly
three) separate commands would help new users to form the right
mental model much sooner than learning one that can be used in
multiple ways.  Another possible approach could be to split the use
of "reset" that does not move "HEAD" into the same half of
"checkout" that does not move "HEAD", i.e. checkout-files.

Re: [PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Junio C Hamano]

Stefan Xenos <sxenos@google.com> writes:

> So - IMO - detaching should always be an explicit action. Some options
> that occur to me:
>
> git switch-branch --detach

That is the most obvious way to spell it, and it is why we have "git
checkout --detach".  If we were to split one half of "checkout" into
"switch-branch", I would support the idea to make switch-branch only
take a branch name and nothing else, allow it to take any commit-ish
to detach the HEAD at that commit in the history graph with the
--detach option".  I also do not think anybody minds explaining the
resulting state to be "on an unnamed branch" (or is it "the" unnamed
branch?  Given that HEAD@{} reflog is a singleton, perhaps the right
mental model is that there is only one unnamed branch per worktree).

> git detach

The detached HEAD state is not all that special to deserve a
separate command.  After all, all history growing commands like
"commit", "cherry-pick", "rebase", "merge", etc. work the same way
on the unnamed branch.

> git switch-branch HEAD

I do not think this one is acceptable.  "git checkout HEAD" does not
detach but works as if you said "git checkout master" (when on the
'master' branch).  And you do not want "git switch-branch HEAD^0" to
be that explicit way to tell Git to detach the HEAD as that will
take us back to square one ("git checkout HEAD^0" is the more
concise and time-honored way to say "git checkout --detach HEAD").

Re: [PATCH v2 5/7] checkout: split options[] array in three pieces

[Junio C Hamano]

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> +static struct option *add_switch_branch_options(struct checkout_opts *opts,
> +						struct option *prevopts)
> +{
> +	struct option options[] = {
>  		OPT_STRING('b', NULL, &opts->new_branch, N_("branch"),
>  			   N_("create and checkout a new branch")),

I think there should be another step to rename the options to more
sensible ones for the context.  In the context of overall "checkout"
command, the 'b' option

	git checkout -b <new-name> <commit-ish>"

that signals that its parameter has something to do with a 'branch'
makes perfect sense.  But when the user uses the new command

	git switch-branch -b <new-name> <commit-ish>

does not convey the more important fact in the context.  In the
orignal context, "this is about a branch" and "we are not checking
out an existing one, but are creating" are both worthwhile thing to
express, but because a single letter option cannot stand for both,
"-b" is the most reasonable choice (compared to calling it "-c" that
stands for "create" that invites "what exactly are you creating?").

In the new context of "switch-branch", it no longer has to be said
that the optional feature is about "branch".  So I would imagine
that users naturally expect this option to be called

	git switch-branch --create <new-name> <commit-ish>

(or -c for short).

I'll just stop with this single example, but I think we should make
sure all the options make sense in the context of new command.

Of course, that means it will NOT be sufficient to just split the
option table into two tables and stitch them together for the single
command.  This option must stay to be "-b" (giving it a synonym
"--create-branch" is OK) in the context of "checkout".

Re: [PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Duy Nguyen]

On Wed, Nov 28, 2018 at 9:30 PM Stefan Beller <sbeller@google.com> wrote:
>
> On Wed, Nov 28, 2018 at 12:09 PM Duy Nguyen <pclouds@gmail.com> wrote:
> >
> > On Wed, Nov 28, 2018 at 9:01 PM Duy Nguyen <pclouds@gmail.com> wrote:
> > > should we do
> > > something about detached HEAD in this switch-branch command (or
> > > whatever its name will be)?
> > >
> > > This is usually a confusing concept to new users
> >
> > And it just occurred to me that perhaps we should call this "unnamed
> > branch" (at least at high UI level) instead of detached HEAD. It is
> > technically not as accurate, but much better to understand.
>
> or 'direct' branch?

makes me think, what is an indirect branch?

> I mean 'detached HEAD' itself is also not correct
> as the HEAD points to a valid commit/tag usually, so it is attached to
> that content. The detachment comes from the implicit "from a branch".

Yeah I guess it's short for "HEAD that is detached from a branch"
-- 
Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Duy Nguyen]

On Thu, Nov 29, 2018 at 6:59 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> Stefan Xenos <sxenos@google.com> writes:
>
> > Although I have no problem with "switch-branch" as a command name,
> > some alternative names we might consider for switch-branch might be:
> >
> > chbranch
> > swbranch
>
> Please never go in that direction.  So far, we made a conscious
> effort to keep the names of most frequently used subcommand to
> proper words that can be understood by coders (IOW, I expect they
> know what 'grep' is, even though that may not be a 'proper word').
>
> > switch
> > branch change (as a subcommand for the "branch" command)
>
> It is more like moving to the branch to work on.  I think 'switch'
> is something SVN veterans may find familiar.  Both are much better
> than ch/swbranch that are overly cryptic.

OK I'll go with switch-branch and restore-files in the next round. And
perhaps consider just 'switch' and 'restore' later.
-- 
Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Duy Nguyen]

On Thu, Nov 29, 2018 at 12:22 AM Stefan Xenos <sxenos@google.com> wrote:
> Some behaviors I'd expect to see from these commands (I haven't yet
> checked to see if you've already done this):
>
> git checkout-files <tree-ish>
> should reset all the files in the repository regardless of the current
> directory - it should produce the same effect as "git reset --hard
> <tree-ish> && git reset HEAD@{1}". It should also delete
> locally-created files that aren't present in <tree-ish>, such that the
> final working tree is exactly identical to what was committed in that
> tree-ish.
>
> git checkout-files foo -- myfile.txt
> should delete myfile.txt if it is present locally but not present in foo.
>
> git checkout-files foo -- .
> should recursively checkout all files in the current folder and all
> subfolders, and delete any locally-created files if they're not
> present in foo.

I think all these are the same as the non-overlay mode Thomas
mentioned [1]. Once he implements that in git-checkout, we can make it
default in checkout-files.

Two things though. I plan to get rid of "--" in checkout-files. The
main use case (I think) is reset from index, so you can just write

 git checkout-files somefiles

and if you want to get it from the tree(-ish) "foo", you do

 git checkout-files --from=foo somefiles

This form is easier to read (and even guess before you read man pages)
and leaves no room for ambiguation.

The second thing is, I plan to forbid "git checkout-files" without
arguments. If you want to reset the entire worktree, do

 git checkout-files :/

or just current dir

 git checkout-files .

Which brings us back to your "git checkout-files <tree-ish>" use case
above. It should be treat the same way in my opinion, so we either do

 git checkout-files --from=tree-ish :/

or

 git checkout-files --from=tree-ish .

But "git checkout-files --from=tree-ish" alone is rejected.


[1] https://public-inbox.org/git/xmqqwoowo1fk.fsf@gitster-ct.c.googlers.com/T/#mdb076d178ccf0ae3dba0fd63143f99278047da93

> Suggestion:
> If git checkout-files overwrites or deletes any locally-modified files
> from the workspace or index, those files could be auto-stashed. That
> would make it easy to restore them in the event of a mistyped command.
> Auto-stashing could be suppressed with a command-line argument (with
> alternate behaviors being fail-if-modified or always-overwrite).

Stashing I think is not the right tool for this. When you stash, you
plan to retrieve it back later but here you should rarely ever need to
unstash until the accident. For recovery from accidents like this, I
have another thing in queue to achieve the same (I'm calling it
"backup log" now). So we will have the same functionality, just with
different means.

> Idea:
> If git checkout-files modifies the submodules file, it could also
> auto-update the submodules. (For example, with something like "git
> submodule update --init --recursive --progress").

This one is tricky because we should deal with submodule autoupdate
consistently across all porcelain commands (or at least common ones),
not just checkout-files. I'd prefer to delay this until later. Once we
figure out what to do with other commands, then we can still change
defaults for checkout-files.
-- 
Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Stefan Beller]

> > Idea:
> > If git checkout-files modifies the submodules file, it could also
> > auto-update the submodules. (For example, with something like "git
> > submodule update --init --recursive --progress").
>
> This one is tricky because we should deal with submodule autoupdate
> consistently across all porcelain commands (or at least common ones),
> not just checkout-files. I'd prefer to delay this until later. Once we
> figure out what to do with other commands, then we can still change
> defaults for checkout-files.

checkout/reset are respecting the submodule.recurse setting for this
already, and as your patches only change the UX frontend

    git -c submodule.recurse checkout-files <pathsspec>

would also touch submodules. Given that deep down in
the submodules it's all about files again, we could think
checkout-files is still a good name.

I think Stefan X. is asking for making submodule.recurse
to default to true, which is indeed unrelated to this.

Stefan

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Duy Nguyen]

On Thu, Nov 29, 2018 at 7:14 PM Stefan Beller <sbeller@google.com> wrote:
>
> > > Idea:
> > > If git checkout-files modifies the submodules file, it could also
> > > auto-update the submodules. (For example, with something like "git
> > > submodule update --init --recursive --progress").
> >
> > This one is tricky because we should deal with submodule autoupdate
> > consistently across all porcelain commands (or at least common ones),
> > not just checkout-files. I'd prefer to delay this until later. Once we
> > figure out what to do with other commands, then we can still change
> > defaults for checkout-files.
>
> checkout/reset are respecting the submodule.recurse setting for this
> already, and as your patches only change the UX frontend
>
>     git -c submodule.recurse checkout-files <pathsspec>
>
> would also touch submodules. Given that deep down in
> the submodules it's all about files again, we could think
> checkout-files is still a good name.
>
> I think Stefan X. is asking for making submodule.recurse
> to default to true, which is indeed unrelated to this.

Yes and I'm concerned that checkout-files now recurses into submodules
this by default but grep for example does not. That just adds more
confusion.
-- 
Duy

Re: [PATCH v2 6/7] checkout: split into switch-branch and checkout-files

[Stefan Xenos]

>
> Which brings us back to your "git checkout-files <tree-ish>" use case
> above. It should be treat the same way in my opinion, so we either do
>
>  git checkout-files --from=tree-ish :/
>
> or
>
>  git checkout-files --from=tree-ish .
>
> But "git checkout-files --from=tree-ish" alone is rejected.

Agreed. Those arguments are better. The gist of my comment was the
treatment of newly created local files rather than the form of the
arguments, but it sounds like you've got that under control, too.

> > Suggestion:
> > If git checkout-files overwrites or deletes any locally-modified files
> > from the workspace or index, those files could be auto-stashed. That
> > would make it easy to restore them in the event of a mistyped command.
> > Auto-stashing could be suppressed with a command-line argument (with
> > alternate behaviors being fail-if-modified or always-overwrite).
>
> Stashing I think is not the right tool for this. When you stash, you
> plan to retrieve it back later but here you should rarely ever need to
> unstash until the accident. For recovery from accidents like this, I
> have another thing in queue to achieve the same (I'm calling it
> "backup log" now). So we will have the same functionality, just with
> different means.

Yes, this makes sense too. You wouldn't want to pollute the stash list
with autogenerated things the user probably doesn't want.

> This one is tricky because we should deal with submodule autoupdate
> consistently across all porcelain commands (or at least common ones),
> not just checkout-files.

This is also a good point. I'd like it if submodules just behaved like
a single giant repository for most commands, but you're right that
this is something that should be done intentionally for all the
commands at one rather than just for a single command.

I also like your new names "switch-branch" and "restore-files".

[PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Nguyễn Thái Ngọc Duy]

v3 sees switch-branch go back to switch-branch (in v2 it was
checkout-branch). checkout-files is also renamed restore-files (v1 was
restore-paths). Hopefully we won't see another rename.

I'll try to summarize the differences between the new commands and
'git checkout' down here, but you're welcome to just head to 07/14 and
read the new man pages.

'git switch-branch'

- does not "do nothing", you have to either switch branch, create a
  new branch, or detach. "git switch-branch" with no arguments is
  rejected.

- implicit detaching is rejected. If you need to detach, you need to
  give --detach. Or stick to 'git checkout'.

- -b/-B is renamed to -c/-C with long option names

- of course does not accept pathspec

'git restore-files'

- takes a ref from --from argument, not as a free ref. As a result,
  '--' is no longer needed. All non-option arguments are pathspec

- pathspec is mandatory, you can't do "git restore-files" without any
  pathspec.

- I just remember -p which is allowed to take no pathspec :( I'll fix
  it later.

- Two more fancy features (the "git checkout --index" being the
  default mode and the backup log for accidental overwrites) are of
  course still missing. But they are coming.

I did not go replace "detached HEAD" with "unnamed branch" (or "no
branch") everywhere because I think a unique term is still good to
refer to this concept. Or maybe "no branch" is good enough. I dunno.

Nguyễn Thái Ngọc Duy (14):
  git-checkout.txt: fix one syntax line
  git-checkout.txt: split detached head section out
  checkout: factor out some code in parse_branchname_arg()
  checkout: make "opts" in cmd_checkout() a pointer
  checkout: move 'confict_style' and 'dwim_..' to checkout_opts
  checkout: split options[] array in three pieces
  checkout: split into switch-branch and restore-files
  switch-branch: better names for -b and -B
  switch-branch: stop accepting pathspec
  switch-branch: reject "do nothing" case
  switch-branch: only allow explicit detached HEAD
  restore-files: take tree-ish from --from option instead
  restore-files: make pathspec mandatory
  doc: promote "git switch-branch" and "git restore-files"

 .gitignore                             |   2 +
 Documentation/config/advice.txt        |  10 +-
 Documentation/config/checkout.txt      |   5 +-
 Documentation/detach-head.txt          | 132 +++++++++
 Documentation/git-branch.txt           |   8 +-
 Documentation/git-check-ref-format.txt |   2 +-
 Documentation/git-checkout.txt         | 140 +--------
 Documentation/git-format-patch.txt     |   2 +-
 Documentation/git-merge-base.txt       |   2 +-
 Documentation/git-rebase.txt           |   2 +-
 Documentation/git-remote.txt           |   2 +-
 Documentation/git-rerere.txt           |  10 +-
 Documentation/git-reset.txt            |  18 +-
 Documentation/git-restore-files.txt    | 167 +++++++++++
 Documentation/git-revert.txt           |   2 +-
 Documentation/git-stash.txt            |   6 +-
 Documentation/git-switch-branch.txt    | 289 +++++++++++++++++++
 Documentation/gitattributes.txt        |   2 +-
 Documentation/gitcli.txt               |   4 +-
 Documentation/gitcore-tutorial.txt     |  18 +-
 Documentation/giteveryday.txt          |  24 +-
 Documentation/githooks.txt             |   5 +-
 Documentation/gittutorial-2.txt        |   2 +-
 Documentation/gittutorial.txt          |   4 +-
 Documentation/revisions.txt            |   2 +-
 Documentation/user-manual.txt          |  54 ++--
 Makefile                               |   2 +
 advice.c                               |  11 +-
 builtin.h                              |   2 +
 builtin/checkout.c                     | 380 ++++++++++++++++++-------
 command-list.txt                       |   4 +-
 git.c                                  |   2 +
 parse-options-cb.c                     |  16 ++
 parse-options.h                        |   3 +-
 sha1-name.c                            |   2 +-
 wt-status.c                            |   2 +-
 36 files changed, 1006 insertions(+), 332 deletions(-)
 create mode 100644 Documentation/detach-head.txt
 create mode 100644 Documentation/git-restore-files.txt
 create mode 100644 Documentation/git-switch-branch.txt

-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 01/14] git-checkout.txt: fix one syntax line

[Nguyễn Thái Ngọc Duy]

<branch> can be omitted in this syntax, and it's actually documented a
few paragraphs down:

  You could omit <branch>, in which case the command degenerates to
  "check out the current branch", which is a glorified no-op with
  rather expensive side-effects to show only the tracking information,
  if exists, for the current branch.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 Documentation/git-checkout.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Documentation/git-checkout.txt b/Documentation/git-checkout.txt
index 801de2f764..65bd1bc50d 100644
--- a/Documentation/git-checkout.txt
+++ b/Documentation/git-checkout.txt
@@ -23,7 +23,7 @@ or the specified tree.  If no paths are given, 'git checkout' will
 also update `HEAD` to set the specified branch as the current
 branch.
 
-'git checkout' <branch>::
+'git checkout' [<branch>]::
 	To prepare for working on <branch>, switch to it by updating
 	the index and the files in the working tree, and by pointing
 	HEAD at the branch. Local modifications to the files in the
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 02/14] git-checkout.txt: split detached head section out

[Nguyễn Thái Ngọc Duy]

This is to be reused by the coming git-switch-branch.txt man page
which also deals with detached HEAD.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 Documentation/detach-head.txt  | 132 ++++++++++++++++++++++++++++++++
 Documentation/git-checkout.txt | 133 +--------------------------------
 2 files changed, 133 insertions(+), 132 deletions(-)
 create mode 100644 Documentation/detach-head.txt

diff --git a/Documentation/detach-head.txt b/Documentation/detach-head.txt
new file mode 100644
index 0000000000..bb6f5d7843
--- /dev/null
+++ b/Documentation/detach-head.txt
@@ -0,0 +1,132 @@
+HEAD normally refers to a named branch (e.g. 'master'). Meanwhile, each
+branch refers to a specific commit. Let's look at a repo with three
+commits, one of them tagged, and with branch 'master' checked out:
+
+------------
+           HEAD (refers to branch 'master')
+            |
+            v
+a---b---c  branch 'master' (refers to commit 'c')
+    ^
+    |
+  tag 'v2.0' (refers to commit 'b')
+------------
+
+When a commit is created in this state, the branch is updated to refer to
+the new commit. Specifically, 'git commit' creates a new commit 'd', whose
+parent is commit 'c', and then updates branch 'master' to refer to new
+commit 'd'. HEAD still refers to branch 'master' and so indirectly now refers
+to commit 'd':
+
+------------
+$ edit; git add; git commit
+
+               HEAD (refers to branch 'master')
+                |
+                v
+a---b---c---d  branch 'master' (refers to commit 'd')
+    ^
+    |
+  tag 'v2.0' (refers to commit 'b')
+------------
+
+It is sometimes useful to be able to checkout a commit that is not at
+the tip of any named branch, or even to create a new commit that is not
+referenced by a named branch. Let's look at what happens when we
+checkout commit 'b' (here we show two ways this may be done):
+
+------------
+$ git checkout v2.0  # or
+$ git checkout master^^
+
+   HEAD (refers to commit 'b')
+    |
+    v
+a---b---c---d  branch 'master' (refers to commit 'd')
+    ^
+    |
+  tag 'v2.0' (refers to commit 'b')
+------------
+
+Notice that regardless of which checkout command we use, HEAD now refers
+directly to commit 'b'. This is known as being in detached HEAD state.
+It means simply that HEAD refers to a specific commit, as opposed to
+referring to a named branch. Let's see what happens when we create a commit:
+
+------------
+$ edit; git add; git commit
+
+     HEAD (refers to commit 'e')
+      |
+      v
+      e
+     /
+a---b---c---d  branch 'master' (refers to commit 'd')
+    ^
+    |
+  tag 'v2.0' (refers to commit 'b')
+------------
+
+There is now a new commit 'e', but it is referenced only by HEAD. We can
+of course add yet another commit in this state:
+
+------------
+$ edit; git add; git commit
+
+         HEAD (refers to commit 'f')
+          |
+          v
+      e---f
+     /
+a---b---c---d  branch 'master' (refers to commit 'd')
+    ^
+    |
+  tag 'v2.0' (refers to commit 'b')
+------------
+
+In fact, we can perform all the normal Git operations. But, let's look
+at what happens when we then checkout master:
+
+------------
+$ git checkout master
+
+               HEAD (refers to branch 'master')
+      e---f     |
+     /          v
+a---b---c---d  branch 'master' (refers to commit 'd')
+    ^
+    |
+  tag 'v2.0' (refers to commit 'b')
+------------
+
+It is important to realize that at this point nothing refers to commit
+'f'. Eventually commit 'f' (and by extension commit 'e') will be deleted
+by the routine Git garbage collection process, unless we create a reference
+before that happens. If we have not yet moved away from commit 'f',
+any of these will create a reference to it:
+
+------------
+$ git checkout -b foo   <1>
+$ git branch foo        <2>
+$ git tag foo           <3>
+------------
+
+<1> creates a new branch 'foo', which refers to commit 'f', and then
+updates HEAD to refer to branch 'foo'. In other words, we'll no longer
+be in detached HEAD state after this command.
+
+<2> similarly creates a new branch 'foo', which refers to commit 'f',
+but leaves HEAD detached.
+
+<3> creates a new tag 'foo', which refers to commit 'f',
+leaving HEAD detached.
+
+If we have moved away from commit 'f', then we must first recover its object
+name (typically by using git reflog), and then we can create a reference to
+it. For example, to see the last two commits to which HEAD referred, we
+can use either of these commands:
+
+------------
+$ git reflog -2 HEAD # or
+$ git log -g -2 HEAD
+------------
diff --git a/Documentation/git-checkout.txt b/Documentation/git-checkout.txt
index 65bd1bc50d..25887a6087 100644
--- a/Documentation/git-checkout.txt
+++ b/Documentation/git-checkout.txt
@@ -306,138 +306,7 @@ leave out at most one of `A` and `B`, in which case it defaults to `HEAD`.
 
 DETACHED HEAD
 -------------
-HEAD normally refers to a named branch (e.g. 'master'). Meanwhile, each
-branch refers to a specific commit. Let's look at a repo with three
-commits, one of them tagged, and with branch 'master' checked out:
-
-------------
-           HEAD (refers to branch 'master')
-            |
-            v
-a---b---c  branch 'master' (refers to commit 'c')
-    ^
-    |
-  tag 'v2.0' (refers to commit 'b')
-------------
-
-When a commit is created in this state, the branch is updated to refer to
-the new commit. Specifically, 'git commit' creates a new commit 'd', whose
-parent is commit 'c', and then updates branch 'master' to refer to new
-commit 'd'. HEAD still refers to branch 'master' and so indirectly now refers
-to commit 'd':
-
-------------
-$ edit; git add; git commit
-
-               HEAD (refers to branch 'master')
-                |
-                v
-a---b---c---d  branch 'master' (refers to commit 'd')
-    ^
-    |
-  tag 'v2.0' (refers to commit 'b')
-------------
-
-It is sometimes useful to be able to checkout a commit that is not at
-the tip of any named branch, or even to create a new commit that is not
-referenced by a named branch. Let's look at what happens when we
-checkout commit 'b' (here we show two ways this may be done):
-
-------------
-$ git checkout v2.0  # or
-$ git checkout master^^
-
-   HEAD (refers to commit 'b')
-    |
-    v
-a---b---c---d  branch 'master' (refers to commit 'd')
-    ^
-    |
-  tag 'v2.0' (refers to commit 'b')
-------------
-
-Notice that regardless of which checkout command we use, HEAD now refers
-directly to commit 'b'. This is known as being in detached HEAD state.
-It means simply that HEAD refers to a specific commit, as opposed to
-referring to a named branch. Let's see what happens when we create a commit:
-
-------------
-$ edit; git add; git commit
-
-     HEAD (refers to commit 'e')
-      |
-      v
-      e
-     /
-a---b---c---d  branch 'master' (refers to commit 'd')
-    ^
-    |
-  tag 'v2.0' (refers to commit 'b')
-------------
-
-There is now a new commit 'e', but it is referenced only by HEAD. We can
-of course add yet another commit in this state:
-
-------------
-$ edit; git add; git commit
-
-	 HEAD (refers to commit 'f')
-	  |
-	  v
-      e---f
-     /
-a---b---c---d  branch 'master' (refers to commit 'd')
-    ^
-    |
-  tag 'v2.0' (refers to commit 'b')
-------------
-
-In fact, we can perform all the normal Git operations. But, let's look
-at what happens when we then checkout master:
-
-------------
-$ git checkout master
-
-               HEAD (refers to branch 'master')
-      e---f     |
-     /          v
-a---b---c---d  branch 'master' (refers to commit 'd')
-    ^
-    |
-  tag 'v2.0' (refers to commit 'b')
-------------
-
-It is important to realize that at this point nothing refers to commit
-'f'. Eventually commit 'f' (and by extension commit 'e') will be deleted
-by the routine Git garbage collection process, unless we create a reference
-before that happens. If we have not yet moved away from commit 'f',
-any of these will create a reference to it:
-
-------------
-$ git checkout -b foo   <1>
-$ git branch foo        <2>
-$ git tag foo           <3>
-------------
-
-<1> creates a new branch 'foo', which refers to commit 'f', and then
-updates HEAD to refer to branch 'foo'. In other words, we'll no longer
-be in detached HEAD state after this command.
-
-<2> similarly creates a new branch 'foo', which refers to commit 'f',
-but leaves HEAD detached.
-
-<3> creates a new tag 'foo', which refers to commit 'f',
-leaving HEAD detached.
-
-If we have moved away from commit 'f', then we must first recover its object
-name (typically by using git reflog), and then we can create a reference to
-it. For example, to see the last two commits to which HEAD referred, we
-can use either of these commands:
-
-------------
-$ git reflog -2 HEAD # or
-$ git log -g -2 HEAD
-------------
+include::detach-head.txt[]
 
 ARGUMENT DISAMBIGUATION
 -----------------------
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 03/14] checkout: factor out some code in parse_branchname_arg()

[Nguyễn Thái Ngọc Duy]

This is in preparation for the new command restore-files, which also
needs to parse opts->source_tree but does not need all the
disambiguation logic.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 51 ++++++++++++++++++++++++++++------------------
 1 file changed, 31 insertions(+), 20 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index acdafc6e4c..1887c996c6 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -990,6 +990,34 @@ static int git_checkout_config(const char *var, const char *value, void *cb)
 	return git_xmerge_config(var, value, NULL);
 }
 
+static void setup_new_branch_info_and_source_tree(
+	struct branch_info *new_branch_info,
+	struct checkout_opts *opts,
+	struct object_id *rev,
+	const char *arg)
+{
+	struct tree **source_tree = &opts->source_tree;
+	struct object_id branch_rev;
+
+	new_branch_info->name = arg;
+	setup_branch_path(new_branch_info);
+
+	if (!check_refname_format(new_branch_info->path, 0) &&
+	    !read_ref(new_branch_info->path, &branch_rev))
+		oidcpy(rev, &branch_rev);
+	else
+		new_branch_info->path = NULL; /* not an existing branch */
+
+	new_branch_info->commit = lookup_commit_reference_gently(the_repository, rev, 1);
+	if (!new_branch_info->commit) {
+		/* not a commit */
+		*source_tree = parse_tree_indirect(rev);
+	} else {
+		parse_commit_or_die(new_branch_info->commit);
+		*source_tree = get_commit_tree(new_branch_info->commit);
+	}
+}
+
 static int parse_branchname_arg(int argc, const char **argv,
 				int dwim_new_local_branch_ok,
 				struct branch_info *new_branch_info,
@@ -997,10 +1025,8 @@ static int parse_branchname_arg(int argc, const char **argv,
 				struct object_id *rev,
 				int *dwim_remotes_matched)
 {
-	struct tree **source_tree = &opts->source_tree;
 	const char **new_branch = &opts->new_branch;
 	int argcount = 0;
-	struct object_id branch_rev;
 	const char *arg;
 	int dash_dash_pos;
 	int has_dash_dash = 0;
@@ -1114,26 +1140,11 @@ static int parse_branchname_arg(int argc, const char **argv,
 	argv++;
 	argc--;
 
-	new_branch_info->name = arg;
-	setup_branch_path(new_branch_info);
-
-	if (!check_refname_format(new_branch_info->path, 0) &&
-	    !read_ref(new_branch_info->path, &branch_rev))
-		oidcpy(rev, &branch_rev);
-	else
-		new_branch_info->path = NULL; /* not an existing branch */
+	setup_new_branch_info_and_source_tree(new_branch_info, opts, rev, arg);
 
-	new_branch_info->commit = lookup_commit_reference_gently(the_repository, rev, 1);
-	if (!new_branch_info->commit) {
-		/* not a commit */
-		*source_tree = parse_tree_indirect(rev);
-	} else {
-		parse_commit_or_die(new_branch_info->commit);
-		*source_tree = get_commit_tree(new_branch_info->commit);
-	}
-
-	if (!*source_tree)                   /* case (1): want a tree */
+	if (!opts->source_tree)                   /* case (1): want a tree */
 		die(_("reference is not a tree: %s"), arg);
+
 	if (!has_dash_dash) {	/* case (3).(d) -> (1) */
 		/*
 		 * Do not complain the most common case
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 04/14] checkout: make "opts" in cmd_checkout() a pointer

[Nguyễn Thái Ngọc Duy]

"opts" will soon be moved out of cmd_checkout(). To keep changes in
that patch smaller, convert "opts" to a pointer and keep the real
thing behind "real_opts".

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 109 +++++++++++++++++++++++----------------------
 1 file changed, 55 insertions(+), 54 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 1887c996c6..1b19328d0a 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -1236,76 +1236,77 @@ static int checkout_branch(struct checkout_opts *opts,
 
 int cmd_checkout(int argc, const char **argv, const char *prefix)
 {
-	struct checkout_opts opts;
+	struct checkout_opts real_opts;
+	struct checkout_opts *opts = &real_opts;
 	struct branch_info new_branch_info;
 	char *conflict_style = NULL;
 	int dwim_new_local_branch = 1;
 	int dwim_remotes_matched = 0;
 	struct option options[] = {
-		OPT__QUIET(&opts.quiet, N_("suppress progress reporting")),
-		OPT_STRING('b', NULL, &opts.new_branch, N_("branch"),
+		OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
+		OPT_STRING('b', NULL, &opts->new_branch, N_("branch"),
 			   N_("create and checkout a new branch")),
-		OPT_STRING('B', NULL, &opts.new_branch_force, N_("branch"),
+		OPT_STRING('B', NULL, &opts->new_branch_force, N_("branch"),
 			   N_("create/reset and checkout a branch")),
-		OPT_BOOL('l', NULL, &opts.new_branch_log, N_("create reflog for new branch")),
-		OPT_BOOL(0, "detach", &opts.force_detach, N_("detach HEAD at named commit")),
-		OPT_SET_INT('t', "track",  &opts.track, N_("set upstream info for new branch"),
+		OPT_BOOL('l', NULL, &opts->new_branch_log, N_("create reflog for new branch")),
+		OPT_BOOL(0, "detach", &opts->force_detach, N_("detach HEAD at named commit")),
+		OPT_SET_INT('t', "track",  &opts->track, N_("set upstream info for new branch"),
 			BRANCH_TRACK_EXPLICIT),
-		OPT_STRING(0, "orphan", &opts.new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
-		OPT_SET_INT_F('2', "ours", &opts.writeout_stage,
+		OPT_STRING(0, "orphan", &opts->new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
+		OPT_SET_INT_F('2', "ours", &opts->writeout_stage,
 			      N_("checkout our version for unmerged files"),
 			      2, PARSE_OPT_NONEG),
-		OPT_SET_INT_F('3', "theirs", &opts.writeout_stage,
+		OPT_SET_INT_F('3', "theirs", &opts->writeout_stage,
 			      N_("checkout their version for unmerged files"),
 			      3, PARSE_OPT_NONEG),
-		OPT__FORCE(&opts.force, N_("force checkout (throw away local modifications)"),
+		OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
 			   PARSE_OPT_NOCOMPLETE),
-		OPT_BOOL('m', "merge", &opts.merge, N_("perform a 3-way merge with the new branch")),
-		OPT_BOOL_F(0, "overwrite-ignore", &opts.overwrite_ignore,
+		OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
+		OPT_BOOL_F(0, "overwrite-ignore", &opts->overwrite_ignore,
 			   N_("update ignored files (default)"),
 			   PARSE_OPT_NOCOMPLETE),
 		OPT_STRING(0, "conflict", &conflict_style, N_("style"),
 			   N_("conflict style (merge or diff3)")),
-		OPT_BOOL('p', "patch", &opts.patch_mode, N_("select hunks interactively")),
-		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts.ignore_skipworktree,
+		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
+		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
 			 N_("do not limit pathspecs to sparse entries only")),
 		OPT_HIDDEN_BOOL(0, "guess", &dwim_new_local_branch,
 				N_("second guess 'git checkout <no-such-branch>'")),
-		OPT_BOOL(0, "ignore-other-worktrees", &opts.ignore_other_worktrees,
+		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
 			 N_("do not check if another worktree is holding the given ref")),
 		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
 			    "checkout", "control recursive updating of submodules",
 			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
-		OPT_BOOL(0, "progress", &opts.show_progress, N_("force progress reporting")),
+		OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
 		OPT_END(),
 	};
 
-	memset(&opts, 0, sizeof(opts));
+	memset(opts, 0, sizeof(*opts));
 	memset(&new_branch_info, 0, sizeof(new_branch_info));
-	opts.overwrite_ignore = 1;
-	opts.prefix = prefix;
-	opts.show_progress = -1;
+	opts->overwrite_ignore = 1;
+	opts->prefix = prefix;
+	opts->show_progress = -1;
 
-	git_config(git_checkout_config, &opts);
+	git_config(git_checkout_config, opts);
 
-	opts.track = BRANCH_TRACK_UNSPECIFIED;
+	opts->track = BRANCH_TRACK_UNSPECIFIED;
 
 	argc = parse_options(argc, argv, prefix, options, checkout_usage,
 			     PARSE_OPT_KEEP_DASHDASH);
 
-	if (opts.show_progress < 0) {
-		if (opts.quiet)
-			opts.show_progress = 0;
+	if (opts->show_progress < 0) {
+		if (opts->quiet)
+			opts->show_progress = 0;
 		else
-			opts.show_progress = isatty(2);
+			opts->show_progress = isatty(2);
 	}
 
 	if (conflict_style) {
-		opts.merge = 1; /* implied */
+		opts->merge = 1; /* implied */
 		git_xmerge_config("merge.conflictstyle", conflict_style, NULL);
 	}
 
-	if ((!!opts.new_branch + !!opts.new_branch_force + !!opts.new_orphan_branch) > 1)
+	if ((!!opts->new_branch + !!opts->new_branch_force + !!opts->new_orphan_branch) > 1)
 		die(_("-b, -B and --orphan are mutually exclusive"));
 
 	/*
@@ -1313,14 +1314,14 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	 * and new_branch_force and new_orphan_branch will tell us which one of
 	 * -b/-B/--orphan is being used.
 	 */
-	if (opts.new_branch_force)
-		opts.new_branch = opts.new_branch_force;
+	if (opts->new_branch_force)
+		opts->new_branch = opts->new_branch_force;
 
-	if (opts.new_orphan_branch)
-		opts.new_branch = opts.new_orphan_branch;
+	if (opts->new_orphan_branch)
+		opts->new_branch = opts->new_orphan_branch;
 
 	/* --track without -b/-B/--orphan should DWIM */
-	if (opts.track != BRANCH_TRACK_UNSPECIFIED && !opts.new_branch) {
+	if (opts->track != BRANCH_TRACK_UNSPECIFIED && !opts->new_branch) {
 		const char *argv0 = argv[0];
 		if (!argc || !strcmp(argv0, "--"))
 			die(_("--track needs a branch name"));
@@ -1329,7 +1330,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		argv0 = strchr(argv0, '/');
 		if (!argv0 || !argv0[1])
 			die(_("missing branch name; try -b"));
-		opts.new_branch = argv0 + 1;
+		opts->new_branch = argv0 + 1;
 	}
 
 	/*
@@ -1348,56 +1349,56 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	if (argc) {
 		struct object_id rev;
 		int dwim_ok =
-			!opts.patch_mode &&
+			!opts->patch_mode &&
 			dwim_new_local_branch &&
-			opts.track == BRANCH_TRACK_UNSPECIFIED &&
-			!opts.new_branch;
+			opts->track == BRANCH_TRACK_UNSPECIFIED &&
+			!opts->new_branch;
 		int n = parse_branchname_arg(argc, argv, dwim_ok,
-					     &new_branch_info, &opts, &rev,
+					     &new_branch_info, opts, &rev,
 					     &dwim_remotes_matched);
 		argv += n;
 		argc -= n;
 	}
 
 	if (argc) {
-		parse_pathspec(&opts.pathspec, 0,
-			       opts.patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
+		parse_pathspec(&opts->pathspec, 0,
+			       opts->patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
 			       prefix, argv);
 
-		if (!opts.pathspec.nr)
+		if (!opts->pathspec.nr)
 			die(_("invalid path specification"));
 
 		/*
 		 * Try to give more helpful suggestion.
 		 * new_branch && argc > 1 will be caught later.
 		 */
-		if (opts.new_branch && argc == 1)
+		if (opts->new_branch && argc == 1)
 			die(_("'%s' is not a commit and a branch '%s' cannot be created from it"),
-				argv[0], opts.new_branch);
+				argv[0], opts->new_branch);
 
-		if (opts.force_detach)
+		if (opts->force_detach)
 			die(_("git checkout: --detach does not take a path argument '%s'"),
 			    argv[0]);
 
-		if (1 < !!opts.writeout_stage + !!opts.force + !!opts.merge)
+		if (1 < !!opts->writeout_stage + !!opts->force + !!opts->merge)
 			die(_("git checkout: --ours/--theirs, --force and --merge are incompatible when\n"
 			      "checking out of the index."));
 	}
 
-	if (opts.new_branch) {
+	if (opts->new_branch) {
 		struct strbuf buf = STRBUF_INIT;
 
-		if (opts.new_branch_force)
-			opts.branch_exists = validate_branchname(opts.new_branch, &buf);
+		if (opts->new_branch_force)
+			opts->branch_exists = validate_branchname(opts->new_branch, &buf);
 		else
-			opts.branch_exists =
-				validate_new_branchname(opts.new_branch, &buf, 0);
+			opts->branch_exists =
+				validate_new_branchname(opts->new_branch, &buf, 0);
 		strbuf_release(&buf);
 	}
 
 	UNLEAK(opts);
-	if (opts.patch_mode || opts.pathspec.nr) {
-		int ret = checkout_paths(&opts, new_branch_info.name);
+	if (opts->patch_mode || opts->pathspec.nr) {
+		int ret = checkout_paths(opts, new_branch_info.name);
 		if (ret && dwim_remotes_matched > 1 &&
 		    advice_checkout_ambiguous_remote_branch_name)
 			advise(_("'%s' matched more than one remote tracking branch.\n"
@@ -1416,6 +1417,6 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 			       dwim_remotes_matched);
 		return ret;
 	} else {
-		return checkout_branch(&opts, &new_branch_info);
+		return checkout_branch(opts, &new_branch_info);
 	}
 }
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 05/14] checkout: move 'confict_style' and 'dwim_..' to checkout_opts

[Nguyễn Thái Ngọc Duy]

These local variables are referenced by struct option[]. This struct
will soon be broken down, moved away and we can't rely on local
variables anymore. Move these two to struct checkout_opts in
preparation for that.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 1b19328d0a..2423fdbf94 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -44,6 +44,8 @@ struct checkout_opts {
 	int ignore_skipworktree;
 	int ignore_other_worktrees;
 	int show_progress;
+	int dwim_new_local_branch;
+
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
 	 * should be updated accordingly.
@@ -55,6 +57,7 @@ struct checkout_opts {
 	int new_branch_log;
 	enum branch_track track;
 	struct diff_options diff_options;
+	char *conflict_style;
 
 	int branch_exists;
 	const char *prefix;
@@ -1239,8 +1242,6 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	struct checkout_opts real_opts;
 	struct checkout_opts *opts = &real_opts;
 	struct branch_info new_branch_info;
-	char *conflict_style = NULL;
-	int dwim_new_local_branch = 1;
 	int dwim_remotes_matched = 0;
 	struct option options[] = {
 		OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
@@ -1265,12 +1266,12 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		OPT_BOOL_F(0, "overwrite-ignore", &opts->overwrite_ignore,
 			   N_("update ignored files (default)"),
 			   PARSE_OPT_NOCOMPLETE),
-		OPT_STRING(0, "conflict", &conflict_style, N_("style"),
+		OPT_STRING(0, "conflict", &opts->conflict_style, N_("style"),
 			   N_("conflict style (merge or diff3)")),
 		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
 		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
 			 N_("do not limit pathspecs to sparse entries only")),
-		OPT_HIDDEN_BOOL(0, "guess", &dwim_new_local_branch,
+		OPT_HIDDEN_BOOL(0, "guess", &opts->dwim_new_local_branch,
 				N_("second guess 'git checkout <no-such-branch>'")),
 		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
 			 N_("do not check if another worktree is holding the given ref")),
@@ -1286,6 +1287,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	opts->overwrite_ignore = 1;
 	opts->prefix = prefix;
 	opts->show_progress = -1;
+	opts->dwim_new_local_branch = 1;
 
 	git_config(git_checkout_config, opts);
 
@@ -1301,9 +1303,9 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 			opts->show_progress = isatty(2);
 	}
 
-	if (conflict_style) {
+	if (opts->conflict_style) {
 		opts->merge = 1; /* implied */
-		git_xmerge_config("merge.conflictstyle", conflict_style, NULL);
+		git_xmerge_config("merge.conflictstyle", opts->conflict_style, NULL);
 	}
 
 	if ((!!opts->new_branch + !!opts->new_branch_force + !!opts->new_orphan_branch) > 1)
@@ -1350,7 +1352,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		struct object_id rev;
 		int dwim_ok =
 			!opts->patch_mode &&
-			dwim_new_local_branch &&
+			opts->dwim_new_local_branch &&
 			opts->track == BRANCH_TRACK_UNSPECIFIED &&
 			!opts->new_branch;
 		int n = parse_branchname_arg(argc, argv, dwim_ok,
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 06/14] checkout: split options[] array in three pieces

[Nguyễn Thái Ngọc Duy]

This is a preparation step for introducing new commands that do parts
of what checkout does. There will be two new commands, one is about
switching branches, detaching HEAD... one about checking out
paths. These share the a subset of command line options. The rest of
command line options are separate.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 77 +++++++++++++++++++++++++++++++++-------------
 parse-options-cb.c | 16 ++++++++++
 parse-options.h    |  3 +-
 3 files changed, 73 insertions(+), 23 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 2423fdbf94..764e1a83a1 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -1237,14 +1237,31 @@ static int checkout_branch(struct checkout_opts *opts,
 	return switch_branches(opts, new_branch_info);
 }
 
-int cmd_checkout(int argc, const char **argv, const char *prefix)
+static struct option *add_common_options(struct checkout_opts *opts,
+					 struct option *prevopts)
 {
-	struct checkout_opts real_opts;
-	struct checkout_opts *opts = &real_opts;
-	struct branch_info new_branch_info;
-	int dwim_remotes_matched = 0;
 	struct option options[] = {
 		OPT__QUIET(&opts->quiet, N_("suppress progress reporting")),
+		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
+			    "checkout", "control recursive updating of submodules",
+			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
+		OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
+		OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
+			   PARSE_OPT_NOCOMPLETE),
+		OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
+		OPT_STRING(0, "conflict", &opts->conflict_style, N_("style"),
+			   N_("conflict style (merge or diff3)")),
+		OPT_END()
+	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
+
+static struct option *add_switch_branch_options(struct checkout_opts *opts,
+						struct option *prevopts)
+{
+	struct option options[] = {
 		OPT_STRING('b', NULL, &opts->new_branch, N_("branch"),
 			   N_("create and checkout a new branch")),
 		OPT_STRING('B', NULL, &opts->new_branch_force, N_("branch"),
@@ -1254,33 +1271,44 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		OPT_SET_INT('t', "track",  &opts->track, N_("set upstream info for new branch"),
 			BRANCH_TRACK_EXPLICIT),
 		OPT_STRING(0, "orphan", &opts->new_orphan_branch, N_("new-branch"), N_("new unparented branch")),
+		OPT_HIDDEN_BOOL(0, "guess", &opts->dwim_new_local_branch,
+				N_("second guess 'git checkout <no-such-branch>'")),
+		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
+			 N_("do not check if another worktree is holding the given ref")),
+		OPT_END()
+	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
+
+static struct option *add_checkout_path_options(struct checkout_opts *opts,
+						struct option *prevopts)
+{
+	struct option options[] = {
 		OPT_SET_INT_F('2', "ours", &opts->writeout_stage,
 			      N_("checkout our version for unmerged files"),
 			      2, PARSE_OPT_NONEG),
 		OPT_SET_INT_F('3', "theirs", &opts->writeout_stage,
 			      N_("checkout their version for unmerged files"),
 			      3, PARSE_OPT_NONEG),
-		OPT__FORCE(&opts->force, N_("force checkout (throw away local modifications)"),
-			   PARSE_OPT_NOCOMPLETE),
-		OPT_BOOL('m', "merge", &opts->merge, N_("perform a 3-way merge with the new branch")),
-		OPT_BOOL_F(0, "overwrite-ignore", &opts->overwrite_ignore,
-			   N_("update ignored files (default)"),
-			   PARSE_OPT_NOCOMPLETE),
-		OPT_STRING(0, "conflict", &opts->conflict_style, N_("style"),
-			   N_("conflict style (merge or diff3)")),
 		OPT_BOOL('p', "patch", &opts->patch_mode, N_("select hunks interactively")),
 		OPT_BOOL(0, "ignore-skip-worktree-bits", &opts->ignore_skipworktree,
 			 N_("do not limit pathspecs to sparse entries only")),
-		OPT_HIDDEN_BOOL(0, "guess", &opts->dwim_new_local_branch,
-				N_("second guess 'git checkout <no-such-branch>'")),
-		OPT_BOOL(0, "ignore-other-worktrees", &opts->ignore_other_worktrees,
-			 N_("do not check if another worktree is holding the given ref")),
-		{ OPTION_CALLBACK, 0, "recurse-submodules", NULL,
-			    "checkout", "control recursive updating of submodules",
-			    PARSE_OPT_OPTARG, option_parse_recurse_submodules_worktree_updater },
-		OPT_BOOL(0, "progress", &opts->show_progress, N_("force progress reporting")),
-		OPT_END(),
+		OPT_END()
 	};
+	struct option *newopts = parse_options_concat(prevopts, options);
+	free(prevopts);
+	return newopts;
+}
+
+int cmd_checkout(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts real_opts;
+	struct checkout_opts *opts = &real_opts;
+	struct branch_info new_branch_info;
+	int dwim_remotes_matched = 0;
+	struct option *options = NULL;
 
 	memset(opts, 0, sizeof(*opts));
 	memset(&new_branch_info, 0, sizeof(new_branch_info));
@@ -1293,6 +1321,11 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 
 	opts->track = BRANCH_TRACK_UNSPECIFIED;
 
+	options = parse_options_dup(options);
+	options = add_common_options(opts, options);
+	options = add_switch_branch_options(opts, options);
+	options = add_checkout_path_options(opts, options);
+
 	argc = parse_options(argc, argv, prefix, options, checkout_usage,
 			     PARSE_OPT_KEEP_DASHDASH);
 
diff --git a/parse-options-cb.c b/parse-options-cb.c
index 8c9edce52f..f46b3cb0a4 100644
--- a/parse-options-cb.c
+++ b/parse-options-cb.c
@@ -121,6 +121,22 @@ int parse_opt_tertiary(const struct option *opt, const char *arg, int unset)
 	return 0;
 }
 
+struct option *parse_options_dup(const struct option *o)
+{
+	struct option *opts;
+	int nr = 0;
+
+	while (o && o->type != OPTION_END) {
+		nr++;
+		o++;
+	}
+
+	CALLOC_ARRAY(opts, nr + 1);
+	memcpy(opts, o - nr, sizeof(*o) * nr);
+	opts[nr].type = OPTION_END;
+	return opts;
+}
+
 struct option *parse_options_concat(struct option *a, struct option *b)
 {
 	struct option *ret;
diff --git a/parse-options.h b/parse-options.h
index 6c4fe2016d..584cb521f5 100644
--- a/parse-options.h
+++ b/parse-options.h
@@ -239,7 +239,8 @@ extern int parse_options_step(struct parse_opt_ctx_t *ctx,
 
 extern int parse_options_end(struct parse_opt_ctx_t *ctx);
 
-extern struct option *parse_options_concat(struct option *a, struct option *b);
+struct option *parse_options_dup(const struct option *a);
+struct option *parse_options_concat(struct option *a, struct option *b);
 
 /*----- some often used options -----*/
 extern int parse_opt_abbrev_cb(const struct option *, const char *, int);
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Nguyễn Thái Ngọc Duy]

"git checkout" doing too many things is a source of confusion for many
users (and it even bites old timers sometimes). To rememdy that, the
command is now split in two: switch-branch and checkout-files. The
good old "git checkout" command is still here and will be until all
(or most of users) are sick of it.

See the new man pages for the final design of these commands. The
actual implementation though is still pretty much the same as "git
checkout". Following patches will adjust their behavior to match the
man pages.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 .gitignore                          |   2 +
 Documentation/git-checkout.txt      |   5 +
 Documentation/git-restore-files.txt | 167 ++++++++++++++++
 Documentation/git-switch-branch.txt | 289 ++++++++++++++++++++++++++++
 Makefile                            |   2 +
 builtin.h                           |   2 +
 builtin/checkout.c                  |  84 ++++++--
 command-list.txt                    |   2 +
 git.c                               |   2 +
 9 files changed, 543 insertions(+), 12 deletions(-)
 create mode 100644 Documentation/git-restore-files.txt
 create mode 100644 Documentation/git-switch-branch.txt

diff --git a/.gitignore b/.gitignore
index 0d77ea5894..c63dcb1427 100644
--- a/.gitignore
+++ b/.gitignore
@@ -143,6 +143,7 @@
 /git-request-pull
 /git-rerere
 /git-reset
+/git-restore-files
 /git-rev-list
 /git-rev-parse
 /git-revert
@@ -167,6 +168,7 @@
 /git-submodule
 /git-submodule--helper
 /git-svn
+/git-switch-branch
 /git-symbolic-ref
 /git-tag
 /git-unpack-file
diff --git a/Documentation/git-checkout.txt b/Documentation/git-checkout.txt
index 25887a6087..25ec7f508f 100644
--- a/Documentation/git-checkout.txt
+++ b/Documentation/git-checkout.txt
@@ -406,6 +406,11 @@ $ edit frotz
 $ git add frotz
 ------------
 
+SEE ALSO
+--------
+linkgit:git-switch-branch[1]
+linkgit:git-restore-files[1]
+
 GIT
 ---
 Part of the linkgit:git[1] suite
diff --git a/Documentation/git-restore-files.txt b/Documentation/git-restore-files.txt
new file mode 100644
index 0000000000..03c1250ad0
--- /dev/null
+++ b/Documentation/git-restore-files.txt
@@ -0,0 +1,167 @@
+git-restore-files(1)
+====================
+
+NAME
+----
+git-restore-files - Restore working tree files
+
+SYNOPSIS
+--------
+[verse]
+'git restore-files' [-f|--ours|--theirs|-m|--conflict=<style>] [--from=<tree-ish>] <pathspec>...
+'git restore-files' [--from=<tree-ish>] <pathspec>...
+'git restore-files' (-p|--patch) [--from=<tree-ish>] [<pathspec>...]
+
+DESCRIPTION
+-----------
+Updates files in the working tree to match the version in the index
+or the specified tree.
+
+'git restore-files' [--from=<tree-ish>] <pathspec>...::
+
+	Overwrite paths in the working tree by replacing with the
+	contents in the index or in the <tree-ish> (most often a
+	commit).  When a <tree-ish> is given, the paths that
+	match the <pathspec> are updated both in the index and in
+	the working tree.
++
+The index may contain unmerged entries because of a previous failed merge.
+By default, if you try to check out such an entry from the index, the
+checkout operation will fail and nothing will be checked out.
+Using `-f` will ignore these unmerged entries.  The contents from a
+specific side of the merge can be checked out of the index by
+using `--ours` or `--theirs`.  With `-m`, changes made to the working tree
+file can be discarded to re-create the original conflicted merge result.
+
+'git restore-files' (-p|--patch) [--from=<tree-ish>] [<pathspec>...]::
+	This is similar to the "check out paths to the working tree
+	from either the index or from a tree-ish" mode described
+	above, but lets you use the interactive interface to show
+	the "diff" output and choose which hunks to use in the
+	result.  See below for the description of `--patch` option.
+
+OPTIONS
+-------
+-q::
+--quiet::
+	Quiet, suppress feedback messages.
+
+--[no-]progress::
+	Progress status is reported on the standard error stream
+	by default when it is attached to a terminal, unless `--quiet`
+	is specified. This flag enables progress reporting even if not
+	attached to a terminal, regardless of `--quiet`.
+
+-f::
+--force::
+	Do not fail upon unmerged entries; instead, unmerged entries
+	are ignored.
+
+--ours::
+--theirs::
+	Check out stage #2 ('ours') or #3 ('theirs') for unmerged
+	paths.
++
+Note that during `git rebase` and `git pull --rebase`, 'ours' and
+'theirs' may appear swapped; `--ours` gives the version from the
+branch the changes are rebased onto, while `--theirs` gives the
+version from the branch that holds your work that is being rebased.
++
+This is because `rebase` is used in a workflow that treats the
+history at the remote as the shared canonical one, and treats the
+work done on the branch you are rebasing as the third-party work to
+be integrated, and you are temporarily assuming the role of the
+keeper of the canonical history during the rebase.  As the keeper of
+the canonical history, you need to view the history from the remote
+as `ours` (i.e. "our shared canonical history"), while what you did
+on your side branch as `theirs` (i.e. "one contributor's work on top
+of it").
+
+--ignore-skip-worktree-bits::
+	In sparse checkout mode, update only entries matched by
+	<paths> and sparse patterns in
+	$GIT_DIR/info/sparse-checkout. This option ignores the sparse
+	patterns and adds back any files in <paths>.
+
+-m::
+--merge::
+	When checking out paths from the index, this option lets you
+	recreate the conflicted merge in the specified paths.
+
+--conflict=<style>::
+	The same as --merge option above, but changes the way the
+	conflicting hunks are presented, overriding the
+	merge.conflictStyle configuration variable.  Possible values are
+	"merge" (default) and "diff3" (in addition to what is shown by
+	"merge" style, shows the original contents).
+
+-p::
+--patch::
+	Interactively select hunks in the difference between the
+	<tree-ish> (or the index, if unspecified) and the working
+	tree.  The chosen hunks are then applied in reverse to the
+	working tree (and if a <tree-ish> was specified, the index).
++
+This means that you can use `git restore-files -p` to selectively
+discard edits from your current working tree. See the ``Interactive
+Mode'' section of linkgit:git-add[1] to learn how to operate the
+`--patch` mode.
+
+--[no-]recurse-submodules::
+	Using --recurse-submodules will update the content of all initialized
+	submodules according to the commit recorded in the superproject. If
+	local modifications in a submodule would be overwritten the checkout
+	will fail unless `-f` is used. If nothing (or --no-recurse-submodules)
+	is used, the work trees of submodules will not be updated.
+	Just like linkgit:git-submodule[1], this will detach the
+	submodules HEAD.
+
+<tree-ish>::
+	Tree to checkout from (when paths are given). If not specified,
+	the index will be used.
+
+EXAMPLES
+--------
+
+. The following sequence checks out the `master` branch, reverts
+the `Makefile` to two revisions back, deletes hello.c by
+mistake, and gets it back from the index.
++
+------------
+$ git switch-branch master                    <1>
+$ git restore-files --from master~2 Makefile  <2>
+$ rm -f hello.c
+$ git restore-files hello.c                   <3>
+------------
++
+<1> switch branch
+<2> take a file out of another commit
+<3> restore hello.c from the index
++
+If you want to check out _all_ C source files out of the index,
+you can say
++
+------------
+$ git restore-files '*.c'
+------------
++
+Note the quotes around `*.c`.  The file `hello.c` will also be
+checked out, even though it is no longer in the working tree,
+because the file globbing is used to match entries in the index
+(not in the working tree by the shell).
++
+If you have an unfortunate branch that is named `hello.c`, this
+step would be confused as an instruction to switch to that branch.
+You should instead write:
++
+------------
+$ git restore-files hello.c
+------------
+
+SEE ALSO
+--------
+linkgit:git-checkout[1]
+
+GIT
+---
+Part of the linkgit:git[1] suite
diff --git a/Documentation/git-switch-branch.txt b/Documentation/git-switch-branch.txt
new file mode 100644
index 0000000000..d5bf5cb37d
--- /dev/null
+++ b/Documentation/git-switch-branch.txt
@@ -0,0 +1,289 @@
+git-switch-branch(1)
+====================
+
+NAME
+----
+git-switch-branch - Switch branches
+
+SYNOPSIS
+--------
+[verse]
+'git switch-branch' [-q] [-f] [-m] <branch>
+'git switch-branch' [-q] [-f] [-m] --detach [<commit>]
+'git switch-branch' [-q] [-f] [-m] [[-b|-B|--orphan] <new_branch>] [<start_point>]
+
+DESCRIPTION
+-----------
+Switch to a specified branch and update files in the working tree to
+match it.
+
+'git switch-branch' <branch>::
+	To prepare for working on <branch>, switch to it by updating
+	the index and the files in the working tree. Local
+	modifications to the files in the working tree are kept, so
+	that they can be committed to the <branch>.
++
+If <branch> is not found but there does exist a tracking branch in
+exactly one remote (call it <remote>) with a matching name, treat as
+equivalent to
++
+------------
+$ git switch-branch -b <branch> --track <remote>/<branch>
+------------
++
+If the branch exists in multiple remotes and one of them is named by
+the `checkout.defaultRemote` configuration variable, we'll use that
+one for the purposes of disambiguation, even if the `<branch>` isn't
+unique across all remotes. Set it to
+e.g. `checkout.defaultRemote=origin` to always checkout remote
+branches from there if `<branch>` is ambiguous but exists on the
+'origin' remote. See also `checkout.defaultRemote` in
+linkgit:git-config[1].
+
+'git switch-branch' -c|-C <new_branch> [<start_point>]::
+
+	Specifying `-c` causes a new branch to be created as if
+	linkgit:git-branch[1] were called and then switched to. In
+	this case you can use the `--track` or `--no-track` options,
+	which will be passed to 'git branch'.  As a convenience,
+	`--track` without `-c` implies branch creation; see the
+	description of `--track` below.
++
+If `-C` is given, <new_branch> is created if it doesn't exist;
+otherwise, it is reset. This is the transactional equivalent of
++
+------------
+$ git branch -f <branch> [<start_point>]
+$ git switch-branch <branch>
+------------
++
+that is to say, the branch is not reset/created unless "git
+switch-branch" is successful.
+
+'git switch-branch' --detach [<commit>]::
+
+	Prepare to work on a unnamed branch on top of <commit> (see
+	"DETACHED HEAD" section), and updating the index and the files
+	in the working tree.  Local modifications to the files in the
+	working tree are kept, so that the resulting working tree will
+	be the state recorded in the commit plus the local
+	modifications.
++
+When the <commit> argument is a branch name, the `--detach` option can
+be used to detach HEAD at the tip of the branch (`git switch-branch
+<branch>` would check out that branch without detaching HEAD).
++
+Omitting <commit> detaches HEAD at the tip of the current branch.
+
+OPTIONS
+-------
+-q::
+--quiet::
+	Quiet, suppress feedback messages.
+
+--[no-]progress::
+	Progress status is reported on the standard error stream
+	by default when it is attached to a terminal, unless `--quiet`
+	is specified. This flag enables progress reporting even if not
+	attached to a terminal, regardless of `--quiet`.
+
+-f::
+--force::
+	Proceed even if the index or the working tree differs from
+	HEAD.  This is used to throw away local changes.
+
+-c <new_branch>::
+--create <new_branch>::
+	Create a new branch named <new_branch> and start it at
+	<start_point>; see linkgit:git-branch[1] for details.
+
+-C <new_branch>::
+--force-create <new_branch>::
+	Creates the branch <new_branch> and start it at <start_point>;
+	if it already exists, then reset it to <start_point>. This is
+	equivalent to running "git branch" with "-f"; see
+	linkgit:git-branch[1] for details.
+
+-t::
+--track::
+	When creating a new branch, set up "upstream" configuration. See
+	"--track" in linkgit:git-branch[1] for details.
++
+If no `-c` option is given, the name of the new branch will be derived
+from the remote-tracking branch, by looking at the local part of the
+refspec configured for the corresponding remote, and then stripping
+the initial part up to the "*".
+This would tell us to use "hack" as the local branch when branching
+off of "origin/hack" (or "remotes/origin/hack", or even
+"refs/remotes/origin/hack").  If the given name has no slash, or the above
+guessing results in an empty name, the guessing is aborted.  You can
+explicitly give a name with `-c` in such a case.
+
+--no-track::
+	Do not set up "upstream" configuration, even if the
+	branch.autoSetupMerge configuration variable is true.
+
+-l::
+	Create the new branch's reflog; see linkgit:git-branch[1] for
+	details.
+
+--detach::
+	Rather than checking out a branch to work on it, check out a
+	commit for inspection and discardable experiments.
+	This is the default behavior of "git checkout <commit>" when
+	<commit> is not a branch name.  See the "DETACHED HEAD" section
+	below for details.
+
+--orphan <new_branch>::
+	Create a new 'orphan' branch, named <new_branch>, started from
+	<start_point> and switch to it.  The first commit made on this
+	new branch will have no parents and it will be the root of a new
+	history totally disconnected from all the other branches and
+	commits.
++
+The index and the working tree are adjusted as if you had previously run
+"git checkout <start_point>".  This allows you to start a new history
+that records a set of paths similar to <start_point> by easily running
+"git commit -a" to make the root commit.
++
+This can be useful when you want to publish the tree from a commit
+without exposing its full history. You might want to do this to publish
+an open source branch of a project whose current tree is "clean", but
+whose full history contains proprietary or otherwise encumbered bits of
+code.
++
+If you want to start a disconnected history that records a set of paths
+that is totally different from the one of <start_point>, then you should
+clear the index and the working tree right after creating the orphan
+branch by running "git rm -rf ." from the top level of the working tree.
+Afterwards you will be ready to prepare your new files, repopulating the
+working tree, by copying them from elsewhere, extracting a tarball, etc.
+
+-m::
+--merge::
+	If you have local modifications to one or more files that are
+	different between the current branch and the branch to which
+	you are switching, the command refuses to switch branches in
+	order to preserve your modifications in context.  However,
+	with this option, a three-way merge between the current
+	branch, your working tree contents, and the new branch is
+	done, and you will be on the new branch.
++
+When a merge conflict happens, the index entries for conflicting
+paths are left unmerged, and you need to resolve the conflicts
+and mark the resolved paths with `git add` (or `git rm` if the merge
+should result in deletion of the path).
+
+--conflict=<style>::
+	The same as --merge option above, but changes the way the
+	conflicting hunks are presented, overriding the
+	merge.conflictStyle configuration variable.  Possible values are
+	"merge" (default) and "diff3" (in addition to what is shown by
+	"merge" style, shows the original contents).
+
+--ignore-other-worktrees::
+	`git switch-branch` refuses when the wanted ref is already
+	checked out by another worktree. This option makes it check
+	the ref out anyway. In other words, the ref can be held by
+	more than one worktree.
+
+--[no-]recurse-submodules::
+	Using --recurse-submodules will update the content of all initialized
+	submodules according to the commit recorded in the superproject. If
+	local modifications in a submodule would be overwritten the checkout
+	will fail unless `-f` is used. If nothing (or --no-recurse-submodules)
+	is used, the work trees of submodules will not be updated.
+	Just like linkgit:git-submodule[1], this will detach the
+	submodules HEAD.
+
+<branch>::
+	Branch to checkout; if it refers to a branch (i.e., a name that,
+	when prepended with "refs/heads/", is a valid ref), then that
+	branch is checked out. Otherwise, if it refers to a valid
+	commit, your HEAD becomes "detached" and you are no longer on
+	any branch (see below for details).
++
+You can use the `"@{-N}"` syntax to refer to the N-th last
+branch/commit checked out using "git checkout" operation. You may
+also specify `-` which is synonymous to `"@{-1}`.
++
+As a special case, you may use `"A...B"` as a shortcut for the
+merge base of `A` and `B` if there is exactly one merge base. You can
+leave out at most one of `A` and `B`, in which case it defaults to `HEAD`.
+
+<new_branch>::
+	Name for the new branch.
+
+<start_point>::
+	The name of a commit at which to start the new branch; see
+	linkgit:git-branch[1] for details. Defaults to HEAD.
+
+DETACHED HEAD
+-------------
+include::detach-head.txt[]
+
+EXAMPLES
+--------
+
+. The following sequence checks out the `master` branch.
++
+------------
+$ git switch-branch master
+------------
++
+
+. After working in the wrong branch, switching to the correct
+branch would be done using:
++
+------------
+$ git switch-branch mytopic
+------------
++
+However, your "wrong" branch and correct "mytopic" branch may
+differ in files that you have modified locally, in which case
+the above checkout would fail like this:
++
+------------
+$ git switch-branch mytopic
+error: You have local changes to 'frotz'; not switching branches.
+------------
++
+You can give the `-m` flag to the command, which would try a
+three-way merge:
++
+------------
+$ git switch-branch -m mytopic
+Auto-merging frotz
+------------
++
+After this three-way merge, the local modifications are _not_
+registered in your index file, so `git diff` would show you what
+changes you made since the tip of the new branch.
+
+. When a merge conflict happens during switching branches with
+the `-m` option, you would see something like this:
++
+------------
+$ git switch-branch -m mytopic
+Auto-merging frotz
+ERROR: Merge conflict in frotz
+fatal: merge program failed
+------------
++
+At this point, `git diff` shows the changes cleanly merged as in
+the previous example, as well as the changes in the conflicted
+files.  Edit and resolve the conflict and mark it resolved with
+`git add` as usual:
++
+------------
+$ edit frotz
+$ git add frotz
+------------
+
+SEE ALSO
+--------
+linkgit:git-checkout[1]
+
+GIT
+---
+Part of the linkgit:git[1] suite
diff --git a/Makefile b/Makefile
index 1a44c811aa..f035dbab9e 100644
--- a/Makefile
+++ b/Makefile
@@ -777,9 +777,11 @@ BUILT_INS += git-format-patch$X
 BUILT_INS += git-fsck-objects$X
 BUILT_INS += git-init$X
 BUILT_INS += git-merge-subtree$X
+BUILT_INS += git-restore-files$X
 BUILT_INS += git-show$X
 BUILT_INS += git-stage$X
 BUILT_INS += git-status$X
+BUILT_INS += git-switch-branch$X
 BUILT_INS += git-whatchanged$X
 
 # what 'all' will build and 'install' will install in gitexecdir,
diff --git a/builtin.h b/builtin.h
index 6538932e99..01ed43ea69 100644
--- a/builtin.h
+++ b/builtin.h
@@ -214,6 +214,7 @@ extern int cmd_remote_fd(int argc, const char **argv, const char *prefix);
 extern int cmd_repack(int argc, const char **argv, const char *prefix);
 extern int cmd_rerere(int argc, const char **argv, const char *prefix);
 extern int cmd_reset(int argc, const char **argv, const char *prefix);
+extern int cmd_restore_files(int argc, const char **argv, const char *prefix);
 extern int cmd_rev_list(int argc, const char **argv, const char *prefix);
 extern int cmd_rev_parse(int argc, const char **argv, const char *prefix);
 extern int cmd_revert(int argc, const char **argv, const char *prefix);
@@ -227,6 +228,7 @@ extern int cmd_show_index(int argc, const char **argv, const char *prefix);
 extern int cmd_status(int argc, const char **argv, const char *prefix);
 extern int cmd_stripspace(int argc, const char **argv, const char *prefix);
 extern int cmd_submodule__helper(int argc, const char **argv, const char *prefix);
+extern int cmd_switch_branch(int argc, const char **argv, const char *prefix);
 extern int cmd_symbolic_ref(int argc, const char **argv, const char *prefix);
 extern int cmd_tag(int argc, const char **argv, const char *prefix);
 extern int cmd_tar_tree(int argc, const char **argv, const char *prefix);
diff --git a/builtin/checkout.c b/builtin/checkout.c
index 764e1a83a1..7dc0f4d3f3 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -33,6 +33,16 @@ static const char * const checkout_usage[] = {
 	NULL,
 };
 
+static const char * const switch_branch_usage[] = {
+	N_("git switch-branch [<options>] [<branch>]"),
+	NULL,
+};
+
+static const char * const restore_files_usage[] = {
+	N_("git restore-files [<options>] [<branch>] -- <file>..."),
+	NULL,
+};
+
 struct checkout_opts {
 	int patch_mode;
 	int quiet;
@@ -1302,31 +1312,23 @@ static struct option *add_checkout_path_options(struct checkout_opts *opts,
 	return newopts;
 }
 
-int cmd_checkout(int argc, const char **argv, const char *prefix)
+static int checkout_main(int argc, const char **argv, const char *prefix,
+			 struct checkout_opts *opts, struct option *options,
+			 const char * const usagestr[])
 {
-	struct checkout_opts real_opts;
-	struct checkout_opts *opts = &real_opts;
 	struct branch_info new_branch_info;
 	int dwim_remotes_matched = 0;
-	struct option *options = NULL;
 
-	memset(opts, 0, sizeof(*opts));
 	memset(&new_branch_info, 0, sizeof(new_branch_info));
 	opts->overwrite_ignore = 1;
 	opts->prefix = prefix;
 	opts->show_progress = -1;
-	opts->dwim_new_local_branch = 1;
 
 	git_config(git_checkout_config, opts);
 
 	opts->track = BRANCH_TRACK_UNSPECIFIED;
 
-	options = parse_options_dup(options);
-	options = add_common_options(opts, options);
-	options = add_switch_branch_options(opts, options);
-	options = add_checkout_path_options(opts, options);
-
-	argc = parse_options(argc, argv, prefix, options, checkout_usage,
+	argc = parse_options(argc, argv, prefix, options, usagestr,
 			     PARSE_OPT_KEEP_DASHDASH);
 
 	if (opts->show_progress < 0) {
@@ -1455,3 +1457,61 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 		return checkout_branch(opts, &new_branch_info);
 	}
 }
+
+int cmd_checkout(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	memset(&opts, 0, sizeof(opts));
+	opts.dwim_new_local_branch = 1;
+
+	options = parse_options_dup(options);
+	options = add_common_options(&opts, options);
+	options = add_switch_branch_options(&opts, options);
+	options = add_checkout_path_options(&opts, options);
+
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, checkout_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
+
+int cmd_switch_branch(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	memset(&opts, 0, sizeof(opts));
+	opts.dwim_new_local_branch = 1;
+
+	options = parse_options_dup(options);
+	options = add_common_options(&opts, options);
+	options = add_switch_branch_options(&opts, options);
+
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, switch_branch_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
+
+int cmd_restore_files(int argc, const char **argv, const char *prefix)
+{
+	struct checkout_opts opts;
+	struct option *options = NULL;
+	int ret;
+
+	memset(&opts, 0, sizeof(opts));
+	opts.dwim_new_local_branch = 1;
+
+	options = parse_options_dup(options);
+	options = add_common_options(&opts, options);
+	options = add_checkout_path_options(&opts, options);
+
+	ret = checkout_main(argc, argv, prefix, &opts,
+			    options, restore_files_usage);
+	FREE_AND_NULL(options);
+	return ret;
+}
diff --git a/command-list.txt b/command-list.txt
index 3a9af104b5..4638802754 100644
--- a/command-list.txt
+++ b/command-list.txt
@@ -151,6 +151,7 @@ git-replace                             ancillarymanipulators           complete
 git-request-pull                        foreignscminterface             complete
 git-rerere                              ancillaryinterrogators
 git-reset                               mainporcelain           worktree
+git-restore-files                       mainporcelain           worktree
 git-revert                              mainporcelain
 git-rev-list                            plumbinginterrogators
 git-rev-parse                           plumbinginterrogators
@@ -171,6 +172,7 @@ git-status                              mainporcelain           info
 git-stripspace                          purehelpers
 git-submodule                           mainporcelain
 git-svn                                 foreignscminterface
+git-switch-branch                       mainporcelain           history
 git-symbolic-ref                        plumbingmanipulators
 git-tag                                 mainporcelain           history
 git-unpack-file                         plumbinginterrogators
diff --git a/git.c b/git.c
index 2f604a41ea..a2be6c3eb5 100644
--- a/git.c
+++ b/git.c
@@ -542,6 +542,7 @@ static struct cmd_struct commands[] = {
 	{ "replace", cmd_replace, RUN_SETUP },
 	{ "rerere", cmd_rerere, RUN_SETUP },
 	{ "reset", cmd_reset, RUN_SETUP },
+	{ "restore-files", cmd_restore_files, RUN_SETUP | NEED_WORK_TREE },
 	{ "rev-list", cmd_rev_list, RUN_SETUP | NO_PARSEOPT },
 	{ "rev-parse", cmd_rev_parse, NO_PARSEOPT },
 	{ "revert", cmd_revert, RUN_SETUP | NEED_WORK_TREE },
@@ -557,6 +558,7 @@ static struct cmd_struct commands[] = {
 	{ "status", cmd_status, RUN_SETUP | NEED_WORK_TREE },
 	{ "stripspace", cmd_stripspace },
 	{ "submodule--helper", cmd_submodule__helper, RUN_SETUP | SUPPORT_SUPER_PREFIX | NO_PARSEOPT },
+	{ "switch-branch", cmd_switch_branch, RUN_SETUP | NEED_WORK_TREE },
 	{ "symbolic-ref", cmd_symbolic_ref, RUN_SETUP },
 	{ "tag", cmd_tag, RUN_SETUP | DELAY_PAGER_CONFIG },
 	{ "unpack-file", cmd_unpack_file, RUN_SETUP | NO_PARSEOPT },
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 08/14] switch-branch: better names for -b and -B

[Nguyễn Thái Ngọc Duy]

The shortcut of these options do not make much sense when used with
switch-branch. And their descriptions are also tied to checkout
out. Move -b/-B to cmd_checkout() and new -c/-C with the same
functionality in cmd_switch_branch()

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 30 +++++++++++++++++++-----------
 1 file changed, 19 insertions(+), 11 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 7dc0f4d3f3..ceb635de36 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -1268,14 +1268,10 @@ static struct option *add_common_options(struct checkout_opts *opts,
 	return newopts;
 }
 
-static struct option *add_switch_branch_options(struct checkout_opts *opts,
-						struct option *prevopts)
+static struct option *add_common_switch_branch_options(
+	struct checkout_opts *opts, struct option *prevopts)
 {
 	struct option options[] = {
-		OPT_STRING('b', NULL, &opts->new_branch, N_("branch"),
-			   N_("create and checkout a new branch")),
-		OPT_STRING('B', NULL, &opts->new_branch_force, N_("branch"),
-			   N_("create/reset and checkout a branch")),
 		OPT_BOOL('l', NULL, &opts->new_branch_log, N_("create reflog for new branch")),
 		OPT_BOOL(0, "detach", &opts->force_detach, N_("detach HEAD at named commit")),
 		OPT_SET_INT('t', "track",  &opts->track, N_("set upstream info for new branch"),
@@ -1461,15 +1457,21 @@ static int checkout_main(int argc, const char **argv, const char *prefix,
 int cmd_checkout(int argc, const char **argv, const char *prefix)
 {
 	struct checkout_opts opts;
-	struct option *options = NULL;
+	struct option *options;
+	struct option checkout_options[] = {
+		OPT_STRING('b', NULL, &opts.new_branch, N_("branch"),
+			   N_("create and checkout a new branch")),
+		OPT_STRING('B', NULL, &opts.new_branch_force, N_("branch"),
+			   N_("create/reset and checkout a branch")),
+	};
 	int ret;
 
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
 
-	options = parse_options_dup(options);
+	options = parse_options_dup(checkout_options);
 	options = add_common_options(&opts, options);
-	options = add_switch_branch_options(&opts, options);
+	options = add_common_switch_branch_options(&opts, options);
 	options = add_checkout_path_options(&opts, options);
 
 	ret = checkout_main(argc, argv, prefix, &opts,
@@ -1482,14 +1484,20 @@ int cmd_switch_branch(int argc, const char **argv, const char *prefix)
 {
 	struct checkout_opts opts;
 	struct option *options = NULL;
+	struct option switch_options[] = {
+		OPT_STRING('c', "create", &opts.new_branch, N_("branch"),
+			   N_("create and switch to a new branch")),
+		OPT_STRING('C', "force-create", &opts.new_branch_force, N_("branch"),
+			   N_("create/reset and switch to a new branch")),
+	};
 	int ret;
 
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
 
-	options = parse_options_dup(options);
+	options = parse_options_dup(switch_options);
 	options = add_common_options(&opts, options);
-	options = add_switch_branch_options(&opts, options);
+	options = add_common_switch_branch_options(&opts, options);
 
 	ret = checkout_main(argc, argv, prefix, &opts,
 			    options, switch_branch_usage);
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 09/14] switch-branch: stop accepting pathspec

[Nguyễn Thái Ngọc Duy]

This command is about switching branch (or creating a new one) and
should not accept pathspec. This helps simplify ambiguation
handling. The other two ("git checkout" and "git restore-files") of
course do accept pathspec as before.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index ceb635de36..880030e929 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -55,6 +55,7 @@ struct checkout_opts {
 	int ignore_other_worktrees;
 	int show_progress;
 	int dwim_new_local_branch;
+	int accept_pathspec;
 
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
@@ -1089,10 +1090,16 @@ static int parse_branchname_arg(int argc, const char **argv,
 	if (!argc)
 		return 0;
 
+	if (!opts->accept_pathspec) {
+		if (argc > 1)
+			die(_("only one reference expected"));
+		has_dash_dash = 1; /* helps disambiguate */
+	}
+
 	arg = argv[0];
 	dash_dash_pos = -1;
 	for (i = 0; i < argc; i++) {
-		if (!strcmp(argv[i], "--")) {
+		if (opts->accept_pathspec && !strcmp(argv[i], "--")) {
 			dash_dash_pos = i;
 			break;
 		}
@@ -1167,7 +1174,7 @@ static int parse_branchname_arg(int argc, const char **argv,
 		 */
 		if (argc)
 			verify_non_filename(opts->prefix, arg);
-	} else {
+	} else if (opts->accept_pathspec) {
 		argcount++;
 		argv++;
 		argc--;
@@ -1468,6 +1475,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
+	opts.accept_pathspec = 1;
 
 	options = parse_options_dup(checkout_options);
 	options = add_common_options(&opts, options);
@@ -1494,6 +1502,7 @@ int cmd_switch_branch(int argc, const char **argv, const char *prefix)
 
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
+	opts.accept_pathspec = 0;
 
 	options = parse_options_dup(switch_options);
 	options = add_common_options(&opts, options);
@@ -1513,6 +1522,7 @@ int cmd_restore_files(int argc, const char **argv, const char *prefix)
 
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
+	opts.accept_pathspec = 1;
 
 	options = parse_options_dup(options);
 	options = add_common_options(&opts, options);
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 10/14] switch-branch: reject "do nothing" case

[Nguyễn Thái Ngọc Duy]

"git checkout" can be executed without any arguments. What it does is
not exactly great: it switches from HEAD to HEAD and showing worktree
modification as a side effect.

Make switch-branch reject this case. You have to either

- really switch a branch
- (explicitly) detach from the current branch
- create a new branch

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 880030e929..c7ae068d2c 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -56,6 +56,7 @@ struct checkout_opts {
 	int show_progress;
 	int dwim_new_local_branch;
 	int accept_pathspec;
+	int switch_branch_doing_nothing_not_ok;
 
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
@@ -1233,6 +1234,13 @@ static int checkout_branch(struct checkout_opts *opts,
 		die(_("Cannot switch branch to a non-commit '%s'"),
 		    new_branch_info->name);
 
+	if (opts->switch_branch_doing_nothing_not_ok &&
+	    !new_branch_info->name &&
+	    !opts->new_branch &&
+	    !opts->new_branch_force &&
+	    !opts->force_detach)
+		die(_("nothing to do"));
+
 	if (new_branch_info->path && !opts->force_detach && !opts->new_branch &&
 	    !opts->ignore_other_worktrees) {
 		int flag;
@@ -1475,6 +1483,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
+	opts.switch_branch_doing_nothing_not_ok = 0;
 	opts.accept_pathspec = 1;
 
 	options = parse_options_dup(checkout_options);
@@ -1503,6 +1512,7 @@ int cmd_switch_branch(int argc, const char **argv, const char *prefix)
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
 	opts.accept_pathspec = 0;
+	opts.switch_branch_doing_nothing_not_ok = 1;
 
 	options = parse_options_dup(switch_options);
 	options = add_common_options(&opts, options);
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 11/14] switch-branch: only allow explicit detached HEAD

[Nguyễn Thái Ngọc Duy]

"git checkout <commit>" will checkout the commit in question and
detach HEAD from the current branch. It is naturally a right thing to
do once you get git references. But detached HEAD is a scary concept
to new users because we show a lot of warnings and stuff, and it could
be hard to get out of (until you know better).

To keep switch-branch a bit more friendly to new users, we only allow
entering detached HEAD mode when --detach is given. "git
switch-branch" must take a branch (unless you create a new branch,
then of course switch-branch can take any commit-ish)

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index c7ae068d2c..fbfebba2d9 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -49,6 +49,7 @@ struct checkout_opts {
 	int merge;
 	int force;
 	int force_detach;
+	int implicit_detach;
 	int writeout_stage;
 	int overwrite_ignore;
 	int ignore_skipworktree;
@@ -1241,6 +1242,13 @@ static int checkout_branch(struct checkout_opts *opts,
 	    !opts->force_detach)
 		die(_("nothing to do"));
 
+	if (!opts->implicit_detach &&
+	    !opts->new_branch &&
+	    !opts->new_branch_force &&
+	    new_branch_info->name &&
+	    !new_branch_info->path)
+		die(_("a branch is expected, got %s"), new_branch_info->name);
+
 	if (new_branch_info->path && !opts->force_detach && !opts->new_branch &&
 	    !opts->ignore_other_worktrees) {
 		int flag;
@@ -1485,6 +1493,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	opts.dwim_new_local_branch = 1;
 	opts.switch_branch_doing_nothing_not_ok = 0;
 	opts.accept_pathspec = 1;
+	opts.implicit_detach = 1;
 
 	options = parse_options_dup(checkout_options);
 	options = add_common_options(&opts, options);
@@ -1513,6 +1522,7 @@ int cmd_switch_branch(int argc, const char **argv, const char *prefix)
 	opts.dwim_new_local_branch = 1;
 	opts.accept_pathspec = 0;
 	opts.switch_branch_doing_nothing_not_ok = 1;
+	opts.implicit_detach = 0;
 
 	options = parse_options_dup(switch_options);
 	options = add_common_options(&opts, options);
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 12/14] restore-files: take tree-ish from --from option instead

[Nguyễn Thái Ngọc Duy]

This is another departure from 'git checkout' syntax, which uses -- to
separate ref and pathspec. The observation is restore-files (or "git
checkout ,, <pathspec>") is most often used to restore some files from
the index. If this is correct, we can simplify it by taking a way the
ref, so that we can write

    git restore-files some-file

without worrying about some-file being a ref and whether we need to do

    git restore-files -- some-file

for safety. If the source of the restore comes from a tree, it will be
in the form of an option with value, e.g.

    git restore-files --from=this-tree some-file

This is of course longer to type than using "--". But hopefully it
will not be used as often, and it is clearly easier to understand.

dwim_new_local_branch is no longer set (or unset) in cmd_restore_files()
because it's irrelevant because we don't really care about dwim-ing.
With accept_ref being unset, dwim can't happen.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 41 ++++++++++++++++++++++++++++++++++-------
 1 file changed, 34 insertions(+), 7 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index fbfebba2d9..7ff9951818 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -39,7 +39,7 @@ static const char * const switch_branch_usage[] = {
 };
 
 static const char * const restore_files_usage[] = {
-	N_("git restore-files [<options>] [<branch>] -- <file>..."),
+	N_("git restore-files [<options>] [--from=<branch>] <file>..."),
 	NULL,
 };
 
@@ -56,6 +56,7 @@ struct checkout_opts {
 	int ignore_other_worktrees;
 	int show_progress;
 	int dwim_new_local_branch;
+	int accept_ref;
 	int accept_pathspec;
 	int switch_branch_doing_nothing_not_ok;
 
@@ -75,6 +76,7 @@ struct checkout_opts {
 	int branch_exists;
 	const char *prefix;
 	struct pathspec pathspec;
+	const char *from_treeish;
 	struct tree *source_tree;
 };
 
@@ -1337,6 +1339,7 @@ static int checkout_main(int argc, const char **argv, const char *prefix,
 {
 	struct branch_info new_branch_info;
 	int dwim_remotes_matched = 0;
+	int parseopt_flags = 0;
 
 	memset(&new_branch_info, 0, sizeof(new_branch_info));
 	opts->overwrite_ignore = 1;
@@ -1347,8 +1350,13 @@ static int checkout_main(int argc, const char **argv, const char *prefix,
 
 	opts->track = BRANCH_TRACK_UNSPECIFIED;
 
-	argc = parse_options(argc, argv, prefix, options, usagestr,
-			     PARSE_OPT_KEEP_DASHDASH);
+	if (!opts->accept_pathspec && !opts->accept_ref)
+		BUG("make up your mind, you need to take _something_");
+	if (opts->accept_pathspec && opts->accept_ref)
+		parseopt_flags = PARSE_OPT_KEEP_DASHDASH;
+
+	argc = parse_options(argc, argv, prefix, options,
+			     usagestr, parseopt_flags);
 
 	if (opts->show_progress < 0) {
 		if (opts->quiet)
@@ -1402,7 +1410,7 @@ static int checkout_main(int argc, const char **argv, const char *prefix,
 	 * including "last branch" syntax and DWIM-ery for names of
 	 * remote branches, erroring out for invalid or ambiguous cases.
 	 */
-	if (argc) {
+	if (argc && opts->accept_ref && opts->accept_pathspec) {
 		struct object_id rev;
 		int dwim_ok =
 			!opts->patch_mode &&
@@ -1414,6 +1422,18 @@ static int checkout_main(int argc, const char **argv, const char *prefix,
 					     &dwim_remotes_matched);
 		argv += n;
 		argc -= n;
+	} else if (!opts->accept_ref && opts->from_treeish) {
+		struct object_id rev;
+
+		if (get_oid_mb(opts->from_treeish, &rev))
+			die(_("could not resolve %s"), opts->from_treeish);
+
+		setup_new_branch_info_and_source_tree(&new_branch_info,
+						      opts, &rev,
+						      opts->from_treeish);
+
+		if (!opts->source_tree)
+			die(_("reference is not a tree: %s"), opts->from_treeish);
 	}
 
 	if (argc) {
@@ -1492,6 +1512,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
 	opts.switch_branch_doing_nothing_not_ok = 0;
+	opts.accept_ref = 1;
 	opts.accept_pathspec = 1;
 	opts.implicit_detach = 1;
 
@@ -1520,6 +1541,7 @@ int cmd_switch_branch(int argc, const char **argv, const char *prefix)
 
 	memset(&opts, 0, sizeof(opts));
 	opts.dwim_new_local_branch = 1;
+	opts.accept_ref = 1;
 	opts.accept_pathspec = 0;
 	opts.switch_branch_doing_nothing_not_ok = 1;
 	opts.implicit_detach = 0;
@@ -1537,14 +1559,19 @@ int cmd_switch_branch(int argc, const char **argv, const char *prefix)
 int cmd_restore_files(int argc, const char **argv, const char *prefix)
 {
 	struct checkout_opts opts;
-	struct option *options = NULL;
+	struct option *options;
+	struct option restore_options[] = {
+		OPT_STRING(0, "from", &opts.from_treeish, "<tree-ish>",
+			   N_("where the checkout from")),
+		OPT_END()
+	};
 	int ret;
 
 	memset(&opts, 0, sizeof(opts));
-	opts.dwim_new_local_branch = 1;
+	opts.accept_ref = 0;
 	opts.accept_pathspec = 1;
 
-	options = parse_options_dup(options);
+	options = parse_options_dup(restore_options);
 	options = add_common_options(&opts, options);
 	options = add_checkout_path_options(&opts, options);
 
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 13/14] restore-files: make pathspec mandatory

[Nguyễn Thái Ngọc Duy]

"git restore-files" without arguments does not make much sense when
it's about restoring files (what files now?). We could default to
either

    git restore-files .

or

    git restore-files :/

Neither is intuitive. Make the user always give pathspec, force the
user to think the scope of restore they want.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 7ff9951818..961a90b1c0 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -59,6 +59,7 @@ struct checkout_opts {
 	int accept_ref;
 	int accept_pathspec;
 	int switch_branch_doing_nothing_not_ok;
+	int empty_pathspec_ok;
 
 	/*
 	 * If new checkout options are added, skip_merge_working_tree
@@ -1436,6 +1437,9 @@ static int checkout_main(int argc, const char **argv, const char *prefix,
 			die(_("reference is not a tree: %s"), opts->from_treeish);
 	}
 
+	if (opts->accept_pathspec && !opts->empty_pathspec_ok && !argc)
+		die(_("pathspec is required"));
+
 	if (argc) {
 		parse_pathspec(&opts->pathspec, 0,
 			       opts->patch_mode ? PATHSPEC_PREFIX_ORIGIN : 0,
@@ -1515,6 +1519,7 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
 	opts.accept_ref = 1;
 	opts.accept_pathspec = 1;
 	opts.implicit_detach = 1;
+	opts.empty_pathspec_ok = 1;
 
 	options = parse_options_dup(checkout_options);
 	options = add_common_options(&opts, options);
@@ -1570,6 +1575,7 @@ int cmd_restore_files(int argc, const char **argv, const char *prefix)
 	memset(&opts, 0, sizeof(opts));
 	opts.accept_ref = 0;
 	opts.accept_pathspec = 1;
+	opts.empty_pathspec_ok = 0;
 
 	options = parse_options_dup(restore_options);
 	options = add_common_options(&opts, options);
-- 
2.20.0.rc1.380.g3eb999425c.dirty

[PATCH v3 14/14] doc: promote "git switch-branch" and "git restore-files"

[Nguyễn Thái Ngọc Duy]

The two new commands "git switch-branch" and "git restore-files" are
added to avoid the confusion of one-command-do-all "git checkout" for
new users. They are also helpful to avoid ambiguation context.

For these reasons, promote them everywhere possible. This includes
documentation, suggestions/advice from other commands...

"git checkout" is also removed from "git help" (i.e. it's no longer
considered a commonly used command)

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 Documentation/config/advice.txt        | 10 +++--
 Documentation/config/checkout.txt      |  5 ++-
 Documentation/git-branch.txt           |  8 ++--
 Documentation/git-check-ref-format.txt |  2 +-
 Documentation/git-format-patch.txt     |  2 +-
 Documentation/git-merge-base.txt       |  2 +-
 Documentation/git-rebase.txt           |  2 +-
 Documentation/git-remote.txt           |  2 +-
 Documentation/git-rerere.txt           | 10 ++---
 Documentation/git-reset.txt            | 18 ++++-----
 Documentation/git-revert.txt           |  2 +-
 Documentation/git-stash.txt            |  6 +--
 Documentation/gitattributes.txt        |  2 +-
 Documentation/gitcli.txt               |  4 +-
 Documentation/gitcore-tutorial.txt     | 18 ++++-----
 Documentation/giteveryday.txt          | 24 ++++++------
 Documentation/githooks.txt             |  5 ++-
 Documentation/gittutorial-2.txt        |  2 +-
 Documentation/gittutorial.txt          |  4 +-
 Documentation/revisions.txt            |  2 +-
 Documentation/user-manual.txt          | 54 +++++++++++++-------------
 advice.c                               | 11 ++++--
 command-list.txt                       |  2 +-
 sha1-name.c                            |  2 +-
 wt-status.c                            |  2 +-
 25 files changed, 104 insertions(+), 97 deletions(-)

diff --git a/Documentation/config/advice.txt b/Documentation/config/advice.txt
index 57fcd4c862..bffc503385 100644
--- a/Documentation/config/advice.txt
+++ b/Documentation/config/advice.txt
@@ -35,7 +35,8 @@ advice.*::
 		state in the output of linkgit:git-status[1], in
 		the template shown when writing commit messages in
 		linkgit:git-commit[1], and in the help message shown
-		by linkgit:git-checkout[1] when switching branch.
+		by linkgit:git-switch-branch[1] or
+		linkgit:git-checkout[1] when switching branch.
 	statusUoption::
 		Advise to consider using the `-u` option to linkgit:git-status[1]
 		when the command takes more than 2 seconds to enumerate untracked
@@ -55,9 +56,10 @@ advice.*::
 		your information is guessed from the system username and
 		domain name.
 	detachedHead::
-		Advice shown when you used linkgit:git-checkout[1] to
-		move to the detach HEAD state, to instruct how to create
-		a local branch after the fact.
+		Advice shown when you used
+		linkgit:git-switch-branch[1] or linkgit:git-checkout[1]
+		to move to the detach HEAD state, to instruct how to
+		create a local branch after the fact.
 	checkoutAmbiguousRemoteBranchName::
 		Advice shown when the argument to
 		linkgit:git-checkout[1] ambiguously resolves to a
diff --git a/Documentation/config/checkout.txt b/Documentation/config/checkout.txt
index c4118fa196..81b0d47ced 100644
--- a/Documentation/config/checkout.txt
+++ b/Documentation/config/checkout.txt
@@ -8,8 +8,9 @@ checkout.defaultRemote::
 	disambiguation. The typical use-case is to set this to
 	`origin`.
 +
-Currently this is used by linkgit:git-checkout[1] when 'git checkout
-<something>' will checkout the '<something>' branch on another remote,
+Currently this is used by linkgit:git-switch-branch[1] and
+linkgit:git-checkout[1] when 'git checkout <something>'
+will checkout the '<something>' branch on another remote,
 and by linkgit:git-worktree[1] when 'git worktree add' refers to a
 remote branch. This setting might be used for other checkout-like
 commands or functionality in the future.
diff --git a/Documentation/git-branch.txt b/Documentation/git-branch.txt
index bf5316ffa9..1564df47d2 100644
--- a/Documentation/git-branch.txt
+++ b/Documentation/git-branch.txt
@@ -48,7 +48,7 @@ The command's second form creates a new branch head named <branchname>
 which points to the current `HEAD`, or <start-point> if given.
 
 Note that this will create the new branch, but it will not switch the
-working tree to it; use "git checkout <newbranch>" to switch to the
+working tree to it; use "git switch-branch <newbranch>" to switch to the
 new branch.
 
 When a local branch is started off a remote-tracking branch, Git sets up the
@@ -194,7 +194,7 @@ This option is only applicable in non-verbose mode.
 +
 This behavior is the default when the start point is a remote-tracking branch.
 Set the branch.autoSetupMerge configuration variable to `false` if you
-want `git checkout` and `git branch` to always behave as if `--no-track`
+want `git switch-branch` and `git branch` to always behave as if `--no-track`
 were given. Set it to `always` if you want this behavior when the
 start-point is either a local or remote-tracking branch.
 
@@ -293,7 +293,7 @@ Start development from a known tag::
 $ git clone git://git.kernel.org/pub/scm/.../linux-2.6 my2.6
 $ cd my2.6
 $ git branch my2.6.14 v2.6.14   <1>
-$ git checkout my2.6.14
+$ git switch-branch my2.6.14
 ------------
 +
 <1> This step and the next one could be combined into a single step with
@@ -319,7 +319,7 @@ NOTES
 -----
 
 If you are creating a branch that you want to checkout immediately, it is
-easier to use the git checkout command with its `-b` option to create
+easier to use the "git switch-branch" command with its `-b` option to create
 a branch and check it out with a single command.
 
 The options `--contains`, `--no-contains`, `--merged` and `--no-merged`
diff --git a/Documentation/git-check-ref-format.txt b/Documentation/git-check-ref-format.txt
index d9de992585..38c2169d7a 100644
--- a/Documentation/git-check-ref-format.txt
+++ b/Documentation/git-check-ref-format.txt
@@ -88,7 +88,7 @@ but it is explicitly forbidden at the beginning of a branch name).
 When run with `--branch` option in a repository, the input is first
 expanded for the ``previous checkout syntax''
 `@{-n}`.  For example, `@{-1}` is a way to refer the last thing that
-was checked out using "git checkout" operation. This option should be
+was checked out using "git switch-branch" operation. This option should be
 used by porcelains to accept this syntax anywhere a branch name is
 expected, so they can act as if you typed the branch name. As an
 exception note that, the ``previous checkout operation'' might result
diff --git a/Documentation/git-format-patch.txt b/Documentation/git-format-patch.txt
index aba4c5febe..0ceaa1173c 100644
--- a/Documentation/git-format-patch.txt
+++ b/Documentation/git-format-patch.txt
@@ -416,7 +416,7 @@ One way to test if your MUA is set up correctly is:
 * Apply it:
 
     $ git fetch <project> master:test-apply
-    $ git checkout test-apply
+    $ git switch-branch test-apply
     $ git reset --hard
     $ git am a.patch
 
diff --git a/Documentation/git-merge-base.txt b/Documentation/git-merge-base.txt
index 9f07f4f6ed..1b25e5d530 100644
--- a/Documentation/git-merge-base.txt
+++ b/Documentation/git-merge-base.txt
@@ -149,7 +149,7 @@ instead.
 Discussion on fork-point mode
 -----------------------------
 
-After working on the `topic` branch created with `git checkout -b
+After working on the `topic` branch created with `git switch-branch -b
 topic origin/master`, the history of remote-tracking branch
 `origin/master` may have been rewound and rebuilt, leading to a
 history of this shape:
diff --git a/Documentation/git-rebase.txt b/Documentation/git-rebase.txt
index 80793bad8d..fe10880633 100644
--- a/Documentation/git-rebase.txt
+++ b/Documentation/git-rebase.txt
@@ -17,7 +17,7 @@ SYNOPSIS
 DESCRIPTION
 -----------
 If <branch> is specified, 'git rebase' will perform an automatic
-`git checkout <branch>` before doing anything else.  Otherwise
+`git switch-branch <branch>` before doing anything else.  Otherwise
 it remains on the current branch.
 
 If <upstream> is not specified, the upstream configured in
diff --git a/Documentation/git-remote.txt b/Documentation/git-remote.txt
index 0cad37fb81..044bbdb27c 100644
--- a/Documentation/git-remote.txt
+++ b/Documentation/git-remote.txt
@@ -230,7 +230,7 @@ $ git branch -r
   staging/master
   staging/staging-linus
   staging/staging-next
-$ git checkout -b staging staging/master
+$ git switch-branch -b staging staging/master
 ...
 ------------
 
diff --git a/Documentation/git-rerere.txt b/Documentation/git-rerere.txt
index df310d2a58..fe9d21b395 100644
--- a/Documentation/git-rerere.txt
+++ b/Documentation/git-rerere.txt
@@ -91,7 +91,7 @@ For such a test, you need to merge master and topic somehow.
 One way to do it is to pull master into the topic branch:
 
 ------------
-	$ git checkout topic
+	$ git switch-branch topic
 	$ git merge master
 
               o---*---o---+ topic
@@ -113,10 +113,10 @@ the upstream might have been advanced since the test merge `+`,
 in which case the final commit graph would look like this:
 
 ------------
-	$ git checkout topic
+	$ git switch-branch topic
 	$ git merge master
 	$ ... work on both topic and master branches
-	$ git checkout master
+	$ git switch-branch master
 	$ git merge topic
 
               o---*---o---+---o---o topic
@@ -136,11 +136,11 @@ merges, you could blow away the test merge, and keep building on
 top of the tip before the test merge:
 
 ------------
-	$ git checkout topic
+	$ git switch-branch topic
 	$ git merge master
 	$ git reset --hard HEAD^ ;# rewind the test merge
 	$ ... work on both topic and master branches
-	$ git checkout master
+	$ git switch-branch master
 	$ git merge topic
 
               o---*---o-------o---o topic
diff --git a/Documentation/git-reset.txt b/Documentation/git-reset.txt
index 2dac95c71a..ca46b4c967 100644
--- a/Documentation/git-reset.txt
+++ b/Documentation/git-reset.txt
@@ -149,9 +149,9 @@ See also the --amend option to linkgit:git-commit[1].
 Undo a commit, making it a topic branch::
 +
 ------------
-$ git branch topic/wip     <1>
-$ git reset --hard HEAD~3  <2>
-$ git checkout topic/wip   <3>
+$ git branch topic/wip          <1>
+$ git reset --hard HEAD~3       <2>
+$ git switch-branch topic/wip   <3>
 ------------
 +
 <1> You have made some commits, but realize they were premature
@@ -232,13 +232,13 @@ working tree are not in any shape to be committed yet, but you
 need to get to the other branch for a quick bugfix.
 +
 ------------
-$ git checkout feature ;# you were working in "feature" branch and
+$ git switch-branch feature ;# you were working in "feature" branch and
 $ work work work       ;# got interrupted
 $ git commit -a -m "snapshot WIP"                 <1>
-$ git checkout master
+$ git switch-branch master
 $ fix fix fix
 $ git commit ;# commit with real log
-$ git checkout feature
+$ git switch-branch feature
 $ git reset --soft HEAD^ ;# go back to WIP state  <2>
 $ git reset                                       <3>
 ------------
@@ -279,18 +279,18 @@ reset it while keeping the changes in your working tree.
 +
 ------------
 $ git tag start
-$ git checkout -b branch1
+$ git switch-branch -b branch1
 $ edit
 $ git commit ...                            <1>
 $ edit
-$ git checkout -b branch2                   <2>
+$ git switch-branch -b branch2              <2>
 $ git reset --keep start                    <3>
 ------------
 +
 <1> This commits your first edits in branch1.
 <2> In the ideal world, you could have realized that the earlier
     commit did not belong to the new topic when you created and switched
-    to branch2 (i.e. "git checkout -b branch2 start"), but nobody is
+    to branch2 (i.e. "git switch-branch -b branch2 start"), but nobody is
     perfect.
 <3> But you can use "reset --keep" to remove the unwanted commit after
     you switched to "branch2".
diff --git a/Documentation/git-revert.txt b/Documentation/git-revert.txt
index 837707a8fd..07ef83866b 100644
--- a/Documentation/git-revert.txt
+++ b/Documentation/git-revert.txt
@@ -26,7 +26,7 @@ effect of some earlier commits (often only a faulty one).  If you want to
 throw away all uncommitted changes in your working directory, you
 should see linkgit:git-reset[1], particularly the `--hard` option.  If
 you want to extract specific files as they were in another commit, you
-should see linkgit:git-checkout[1], specifically the `git checkout
+should see linkgit:git-checkout[1], specifically the `git restore-files
 <commit> -- <filename>` syntax.  Take care with these alternatives as
 both will discard uncommitted changes in your working directory.
 
diff --git a/Documentation/git-stash.txt b/Documentation/git-stash.txt
index 7ef8c47911..ea226979b1 100644
--- a/Documentation/git-stash.txt
+++ b/Documentation/git-stash.txt
@@ -235,12 +235,12 @@ return to your original branch to make the emergency fix, like this:
 +
 ----------------------------------------------------------------
 # ... hack hack hack ...
-$ git checkout -b my_wip
+$ git switch-branch -b my_wip
 $ git commit -a -m "WIP"
-$ git checkout master
+$ git switch-branch master
 $ edit emergency fix
 $ git commit -a -m "Fix in a hurry"
-$ git checkout my_wip
+$ git switch-branch my_wip
 $ git reset --soft HEAD^
 # ... continue hacking ...
 ----------------------------------------------------------------
diff --git a/Documentation/gitattributes.txt b/Documentation/gitattributes.txt
index b8392fc330..df62bd8019 100644
--- a/Documentation/gitattributes.txt
+++ b/Documentation/gitattributes.txt
@@ -112,7 +112,7 @@ Checking-out and checking-in
 
 These attributes affect how the contents stored in the
 repository are copied to the working tree files when commands
-such as 'git checkout' and 'git merge' run.  They also affect how
+such as 'git switch-branch' and 'git merge' run.  They also affect how
 Git stores the contents you prepare in the working tree in the
 repository upon 'git add' and 'git commit'.
 
diff --git a/Documentation/gitcli.txt b/Documentation/gitcli.txt
index 592e06d839..491eb91c2e 100644
--- a/Documentation/gitcli.txt
+++ b/Documentation/gitcli.txt
@@ -47,8 +47,8 @@ disambiguating `--` at appropriate places.
    things:
 +
 --------------------------------
-$ git checkout -- *.c
-$ git checkout -- \*.c
+$ git restore-files -- *.c
+$ git restore-files -- \*.c
 --------------------------------
 +
 The former lets your shell expand the fileglob, and you are asking
diff --git a/Documentation/gitcore-tutorial.txt b/Documentation/gitcore-tutorial.txt
index e29a9effcc..49a8b5aa52 100644
--- a/Documentation/gitcore-tutorial.txt
+++ b/Documentation/gitcore-tutorial.txt
@@ -741,7 +741,7 @@ used earlier, and create a branch in it. You do that by simply just
 saying that you want to check out a new branch:
 
 ------------
-$ git checkout -b mybranch
+$ git branch mybranch
 ------------
 
 will create a new branch based at the current `HEAD` position, and switch
@@ -755,7 +755,7 @@ just telling 'git checkout' what the base of the checkout would be.
 In other words, if you have an earlier tag or branch, you'd just do
 
 ------------
-$ git checkout -b mybranch earlier-commit
+$ git switch-branch -b mybranch earlier-commit
 ------------
 
 and it would create the new branch `mybranch` at the earlier commit,
@@ -765,7 +765,7 @@ and check out the state at that time.
 You can always just jump back to your original `master` branch by doing
 
 ------------
-$ git checkout master
+$ git switch-branch master
 ------------
 
 (or any other branch-name, for that matter) and if you forget which
@@ -794,7 +794,7 @@ $ git branch <branchname> [startingpoint]
 
 which will simply _create_ the branch, but will not do anything further.
 You can then later -- once you decide that you want to actually develop
-on that branch -- switch to that branch with a regular 'git checkout'
+on that branch -- switch to that branch with a regular 'git switch-branch
 with the branchname as the argument.
 
 
@@ -808,7 +808,7 @@ being the same as the original `master` branch, let's make sure we're in
 that branch, and do some work there.
 
 ------------------------------------------------
-$ git checkout mybranch
+$ git switch-branch mybranch
 $ echo "Work, work, work" >>hello
 $ git commit -m "Some work." -i hello
 ------------------------------------------------
@@ -825,7 +825,7 @@ does some work in the original branch, and simulate that by going back
 to the master branch, and editing the same file differently there:
 
 ------------
-$ git checkout master
+$ git switch-branch master
 ------------
 
 Here, take a moment to look at the contents of `hello`, and notice how they
@@ -958,7 +958,7 @@ to the `master` branch. Let's go back to `mybranch`, and run
 'git merge' to get the "upstream changes" back to your branch.
 
 ------------
-$ git checkout mybranch
+$ git switch-branch mybranch
 $ git merge -m "Merge upstream changes." master
 ------------
 
@@ -1133,9 +1133,9 @@ Remember, before running 'git merge', our `master` head was at
 work." commit.
 
 ------------
-$ git checkout mybranch
+$ git switch-branch mybranch
 $ git reset --hard master^2
-$ git checkout master
+$ git switch-branch master
 $ git reset --hard master^
 ------------
 
diff --git a/Documentation/giteveryday.txt b/Documentation/giteveryday.txt
index 9f2528fc8c..9d64544bb9 100644
--- a/Documentation/giteveryday.txt
+++ b/Documentation/giteveryday.txt
@@ -80,9 +80,9 @@ $ git tag v2.43 <2>
 Create a topic branch and develop.::
 +
 ------------
-$ git checkout -b alsa-audio <1>
+$ git branch alsa-audio <1>
 $ edit/compile/test
-$ git checkout -- curses/ux_audio_oss.c <2>
+$ git restore-files -- curses/ux_audio_oss.c <2>
 $ git add curses/ux_audio_alsa.c <3>
 $ edit/compile/test
 $ git diff HEAD <4>
@@ -90,7 +90,7 @@ $ git commit -a -s <5>
 $ edit/compile/test
 $ git diff HEAD^ <6>
 $ git commit -a --amend <7>
-$ git checkout master <8>
+$ git switch-branch master <8>
 $ git merge alsa-audio <9>
 $ git log --since='3 days ago' <10>
 $ git log v2.43.. curses/ <11>
@@ -148,11 +148,11 @@ Clone the upstream and work on it.  Feed changes to upstream.::
 ------------
 $ git clone git://git.kernel.org/pub/scm/.../torvalds/linux-2.6 my2.6
 $ cd my2.6
-$ git checkout -b mine master <1>
+$ git switch-branch -b mine master <1>
 $ edit/compile/test; git commit -a -s <2>
 $ git format-patch master <3>
 $ git send-email --to="person <email@example.com>" 00*.patch <4>
-$ git checkout master <5>
+$ git switch-branch master <5>
 $ git pull <6>
 $ git log -p ORIG_HEAD.. arch/i386 include/asm-i386 <7>
 $ git ls-remote --heads http://git.kernel.org/.../jgarzik/libata-dev.git <8>
@@ -194,7 +194,7 @@ satellite$ edit/compile/test/commit
 satellite$ git push origin <4>
 
 mothership$ cd frotz
-mothership$ git checkout master
+mothership$ git switch-branch master
 mothership$ git merge satellite/master <5>
 ------------
 +
@@ -216,7 +216,7 @@ machine into the master branch.
 Branch off of a specific tag.::
 +
 ------------
-$ git checkout -b private2.6.14 v2.6.14 <1>
+$ git switch-branch -b private2.6.14 v2.6.14 <1>
 $ edit/compile/test; git commit -a
 $ git checkout master
 $ git cherry-pick v2.6.14..private2.6.14 <2>
@@ -274,14 +274,14 @@ $ mailx <3>
 & s 2 3 4 5 ./+to-apply
 & s 7 8 ./+hold-linus
 & q
-$ git checkout -b topic/one master
+$ git switch-branch -b topic/one master
 $ git am -3 -i -s ./+to-apply <4>
 $ compile/test
-$ git checkout -b hold/linus && git am -3 -i -s ./+hold-linus <5>
-$ git checkout topic/one && git rebase master <6>
-$ git checkout pu && git reset --hard next <7>
+$ git switch-branch -b hold/linus && git am -3 -i -s ./+hold-linus <5>
+$ git switch-branch topic/one && git rebase master <6>
+$ git switch-branch pu && git reset --hard next <7>
 $ git merge topic/one topic/two && git merge hold/linus <8>
-$ git checkout maint
+$ git switch-branch maint
 $ git cherry-pick master~4 <9>
 $ compile/test
 $ git tag -s -m "GIT 0.99.9x" v0.99.9x <10>
diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt
index 959044347e..3939ec774a 100644
--- a/Documentation/githooks.txt
+++ b/Documentation/githooks.txt
@@ -166,7 +166,7 @@ worktree.  The hook is given three parameters: the ref of the previous HEAD,
 the ref of the new HEAD (which may or may not have changed), and a flag
 indicating whether the checkout was a branch checkout (changing branches,
 flag=1) or a file checkout (retrieving a file from the index, flag=0).
-This hook cannot affect the outcome of `git checkout`.
+This hook cannot affect the outcome of `git switch-branch` or `git checkout`.
 
 It is also run after linkgit:git-clone[1], unless the `--no-checkout` (`-n`) option is
 used. The first parameter given to the hook is the null-ref, the second the
@@ -402,7 +402,8 @@ exit with a zero status.
 For example, the hook can simply run `git read-tree -u -m HEAD "$1"`
 in order to emulate `git fetch` that is run in the reverse direction
 with `git push`, as the two-tree form of `git read-tree -u -m` is
-essentially the same as `git checkout` that switches branches while
+essentially the same as `git switch-branch` or `git checkout`
+that switches branches while
 keeping the local changes in the working tree that do not interfere
 with the difference between the branches.
 
diff --git a/Documentation/gittutorial-2.txt b/Documentation/gittutorial-2.txt
index e0976f6017..1ec14da6b4 100644
--- a/Documentation/gittutorial-2.txt
+++ b/Documentation/gittutorial-2.txt
@@ -376,7 +376,7 @@ Changes to be committed:
 
 Changes not staged for commit:
   (use "git add <file>..." to update what will be committed)
-  (use "git checkout -- <file>..." to discard changes in working directory)
+  (use "git restore-files -- <file>..." to discard changes in working directory)
 
 	modified:   file.txt
 
diff --git a/Documentation/gittutorial.txt b/Documentation/gittutorial.txt
index 242de31cb6..396e55c191 100644
--- a/Documentation/gittutorial.txt
+++ b/Documentation/gittutorial.txt
@@ -207,7 +207,7 @@ automatically.  The asterisk marks the branch you are currently on;
 type
 
 ------------------------------------------------
-$ git checkout experimental
+$ git switch-branch experimental
 ------------------------------------------------
 
 to switch to the experimental branch.  Now edit a file, commit the
@@ -216,7 +216,7 @@ change, and switch back to the master branch:
 ------------------------------------------------
 (edit file)
 $ git commit -a
-$ git checkout master
+$ git switch-branch master
 ------------------------------------------------
 
 Check that the change you made is no longer visible, since it was
diff --git a/Documentation/revisions.txt b/Documentation/revisions.txt
index 72daa20e76..f55502cd50 100644
--- a/Documentation/revisions.txt
+++ b/Documentation/revisions.txt
@@ -115,7 +115,7 @@ Here's an example to make it more clear:
 ------------------------------
 $ git config push.default current
 $ git config remote.pushdefault myfork
-$ git checkout -b mybranch origin/master
+$ git switch-branch -b mybranch origin/master
 
 $ git rev-parse --symbolic-full-name @{upstream}
 refs/remotes/origin/master
diff --git a/Documentation/user-manual.txt b/Documentation/user-manual.txt
index eff7890274..e3ff98077d 100644
--- a/Documentation/user-manual.txt
+++ b/Documentation/user-manual.txt
@@ -125,7 +125,7 @@ Create a new branch head pointing to one of these versions and check it
 out using linkgit:git-checkout[1]:
 
 ------------------------------------------------
-$ git checkout -b new v2.6.13
+$ git switch-branch -b new v2.6.13
 ------------------------------------------------
 
 The working directory then reflects the contents that the project had
@@ -282,10 +282,10 @@ a summary of the commands:
 	this command will fail with a warning.
 `git branch -D <branch>`::
 	delete the branch `<branch>` irrespective of its merged status.
-`git checkout <branch>`::
+`git switch-branch <branch>`::
 	make the current branch `<branch>`, updating the working
 	directory to reflect the version referenced by `<branch>`.
-`git checkout -b <new> <start-point>`::
+`git switch-branch -b <new> <start-point>`::
 	create a new branch `<new>` referencing `<start-point>`, and
 	check it out.
 
@@ -302,12 +302,12 @@ ref: refs/heads/master
 Examining an old version without creating a new branch
 ------------------------------------------------------
 
-The `git checkout` command normally expects a branch head, but will also
+The `git switch-branch` command normally expects a branch head, but will also
 accept an arbitrary commit; for example, you can check out the commit
 referenced by a tag:
 
 ------------------------------------------------
-$ git checkout v2.6.17
+$ git switch-branch v2.6.17
 Note: checking out 'v2.6.17'.
 
 You are in 'detached HEAD' state. You can look around, make experimental
@@ -317,7 +317,7 @@ state without impacting any branches by performing another checkout.
 If you want to create a new branch to retain commits you create, you may
 do so (now or later) by using -b with the checkout command again. Example:
 
-  git checkout -b new_branch_name
+  git switch-branch -b new_branch_name
 
 HEAD is now at 427abfa Linux v2.6.17
 ------------------------------------------------
@@ -373,7 +373,7 @@ You might want to build on one of these remote-tracking branches
 on a branch of your own, just as you would for a tag:
 
 ------------------------------------------------
-$ git checkout -b my-todo-copy origin/todo
+$ git switch-branch -b my-todo-copy origin/todo
 ------------------------------------------------
 
 You can also check out `origin/todo` directly to examine it or
@@ -1523,12 +1523,12 @@ Checking out an old version of a file
 
 In the process of undoing a previous bad change, you may find it
 useful to check out an older version of a particular file using
-linkgit:git-checkout[1].  We've used `git checkout` before to switch
+linkgit:git-checkout[1].  We've used `git switch-branch` before to switch
 branches, but it has quite different behavior if it is given a path
 name: the command
 
 -------------------------------------------------
-$ git checkout HEAD^ path/to/file
+$ git restore-files HEAD^ path/to/file
 -------------------------------------------------
 
 replaces path/to/file by the contents it had in the commit HEAD^, and
@@ -2211,8 +2211,8 @@ $ git branch --track release origin/master
 These can be easily kept up to date using linkgit:git-pull[1].
 
 -------------------------------------------------
-$ git checkout test && git pull
-$ git checkout release && git pull
+$ git switch-branch test && git pull
+$ git switch-branch release && git pull
 -------------------------------------------------
 
 Important note!  If you have any local changes in these branches, then
@@ -2264,7 +2264,7 @@ tested changes
 2) help future bug hunters that use `git bisect` to find problems
 
 -------------------------------------------------
-$ git checkout -b speed-up-spinlocks v2.6.35
+$ git switch-branch -b speed-up-spinlocks v2.6.35
 -------------------------------------------------
 
 Now you apply the patch(es), run some tests, and commit the change(s).  If
@@ -2279,7 +2279,7 @@ When you are happy with the state of this change, you can merge it into the
 "test" branch in preparation to make it public:
 
 -------------------------------------------------
-$ git checkout test && git merge speed-up-spinlocks
+$ git switch-branch test && git merge speed-up-spinlocks
 -------------------------------------------------
 
 It is unlikely that you would have any conflicts here ... but you might if you
@@ -2291,7 +2291,7 @@ see the value of keeping each patch (or patch series) in its own branch.  It
 means that the patches can be moved into the `release` tree in any order.
 
 -------------------------------------------------
-$ git checkout release && git merge speed-up-spinlocks
+$ git switch-branch release && git merge speed-up-spinlocks
 -------------------------------------------------
 
 After a while, you will have a number of branches, and despite the
@@ -2358,7 +2358,7 @@ Here are some of the scripts that simplify all this even further.
 
 case "$1" in
 test|release)
-	git checkout $1 && git pull . origin
+	git switch-branch $1 && git pull . origin
 	;;
 origin)
 	before=$(git rev-parse refs/remotes/origin/master)
@@ -2400,7 +2400,7 @@ test|release)
 		echo $1 already merged into $2 1>&2
 		exit 1
 	fi
-	git checkout $2 && git pull . $1
+	git switch-branch $2 && git pull . $1
 	;;
 *)
 	usage
@@ -2512,7 +2512,7 @@ Suppose that you create a branch `mywork` on a remote-tracking branch
 `origin`, and create some commits on top of it:
 
 -------------------------------------------------
-$ git checkout -b mywork origin
+$ git switch-branch -b mywork origin
 $ vi file.txt
 $ git commit
 $ vi otherfile.txt
@@ -2552,7 +2552,7 @@ commits without any merges, you may instead choose to use
 linkgit:git-rebase[1]:
 
 -------------------------------------------------
-$ git checkout mywork
+$ git switch-branch mywork
 $ git rebase origin
 -------------------------------------------------
 
@@ -3668,13 +3668,13 @@ change within the submodule, and then update the superproject to reference the
 new commit:
 
 -------------------------------------------------
-$ git checkout master
+$ git switch-branch master
 -------------------------------------------------
 
 or
 
 -------------------------------------------------
-$ git checkout -b fix-up
+$ git switch-branch -b fix-up
 -------------------------------------------------
 
 then
@@ -4194,7 +4194,7 @@ start.
 A good place to start is with the contents of the initial commit, with:
 
 ----------------------------------------------------
-$ git checkout e83c5163
+$ git switch-branch e83c5163
 ----------------------------------------------------
 
 The initial revision lays the foundation for almost everything Git has
@@ -4437,10 +4437,10 @@ Managing branches
 -----------------
 
 -----------------------------------------------
-$ git branch	     # list all local branches in this repo
-$ git checkout test  # switch working directory to branch "test"
-$ git branch new     # create branch "new" starting at current HEAD
-$ git branch -d new  # delete branch "new"
+$ git branch			# list all local branches in this repo
+$ git switch-branch test	# switch working directory to branch "test"
+$ git branch new		# create branch "new" starting at current HEAD
+$ git branch -d new		# delete branch "new"
 -----------------------------------------------
 
 Instead of basing a new branch on current HEAD (the default), use:
@@ -4456,7 +4456,7 @@ $ git branch new test~10 # ten commits before tip of branch "test"
 Create and switch to a new branch at the same time:
 
 -----------------------------------------------
-$ git checkout -b new v2.6.15
+$ git switch-branch -b new v2.6.15
 -----------------------------------------------
 
 Update and examine branches from the repository you cloned from:
@@ -4467,7 +4467,7 @@ $ git branch -r		# list
   origin/master
   origin/next
   ...
-$ git checkout -b masterwork origin/master
+$ git switch-branch -b masterwork origin/master
 -----------------------------------------------
 
 Fetch a branch from a different repository, and give it a new
diff --git a/advice.c b/advice.c
index 5f35656409..578ea31c7e 100644
--- a/advice.c
+++ b/advice.c
@@ -189,13 +189,16 @@ void NORETURN die_conclude_merge(void)
 void detach_advice(const char *new_name)
 {
 	const char *fmt =
-	_("Note: checking out '%s'.\n\n"
+	_("Note: checking out '%s'.\n"
+	"\n"
 	"You are in 'detached HEAD' state. You can look around, make experimental\n"
 	"changes and commit them, and you can discard any commits you make in this\n"
-	"state without impacting any branches by performing another checkout.\n\n"
+	"state without impacting any branches by performing another checkout.\n"
+	"\n"
 	"If you want to create a new branch to retain commits you create, you may\n"
-	"do so (now or later) by using -b with the checkout command again. Example:\n\n"
-	"  git checkout -b <new-branch-name>\n\n");
+	"do so (now or later) by using -b with the checkout command again. Example:\n"
+	"\n"
+	"  git switch-branch -b <new-branch-name>\n\n");
 
 	fprintf(stderr, fmt, new_name);
 }
diff --git a/command-list.txt b/command-list.txt
index 4638802754..d1fb1d551d 100644
--- a/command-list.txt
+++ b/command-list.txt
@@ -59,7 +59,7 @@ git-cat-file                            plumbinginterrogators
 git-check-attr                          purehelpers
 git-check-ignore                        purehelpers
 git-check-mailmap                       purehelpers
-git-checkout                            mainporcelain           history
+git-checkout                            mainporcelain
 git-checkout-index                      plumbingmanipulators
 git-check-ref-format                    purehelpers
 git-cherry                              plumbinginterrogators          complete
diff --git a/sha1-name.c b/sha1-name.c
index faa60f69e3..4e4e14a45c 100644
--- a/sha1-name.c
+++ b/sha1-name.c
@@ -771,7 +771,7 @@ static int get_oid_basic(const char *str, int len, struct object_id *oid,
 	"because it will be ignored when you just specify 40-hex. These refs\n"
 	"may be created by mistake. For example,\n"
 	"\n"
-	"  git checkout -b $br $(git rev-parse ...)\n"
+	"  git switch-branch -b $br $(git rev-parse ...)\n"
 	"\n"
 	"where \"$br\" is somehow empty and a 40-hex ref is created. Please\n"
 	"examine these refs and maybe delete them. Turn this message off by\n"
diff --git a/wt-status.c b/wt-status.c
index a24711374c..c615cac607 100644
--- a/wt-status.c
+++ b/wt-status.c
@@ -224,7 +224,7 @@ static void wt_longstatus_print_dirty_header(struct wt_status *s,
 		status_printf_ln(s, c, _("  (use \"git add <file>...\" to update what will be committed)"));
 	else
 		status_printf_ln(s, c, _("  (use \"git add/rm <file>...\" to update what will be committed)"));
-	status_printf_ln(s, c, _("  (use \"git checkout -- <file>...\" to discard changes in working directory)"));
+	status_printf_ln(s, c, _("  (use \"git restore-files <file>...\" to discard changes in working directory)"));
 	if (has_dirty_submodules)
 		status_printf_ln(s, c, _("  (commit or discard the untracked or modified content in submodules)"));
 	status_printf_ln(s, c, "%s", "");
-- 
2.20.0.rc1.380.g3eb999425c.dirty

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Ævar Arnfjörð Bjarmason]

On Thu, Nov 29 2018, Nguyễn Thái Ngọc Duy wrote:

> v3 sees switch-branch go back to switch-branch (in v2 it was
> checkout-branch). checkout-files is also renamed restore-files (v1 was
> restore-paths). Hopefully we won't see another rename.
>
> I'll try to summarize the differences between the new commands and
> 'git checkout' down here, but you're welcome to just head to 07/14 and
> read the new man pages.
>
> 'git switch-branch'
>
> - does not "do nothing", you have to either switch branch, create a
>   new branch, or detach. "git switch-branch" with no arguments is
>   rejected.
>
> - implicit detaching is rejected. If you need to detach, you need to
>   give --detach. Or stick to 'git checkout'.
>
> - -b/-B is renamed to -c/-C with long option names
>
> - of course does not accept pathspec
>
> 'git restore-files'
>
> - takes a ref from --from argument, not as a free ref. As a result,
>   '--' is no longer needed. All non-option arguments are pathspec
>
> - pathspec is mandatory, you can't do "git restore-files" without any
>   pathspec.
>
> - I just remember -p which is allowed to take no pathspec :( I'll fix
>   it later.
>
> - Two more fancy features (the "git checkout --index" being the
>   default mode and the backup log for accidental overwrites) are of
>   course still missing. But they are coming.
>
> I did not go replace "detached HEAD" with "unnamed branch" (or "no
> branch") everywhere because I think a unique term is still good to
> refer to this concept. Or maybe "no branch" is good enough. I dunno.

I finally tracked down
https://redfin.engineering/two-commits-that-wrecked-the-user-experience-of-git-f0075b77eab1
which I'd remembered reading and couldn't find again in these
discussions. Re-reading it while one may not 100% agree with the
author's opinion, it's an interesting rabbit hole.

I also didn't know about EasyGit, or that Elijah Newren had written
it. I haven't seen him chime in on this series, and would be interested
to see what he thinks about it.

Re the naming question in
https://public-inbox.org/git/87o9abzv46.fsf@evledraar.gmail.com/ &
seeing that eg-switch exists, I wonder if just s/switch-branch/switch/
makes more sense.

Assuming greenfield development (which we definitely don't have), I
don't like the "restore-files" name, but the alternative that makes
sense is "checkout". Then this "--from" argument could become "git
checkout-tree <treeish> -- <pathspec>", and we'd have:

    git switch <branchish>
    git checkout <pathspec>
    git checkout-tree <treeish> -- <pathspec>

Or maybe that sucks, anyway what I was going to suggest is *if* others
think that made sense as a "if we designed git today" endgame whether we
could have an opt-in setting where you set e.g. core.uiVersion=3 (in
anticipation of Git 3.0) and you'd get that behavior. There could be
some other setting where core.uiVersion would use the old behavior (or
with another setting, only warn) if we weren't connected to a terminal.

I.e. I'm thinking of this as step #2 in a #3 step series. Where this is
the fully backwards compatible UI improvement, but someone who'd
e.g. use EasyGit or didn't have backwards compatibility concerns could
enable step #3 and opt-in to a mode where we'd fixed a bunch of UI warts
in a backwards-incompatible way.

What would that mode look like? I'd to work on piling that on top of
this :)

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Ævar Arnfjörð Bjarmason]

On Thu, Nov 29 2018, Ævar Arnfjörð Bjarmason wrote:

> On Thu, Nov 29 2018, Nguyễn Thái Ngọc Duy wrote:
>
>> v3 sees switch-branch go back to switch-branch (in v2 it was
>> checkout-branch). checkout-files is also renamed restore-files (v1 was
>> restore-paths). Hopefully we won't see another rename.
>>
>> I'll try to summarize the differences between the new commands and
>> 'git checkout' down here, but you're welcome to just head to 07/14 and
>> read the new man pages.
>>
>> 'git switch-branch'
>>
>> - does not "do nothing", you have to either switch branch, create a
>>   new branch, or detach. "git switch-branch" with no arguments is
>>   rejected.
>>
>> - implicit detaching is rejected. If you need to detach, you need to
>>   give --detach. Or stick to 'git checkout'.
>>
>> - -b/-B is renamed to -c/-C with long option names
>>
>> - of course does not accept pathspec
>>
>> 'git restore-files'
>>
>> - takes a ref from --from argument, not as a free ref. As a result,
>>   '--' is no longer needed. All non-option arguments are pathspec
>>
>> - pathspec is mandatory, you can't do "git restore-files" without any
>>   pathspec.
>>
>> - I just remember -p which is allowed to take no pathspec :( I'll fix
>>   it later.
>>
>> - Two more fancy features (the "git checkout --index" being the
>>   default mode and the backup log for accidental overwrites) are of
>>   course still missing. But they are coming.
>>
>> I did not go replace "detached HEAD" with "unnamed branch" (or "no
>> branch") everywhere because I think a unique term is still good to
>> refer to this concept. Or maybe "no branch" is good enough. I dunno.
>
> I finally tracked down
> https://redfin.engineering/two-commits-that-wrecked-the-user-experience-of-git-f0075b77eab1
> which I'd remembered reading and couldn't find again in these
> discussions. Re-reading it while one may not 100% agree with the
> author's opinion, it's an interesting rabbit hole.
>
> I also didn't know about EasyGit, or that Elijah Newren had written
> it. I haven't seen him chime in on this series, and would be interested
> to see what he thinks about it.
>
> Re the naming question in
> https://public-inbox.org/git/87o9abzv46.fsf@evledraar.gmail.com/ &
> seeing that eg-switch exists, I wonder if just s/switch-branch/switch/
> makes more sense.
>
> Assuming greenfield development (which we definitely don't have), I
> don't like the "restore-files" name, but the alternative that makes
> sense is "checkout". Then this "--from" argument could become "git
> checkout-tree <treeish> -- <pathspec>", and we'd have:
>
>     git switch <branchish>
>     git checkout <pathspec>
>     git checkout-tree <treeish> -- <pathspec>
>
> Or maybe that sucks, anyway what I was going to suggest is *if* others
> think that made sense as a "if we designed git today" endgame whether we
> could have an opt-in setting where you set e.g. core.uiVersion=3 (in
> anticipation of Git 3.0) and you'd get that behavior. There could be
> some other setting where core.uiVersion would use the old behavior (or
> with another setting, only warn) if we weren't connected to a terminal.
>
> I.e. I'm thinking of this as step #2 in a #3 step series. Where this is
> the fully backwards compatible UI improvement, but someone who'd
> e.g. use EasyGit or didn't have backwards compatibility concerns could
> enable step #3 and opt-in to a mode where we'd fixed a bunch of UI warts
> in a backwards-incompatible way.
>
> What would that mode look like? I'd to work on piling that on top of
> this :)

(Digging some more)

There's also more interesting prior art at https://gitless.com/ (CC'd
authors) and 2x research papers linked at the bottom of that page which
were briefly discussed on-list before:
https://public-inbox.org/git/20160930191413.002049b94b3908b15881b77f@domain007.com/

The "gitless" UI has a "gl checkout" which just takes paths as I was
musing about above (and that Redfin post also suggests).

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Dan Fabulich]

Assuming the great day has come to think about this, one thing I'd love to do is to unify the name of the index/stage/cache in command-line parameters and the documentation.

The index/stage/cache should have one canonical name, and the documentation should support that consistently. My taste is to call it the "stage," but references to --index, --keep-index, --no-index, etc. are all over the code, as well as legacy references to "--cached".

* You can 'git rm --cached myfile' but you can't 'git rm --staged myfile' or 'git rm --index myfile'.

* You can 'git diff --no-index' or you can 'git diff --cached' with 'git diff --staged' as a synonym, deprioritized in the documentation ("--staged is a synonym of --cached"). But you can't 'git diff --index' or 'git diff --no-stage' or 'git diff --no-cache'.

* You can 'git stage' but 'git help stage' is only one line long, declaring that it's a synonym for 'git add'. 'add' appears in the 'git help' common commands, but not 'git stage'. There's no built-in 'git unstage', and certainly no appetite for 'git index' or 'git cache'.

* Not to mention all of the plumbing commands: checkout-index, diff-index, index-pack, merge-index, show-index, and update-index.

My understanding based on historical threads is that changes like this would be unwelcome, even just to add synonyms and standardize the documentation around "stage" as the term (leaving the plumbing commands alone), but if documentation+synonym patches are welcome here, I'd be very enthusiastic.

-Dan

> On Nov 29, 2018, at 3:05 PM, Ævar Arnfjörð Bjarmason <avarab@gmail.com> wrote:
> 
> 
> On Thu, Nov 29 2018, Nguyễn Thái Ngọc Duy wrote:
> 
>> v3 sees switch-branch go back to switch-branch (in v2 it was
>> checkout-branch). checkout-files is also renamed restore-files (v1 was
>> restore-paths). Hopefully we won't see another rename.
>> 
>> I'll try to summarize the differences between the new commands and
>> 'git checkout' down here, but you're welcome to just head to 07/14 and
>> read the new man pages.
>> 
>> 'git switch-branch'
>> 
>> - does not "do nothing", you have to either switch branch, create a
>>  new branch, or detach. "git switch-branch" with no arguments is
>>  rejected.
>> 
>> - implicit detaching is rejected. If you need to detach, you need to
>>  give --detach. Or stick to 'git checkout'.
>> 
>> - -b/-B is renamed to -c/-C with long option names
>> 
>> - of course does not accept pathspec
>> 
>> 'git restore-files'
>> 
>> - takes a ref from --from argument, not as a free ref. As a result,
>>  '--' is no longer needed. All non-option arguments are pathspec
>> 
>> - pathspec is mandatory, you can't do "git restore-files" without any
>>  pathspec.
>> 
>> - I just remember -p which is allowed to take no pathspec :( I'll fix
>>  it later.
>> 
>> - Two more fancy features (the "git checkout --index" being the
>>  default mode and the backup log for accidental overwrites) are of
>>  course still missing. But they are coming.
>> 
>> I did not go replace "detached HEAD" with "unnamed branch" (or "no
>> branch") everywhere because I think a unique term is still good to
>> refer to this concept. Or maybe "no branch" is good enough. I dunno.
> 
> I finally tracked down
> https://redfin.engineering/two-commits-that-wrecked-the-user-experience-of-git-f0075b77eab1
> which I'd remembered reading and couldn't find again in these
> discussions. Re-reading it while one may not 100% agree with the
> author's opinion, it's an interesting rabbit hole.
> 
> I also didn't know about EasyGit, or that Elijah Newren had written
> it. I haven't seen him chime in on this series, and would be interested
> to see what he thinks about it.
> 
> Re the naming question in
> https://public-inbox.org/git/87o9abzv46.fsf@evledraar.gmail.com/ &
> seeing that eg-switch exists, I wonder if just s/switch-branch/switch/
> makes more sense.
> 
> Assuming greenfield development (which we definitely don't have), I
> don't like the "restore-files" name, but the alternative that makes
> sense is "checkout". Then this "--from" argument could become "git
> checkout-tree <treeish> -- <pathspec>", and we'd have:
> 
>    git switch <branchish>
>    git checkout <pathspec>
>    git checkout-tree <treeish> -- <pathspec>
> 
> Or maybe that sucks, anyway what I was going to suggest is *if* others
> think that made sense as a "if we designed git today" endgame whether we
> could have an opt-in setting where you set e.g. core.uiVersion=3 (in
> anticipation of Git 3.0) and you'd get that behavior. There could be
> some other setting where core.uiVersion would use the old behavior (or
> with another setting, only warn) if we weren't connected to a terminal.
> 
> I.e. I'm thinking of this as step #2 in a #3 step series. Where this is
> the fully backwards compatible UI improvement, but someone who'd
> e.g. use EasyGit or didn't have backwards compatibility concerns could
> enable step #3 and opt-in to a mode where we'd fixed a bunch of UI warts
> in a backwards-incompatible way.
> 
> What would that mode look like? I'd to work on piling that on top of
> this :)

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Dan Fabulich]

Other thoughts on a global UI rethink:

One of the most common complaints I hear about git is the conceptual difficulty required in undoing changes. https://ohshitgit.com/

> Git is hard: screwing up is easy, and figuring out how to fix your mistakes is fucking impossible. Git documentation has this chicken and egg problem where you can't search for how to get yourself out of a mess, unless you *already know the name of the thing you need to know about* in order to fix your problem.

A significant fraction of the top-voted questions on StackOverflow are about undoing changes. https://stackoverflow.com/questions/tagged/git

What if there were a 'git undo' command that could unify the answers to all of these questions?

git undo stage
git undo rm
git undo edit (checkout files from the stage)

git undo commit (prompt the user whether to revert or reset)
git undo reset
git undo checkout

git undo merge
git undo pull
git undo push (prompt the user whether to force push back to the past or whether to revert pushed commits)
git undo rebase

git undo undo

git undo clean
git undo delete-branch
git undo delete-stash

Some of these would be trivial effort, but a lot of them would require fundamental changes in the way git operates. (You can't undo a clean right now because the files are just destroyed.)

-Dan

> On Nov 29, 2018, at 3:05 PM, Ævar Arnfjörð Bjarmason <avarab@gmail.com> wrote:
> 
> 
> On Thu, Nov 29 2018, Nguyễn Thái Ngọc Duy wrote:
> 
>> v3 sees switch-branch go back to switch-branch (in v2 it was
>> checkout-branch). checkout-files is also renamed restore-files (v1 was
>> restore-paths). Hopefully we won't see another rename.
>> 
>> I'll try to summarize the differences between the new commands and
>> 'git checkout' down here, but you're welcome to just head to 07/14 and
>> read the new man pages.
>> 
>> 'git switch-branch'
>> 
>> - does not "do nothing", you have to either switch branch, create a
>>  new branch, or detach. "git switch-branch" with no arguments is
>>  rejected.
>> 
>> - implicit detaching is rejected. If you need to detach, you need to
>>  give --detach. Or stick to 'git checkout'.
>> 
>> - -b/-B is renamed to -c/-C with long option names
>> 
>> - of course does not accept pathspec
>> 
>> 'git restore-files'
>> 
>> - takes a ref from --from argument, not as a free ref. As a result,
>>  '--' is no longer needed. All non-option arguments are pathspec
>> 
>> - pathspec is mandatory, you can't do "git restore-files" without any
>>  pathspec.
>> 
>> - I just remember -p which is allowed to take no pathspec :( I'll fix
>>  it later.
>> 
>> - Two more fancy features (the "git checkout --index" being the
>>  default mode and the backup log for accidental overwrites) are of
>>  course still missing. But they are coming.
>> 
>> I did not go replace "detached HEAD" with "unnamed branch" (or "no
>> branch") everywhere because I think a unique term is still good to
>> refer to this concept. Or maybe "no branch" is good enough. I dunno.
> 
> I finally tracked down
> https://redfin.engineering/two-commits-that-wrecked-the-user-experience-of-git-f0075b77eab1
> which I'd remembered reading and couldn't find again in these
> discussions. Re-reading it while one may not 100% agree with the
> author's opinion, it's an interesting rabbit hole.
> 
> I also didn't know about EasyGit, or that Elijah Newren had written
> it. I haven't seen him chime in on this series, and would be interested
> to see what he thinks about it.
> 
> Re the naming question in
> https://public-inbox.org/git/87o9abzv46.fsf@evledraar.gmail.com/ &
> seeing that eg-switch exists, I wonder if just s/switch-branch/switch/
> makes more sense.
> 
> Assuming greenfield development (which we definitely don't have), I
> don't like the "restore-files" name, but the alternative that makes
> sense is "checkout". Then this "--from" argument could become "git
> checkout-tree <treeish> -- <pathspec>", and we'd have:
> 
>    git switch <branchish>
>    git checkout <pathspec>
>    git checkout-tree <treeish> -- <pathspec>
> 
> Or maybe that sucks, anyway what I was going to suggest is *if* others
> think that made sense as a "if we designed git today" endgame whether we
> could have an opt-in setting where you set e.g. core.uiVersion=3 (in
> anticipation of Git 3.0) and you'd get that behavior. There could be
> some other setting where core.uiVersion would use the old behavior (or
> with another setting, only warn) if we weren't connected to a terminal.
> 
> I.e. I'm thinking of this as step #2 in a #3 step series. Where this is
> the fully backwards compatible UI improvement, but someone who'd
> e.g. use EasyGit or didn't have backwards compatibility concerns could
> enable step #3 and opt-in to a mode where we'd fixed a bunch of UI warts
> in a backwards-incompatible way.
> 
> What would that mode look like? I'd to work on piling that on top of
> this :)

Re: [PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Junio C Hamano]

Duy Nguyen <pclouds@gmail.com> writes:

> On Wed, Nov 28, 2018 at 9:01 PM Duy Nguyen <pclouds@gmail.com> wrote:
>> should we do
>> something about detached HEAD in this switch-branch command (or
>> whatever its name will be)?
>>
>> This is usually a confusing concept to new users
>
> And it just occurred to me that perhaps we should call this "unnamed
> branch" (at least at high UI level) instead of detached HEAD. It is
> technically not as accurate, but much better to understand.

As I said elsewhere in nearby thread, I agree that "unnamed branch"
is a reasonable way to explain what the state the user is in.  It is
not incorrect per-se that HEAD is detached from anything in refs/ in
such a state, but that is an implementation detail of how the
worktree gets on the unnamed branch (which lasts until the worktree
next gets on a named branch, at which point the unnamed branch
disappears).

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Junio C Hamano]

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> 'git switch-branch'
>
> - implicit detaching is rejected. If you need to detach, you need to
>   give --detach. Or stick to 'git checkout'.

OK.  Is "auto-vivify the named branch based on a remote-tracking"
also rejected, as it is a confusing behaviour that is a too subtle
and implicit, just like the detaching head is, and require --guess
or sticking to 'git checkout'?  I think it should.

> - -b/-B is renamed to -c/-C with long option names

I did not expect that these two are the only options that would be
out of place with the command name split, but presumably you looked
at all options for both of the two new commands to see if they made
sense in the new context?

> 'git restore-files'
>
> - takes a ref from --from argument, not as a free ref. As a result,
>   '--' is no longer needed. All non-option arguments are pathspec

OK.  That does make things easier to teach, as there is no need for
disambiguation.

> - pathspec is mandatory, you can't do "git restore-files" without any
>   pathspec.
>
> - I just remember -p which is allowed to take no pathspec :( I'll fix
>   it later.

Or leave it out of restore-files as a more advanced feature (just
like detaching with HEAD^0 is left out of switch-branch) that the
user can stick to 'git checkout' to use.

> - Two more fancy features (the "git checkout --index" being the
>   default mode and the backup log for accidental overwrites) are of
>   course still missing. But they are coming.

I am unsure about the wisdom of calling it "--index", though.

The "--index" option is "the command can work only on the index, or
only on the working tree files, or on both the index and the working
tree files, and this option tells it to work in the 'both the index
and the working tree files' mode", but when "restore-files" touches
paths, it always modifies both the index and the working tree, so
the "--index" option does not capture the differences well in this
context [*1*].  As I saw this was described as "not using the usual
'overlay' semantics [*2*]", perhaps --overlay/--no-overlay option
that defaults to --no-overlay is easier to explain.

    side note 1.  I think the original mention of "--index" came in
    the context of contrasting "git reset" with "git checkout".
    "git reset (--hard|--mixed) -- <pathspec>" (that does not move
    HEAD), which does not but perhaps should exist, is very much
    like "git checkout -- <pathspec>", and if "reset" were written
    after the "--index/--cached" convention was established, "reset
    --hard" would have called "reset --index" while "reset --mixed"
    would have been "reset --cached" (i.e. only work on the index
    and not on the working tree).  And "reset --index <directory>"
    would have worked by removing paths in <directory> that are not
    in the HEAD and updating paths in <directory> that are in the
    HEAD, i.e. identical to the non overlay behaviour proposed for
    the "git checkout" command.  So calling the non overlay mode
    "--index" makes sense in the context of discussing "git reset",
    and not in the context of "git checkout".

    side note 2.  "git checkout <tree-ish> <pathspec>" grabs entries
    from the <tree-ish> that patch <pathspec> and adds them to the
    index and checks them out to the working tree.  If the original
    index has entries that match <pathspec> but do not appear in
    <tree-ish>, they are left in the result.  That is "overlaying
    what was taken from the <tree-ish> on top of what is in the
    index".

Having said all that, I will not be looking at the series until 2.20
final.  And I hope more people do the same to concentrate on helping
us prevent the last minute glitch slipping in the final release.

Thanks.

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Duy Nguyen]

On Fri, Nov 30, 2018 at 12:05 AM Ævar Arnfjörð Bjarmason
<avarab@gmail.com> wrote:
> Assuming greenfield development (which we definitely don't have), I
> don't like the "restore-files" name, but the alternative that makes
> sense is "checkout". Then this "--from" argument could become "git
> checkout-tree <treeish> -- <pathspec>", and we'd have:
>
>     git switch <branchish>
>     git checkout <pathspec>
>     git checkout-tree <treeish> -- <pathspec>
>
> Or maybe that sucks, anyway what I was going to suggest is *if* others
> think that made sense as a "if we designed git today" endgame whether we
> could have an opt-in setting where you set e.g. core.uiVersion=3 (in
> anticipation of Git 3.0) and you'd get that behavior. There could be
> some other setting where core.uiVersion would use the old behavior (or
> with another setting, only warn) if we weren't connected to a terminal.

core.uiVersion is a big no no to me. I don't want to go to someone's
terminal, type something and have a total surprise because they set
different ui version. If you want a total UI redesign, go with a new
prefix, like "ng" (for new git) or something instead of "git".
-- 
Duy

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Duy Nguyen]

On Fri, Nov 30, 2018 at 3:16 AM Junio C Hamano <gitster@pobox.com> wrote:
>
> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>
> > 'git switch-branch'
> >
> > - implicit detaching is rejected. If you need to detach, you need to
> >   give --detach. Or stick to 'git checkout'.
>
> OK.  Is "auto-vivify the named branch based on a remote-tracking"
> also rejected, as it is a confusing behaviour that is a too subtle
> and implicit, just like the detaching head is, and require --guess
> or sticking to 'git checkout'?  I think it should.

This touches the "remote" concept which I think is another confusing
thing for new people (your "master" is not the same as the server's
"master", aka origin/master) and perhaps this dwim thing helps.
Frankly I don't do dwim much so I don't know if it's that often used.

> > - -b/-B is renamed to -c/-C with long option names
>
> I did not expect that these two are the only options that would be
> out of place with the command name split, but presumably you looked
> at all options for both of the two new commands to see if they made
> sense in the new context?

Yeah (at least the description in struct option[] array)
--
Duy

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Junio C Hamano]

Duy Nguyen <pclouds@gmail.com> writes:

>>
>> OK.  Is "auto-vivify the named branch based on a remote-tracking"
>> also rejected, as it is a confusing behaviour that is a too subtle
>> and implicit, just like the detaching head is, and require --guess
>> or sticking to 'git checkout'?  I think it should.
>
> This touches the "remote" concept which I think is another confusing
> thing for new people (your "master" is not the same as the server's
> "master", aka origin/master) and perhaps this dwim thing helps.
> Frankly I don't do dwim much so I don't know if it's that often used.

I actually think a user who sees a DWIM without understanding what
the user wants to do would perceive magic that sometimes works and
sometimes does not, and some other times it does a random thing that
the user does not even understand what is going on.  And such a
random magic that sometimes works, even if the "sometimes" is "most
of the time", say 85% of the time, would not help user form the
right mental model.

"git checkout master~2" that DWIMs to "git checkout --deatch
master~2", but does totally different thing when "git checkout
master" is given, leaving the user confused "what is so different
between these two?".  Until the user realizes 'master' can serve
both as a branch name and a name for a commit object, while master~2
can only be a name for a commit object and is not a branch name, the
behaviour of the command will stay to be mysterious and DWIMmage
would not help user form the right mental model.  I earlier said
that I agree with your decision to leave the implied form out of
switch-branch for exactly that reason.  

The behaviour falls into the same category as "git checkout frotz"
that DWIMS to "git checkout -b frotz -t remotes/origin/frotz", which
also is mysterious until the user understands your 'master' is unique
and is different from 'master' to everybody else, I would think.

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Junio C Hamano]

Duy Nguyen <pclouds@gmail.com> writes:

> core.uiVersion is a big no no to me. I don't want to go to someone's
> terminal, type something and have a total surprise because they set
> different ui version. If you want a total UI redesign, go with a new
> prefix, like "ng" (for new git) or something instead of "git".

Yup, very good point to keep in mind.

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Duy Nguyen]

On Fri, Nov 30, 2018 at 1:16 AM Dan Fabulich <dan@fabulich.com> wrote:
>
> Other thoughts on a global UI rethink:
>
> One of the most common complaints I hear about git is the conceptual difficulty required in undoing changes. https://ohshitgit.com/
>
> > Git is hard: screwing up is easy, and figuring out how to fix your mistakes is fucking impossible. Git documentation has this chicken and egg problem where you can't search for how to get yourself out of a mess, unless you *already know the name of the thing you need to know about* in order to fix your problem.
>
> A significant fraction of the top-voted questions on StackOverflow are about undoing changes. https://stackoverflow.com/questions/tagged/git
>
> What if there were a 'git undo' command that could unify the answers to all of these questions?
>
> git undo stage
> git undo rm
> git undo edit (checkout files from the stage)
>
> git undo commit (prompt the user whether to revert or reset)
> git undo reset
> git undo checkout
>
> git undo merge
> git undo pull
> git undo push (prompt the user whether to force push back to the past or whether to revert pushed commits)
> git undo rebase
>
> git undo undo
>
> git undo clean
> git undo delete-branch
> git undo delete-stash
>
> Some of these would be trivial effort, but a lot of them would require fundamental changes in the way git operates. (You can't undo a clean right now because the files are just destroyed.)

We're getting there. The biggest problem I have is how this "git undo"
should work, not the changes behind to support it. For example, if I
pulled then did some rebase, what would "git undo pull" do?
-- 
Duy

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Ævar Arnfjörð Bjarmason]

On Fri, Nov 30 2018, Duy Nguyen wrote:

> On Fri, Nov 30, 2018 at 12:05 AM Ævar Arnfjörð Bjarmason
> <avarab@gmail.com> wrote:
>> Assuming greenfield development (which we definitely don't have), I
>> don't like the "restore-files" name, but the alternative that makes
>> sense is "checkout". Then this "--from" argument could become "git
>> checkout-tree <treeish> -- <pathspec>", and we'd have:
>>
>>     git switch <branchish>
>>     git checkout <pathspec>
>>     git checkout-tree <treeish> -- <pathspec>
>>
>> Or maybe that sucks, anyway what I was going to suggest is *if* others
>> think that made sense as a "if we designed git today" endgame whether we
>> could have an opt-in setting where you set e.g. core.uiVersion=3 (in
>> anticipation of Git 3.0) and you'd get that behavior. There could be
>> some other setting where core.uiVersion would use the old behavior (or
>> with another setting, only warn) if we weren't connected to a terminal.
>
> core.uiVersion is a big no no to me. I don't want to go to someone's
> terminal, type something and have a total surprise because they set
> different ui version. If you want a total UI redesign, go with a new
> prefix, like "ng" (for new git) or something instead of "git".

I don't think that's a viable way forward. First, we're not talking
about a total change of the UI. Most the main porcelain will stay the
same. So we'd have a "ng" that's >98% the same UI, and then if we
discover warts in in 10 years we'd like to fix then what do wo do? Ship
"nng" and so forth?

We already have this UI problem with various config variables that
change things. I think we should just solve this general issue by a
combination of:

 a) Accepting that over the long term we will have Git's UI changing,
    sometimes in breaking ways (with sensible phase-in), hopefully for
    the better. E.g. we had this with "git-init" v.s. "git init". This
    is similar, there'd be an error due to ambiguity with a "hint"
    saying use the new thing if you e.g. feed "git checkout" a branch
    name.

 b) For the general problem of "user has exotic config" we should learn
    a "git -Q <cmd>" option similar to Emacs, which is another highly
    customizable piece of software that has a "don't read user config"
    escape hatch.

    That's a bit more complex than for Emacs since we need to actually
    read some config (e.g. remote config from .git/config), and maybe
    opt to keep some stuff like user.*. But there's no reason we can't
    have such a black/whitelist of config & env variables that impact us
    with a switch to get "make it as if nothing was configured" for
    whatever sane version of that we'd come up with.

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Duy Nguyen]

On Fri, Nov 30, 2018 at 12:29 PM Ævar Arnfjörð Bjarmason
<avarab@gmail.com> wrote:
>
>
> On Fri, Nov 30 2018, Duy Nguyen wrote:
>
> > On Fri, Nov 30, 2018 at 12:05 AM Ævar Arnfjörð Bjarmason
> > <avarab@gmail.com> wrote:
> >> Assuming greenfield development (which we definitely don't have), I
> >> don't like the "restore-files" name, but the alternative that makes
> >> sense is "checkout". Then this "--from" argument could become "git
> >> checkout-tree <treeish> -- <pathspec>", and we'd have:
> >>
> >>     git switch <branchish>
> >>     git checkout <pathspec>
> >>     git checkout-tree <treeish> -- <pathspec>
> >>
> >> Or maybe that sucks, anyway what I was going to suggest is *if* others
> >> think that made sense as a "if we designed git today" endgame whether we
> >> could have an opt-in setting where you set e.g. core.uiVersion=3 (in
> >> anticipation of Git 3.0) and you'd get that behavior. There could be
> >> some other setting where core.uiVersion would use the old behavior (or
> >> with another setting, only warn) if we weren't connected to a terminal.
> >
> > core.uiVersion is a big no no to me. I don't want to go to someone's
> > terminal, type something and have a total surprise because they set
> > different ui version. If you want a total UI redesign, go with a new
> > prefix, like "ng" (for new git) or something instead of "git".
>
> I don't think that's a viable way forward. First, we're not talking
> about a total change of the UI. Most the main porcelain will stay the
> same. So we'd have a "ng" that's >98% the same UI, and then if we
> discover warts in in 10 years we'd like to fix then what do wo do? Ship
> "nng" and so forth?

Yes. So think it through and try not to do it often.

> We already have this UI problem with various config variables that
> change things. I think we should just solve this general issue by a
> combination of:
>
>  a) Accepting that over the long term we will have Git's UI changing,
>     sometimes in breaking ways (with sensible phase-in), hopefully for
>     the better. E.g. we had this with "git-init" v.s. "git init". This
>     is similar, there'd be an error due to ambiguity with a "hint"
>     saying use the new thing if you e.g. feed "git checkout" a branch
>     name.

And I hate adding a zillion of config keys to change little things
like this. Now you use this to change bigger behavior. No.
-- 
Duy

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Thomas Gummerer]

On 11/30, Junio C Hamano wrote:
> 
> I am unsure about the wisdom of calling it "--index", though.
> 
> The "--index" option is "the command can work only on the index, or
> only on the working tree files, or on both the index and the working
> tree files, and this option tells it to work in the 'both the index
> and the working tree files' mode", but when "restore-files" touches
> paths, it always modifies both the index and the working tree, so
> the "--index" option does not capture the differences well in this
> context [*1*].  As I saw this was described as "not using the usual
> 'overlay' semantics [*2*]", perhaps --overlay/--no-overlay option
> that defaults to --no-overlay is easier to explain.

Agreed, I think --{no-,}overlay is a much better name for the option,
I'll use that for my patch series (I hope to send that soon after 2.20
is released).

I must admit that I was not aware that the mode is called overlay
mode, before you explained it to me, so I wouldn't expect most users
to know either.  But as it's easy to explain that probably doesn't
matter much.

>     side note 1.  I think the original mention of "--index" came in
>     the context of contrasting "git reset" with "git checkout".
>     "git reset (--hard|--mixed) -- <pathspec>" (that does not move
>     HEAD), which does not but perhaps should exist, is very much
>     like "git checkout -- <pathspec>", and if "reset" were written
>     after the "--index/--cached" convention was established, "reset
>     --hard" would have called "reset --index" while "reset --mixed"
>     would have been "reset --cached" (i.e. only work on the index
>     and not on the working tree).  And "reset --index <directory>"
>     would have worked by removing paths in <directory> that are not
>     in the HEAD and updating paths in <directory> that are in the
>     HEAD, i.e. identical to the non overlay behaviour proposed for
>     the "git checkout" command.  So calling the non overlay mode
>     "--index" makes sense in the context of discussing "git reset",
>     and not in the context of "git checkout".
> 
>     side note 2.  "git checkout <tree-ish> <pathspec>" grabs entries
>     from the <tree-ish> that patch <pathspec> and adds them to the
>     index and checks them out to the working tree.  If the original
>     index has entries that match <pathspec> but do not appear in
>     <tree-ish>, they are left in the result.  That is "overlaying
>     what was taken from the <tree-ish> on top of what is in the
>     index".
> 
> Having said all that, I will not be looking at the series until 2.20
> final.  And I hope more people do the same to concentrate on helping
> us prevent the last minute glitch slipping in the final release.
> 
> Thanks.

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Junio C Hamano]

Thomas Gummerer <t.gummerer@gmail.com> writes:

> Agreed, I think --{no-,}overlay is a much better name for the option,
> I'll use that for my patch series (I hope to send that soon after 2.20
> is released).

OK.

> I must admit that I was not aware that the mode is called overlay
> mode, before you explained it to me, so I wouldn't expect most users
> to know either.  But as it's easy to explain that probably doesn't
> matter much.

I do not think "the mode is called the overlay mode" is so accurate
a description.  I think I've seen the word 'overlay' used to
describe the behaviour in earlier discussions, but because there is
no 'non-overlay' mode exists in versions of 'git checkout' the
end-users have, the users won't even be aware of the possibility
that mode different from what they are used to see could exist, or
that the mode that they are used to see could be called/explained as
the 'overlay' mode.  IOW, we should pick the best phrase to explain
the behaviour we can use when coming up with the command line
option, and that phrase does not have to be 'overlay'---there is no
"using the word 'overlay' for this is good because the users are
familiar with the existing use of the word", simply because there
isn't such familiarilty ;-)

Re: [PATCH/RFC v2 0/7] Introduce new commands switch-branch and checkout-files

[Stefan Beller]

On Thu, Nov 29, 2018 at 7:33 AM Duy Nguyen <pclouds@gmail.com> wrote:
>
> On Wed, Nov 28, 2018 at 9:30 PM Stefan Beller <sbeller@google.com> wrote:
> >
> > On Wed, Nov 28, 2018 at 12:09 PM Duy Nguyen <pclouds@gmail.com> wrote:
> > >
> > > On Wed, Nov 28, 2018 at 9:01 PM Duy Nguyen <pclouds@gmail.com> wrote:
> > > > should we do
> > > > something about detached HEAD in this switch-branch command (or
> > > > whatever its name will be)?
> > > >
> > > > This is usually a confusing concept to new users
> > >
> > > And it just occurred to me that perhaps we should call this "unnamed
> > > branch" (at least at high UI level) instead of detached HEAD. It is
> > > technically not as accurate, but much better to understand.
> >
> > or 'direct' branch?
>
> makes me think, what is an indirect branch?

I drew the term from HEAD pointing to a branch pointing
to a commit, i.e. HEAD indirectly points to a commit, but
in 'direct' branch mode, HEAD contains the commit id.

So indirect branch would work for our current 'real' branches.

When asked out of context of this discussion, I might add
yet another layer of abstraction to make an 'indirect branch',
i.e. HEAD pointing to a symbolic ref that points at a branch
that points to a commit.

The term symref is what we currently use
(Just looked into gitglossary, where we distinguish
symbolic refs from pseudorefs) for hat I would have called
an indirect branch as well.

So maybe we need to measure the level of indirection
("How often do we need to dereference the ref/object to get
a commit oid?") to come to terms in how to describe
the use cases easily.

Here is a fun-one:
  git checkout <symbolic-ref>
  git checkout --detach

Currently the --detach option detaches HEAD from
branch pointing at the object id, i.e. it is the same as
  git checkout <oid>

whereas when we focus on the levels of indirection
it would also be reasonable to have
  git checkout <branch>
as a reasonable alternative, where <branch> is the
branch that is pointed at from the <symbolic ref>.

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Elijah Newren]

On Thu, Nov 29, 2018 at 2:03 PM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
>
> "git checkout" doing too many things is a source of confusion for many
> users (and it even bites old timers sometimes). To rememdy that, the
> command is now split in two: switch-branch and checkout-files. The

"checkout-files" here....(will comment more on this below)

> good old "git checkout" command is still here and will be until all
> (or most of users) are sick of it.
>
> See the new man pages for the final design of these commands. The
> actual implementation though is still pretty much the same as "git
> checkout". Following patches will adjust their behavior to match the
> man pages.
>
> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> ---
>  .gitignore                          |   2 +
>  Documentation/git-checkout.txt      |   5 +
>  Documentation/git-restore-files.txt | 167 ++++++++++++++++
>  Documentation/git-switch-branch.txt | 289 ++++++++++++++++++++++++++++
>  Makefile                            |   2 +
>  builtin.h                           |   2 +
>  builtin/checkout.c                  |  84 ++++++--
>  command-list.txt                    |   2 +
>  git.c                               |   2 +
>  9 files changed, 543 insertions(+), 12 deletions(-)
>  create mode 100644 Documentation/git-restore-files.txt
>  create mode 100644 Documentation/git-switch-branch.txt
>
> diff --git a/.gitignore b/.gitignore
> index 0d77ea5894..c63dcb1427 100644
> --- a/.gitignore
> +++ b/.gitignore
> @@ -143,6 +143,7 @@
>  /git-request-pull
>  /git-rerere
>  /git-reset
> +/git-restore-files

...and "restore-files" here.  Should be consistent with whatever name you pick.

>  /git-rev-list
>  /git-rev-parse
>  /git-revert
> @@ -167,6 +168,7 @@
>  /git-submodule
>  /git-submodule--helper
>  /git-svn
> +/git-switch-branch
>  /git-symbolic-ref
>  /git-tag
>  /git-unpack-file
> diff --git a/Documentation/git-checkout.txt b/Documentation/git-checkout.txt
> index 25887a6087..25ec7f508f 100644
> --- a/Documentation/git-checkout.txt
> +++ b/Documentation/git-checkout.txt
> @@ -406,6 +406,11 @@ $ edit frotz
>  $ git add frotz
>  ------------
>
> +SEE ALSO
> +--------
> +linkgit:git-switch-branch[1]
> +linkgit:git-restore-files[1]
> +
>  GIT
>  ---
>  Part of the linkgit:git[1] suite
> diff --git a/Documentation/git-restore-files.txt b/Documentation/git-restore-files.txt
> new file mode 100644
> index 0000000000..03c1250ad0
> --- /dev/null
> +++ b/Documentation/git-restore-files.txt
> @@ -0,0 +1,167 @@
> +git-restore-files(1)
> +====================
> +
> +NAME
> +----
> +git-restore-files - Restore working tree files
> +
> +SYNOPSIS
> +--------
> +[verse]
> +'git restore-files' [-f|--ours|--theirs|-m|--conflict=<style>] [--from=<tree-ish>] <pathspec>...

Suggesting that you can use both --ours and --from?  Or -f and --from?
 That seems bad; see below for more on this...

> +'git restore-files' [--from=<tree-ish>] <pathspec>...

Isn't this already inferred by the previous line?  Or was the
inclusion of --from on the previous line in error?  Looking at the
git-checkout manpage, it looks like you may have just been copying an
existing weirdness, but it needs to be fixed.  ;-)

> +'git restore-files' (-p|--patch) [--from=<tree-ish>] [<pathspec>...]
> +
> +DESCRIPTION
> +-----------
> +Updates files in the working tree to match the version in the index
> +or the specified tree.
> +
> +'git restore-files' [--from=<tree-ish>] <pathspec>...::

<tree-ish> and <pathspec>?  I understand <commit-ish> and <pathspec>,
or <tree-ish> but have no clue why it'd be okay to specify <tree-ish>
and <pathspec> together.  What does that even mean?

Also, rather than fixing from <tree-ish> to <commit-ish> or <commit>,
perhaps we should just use <revision> here?  (I'm thinking of git
rev-parse's "Specifying revisions", which suggests "revisions" as a
good substitute for "commit-ish" that isn't quite so weird for new
users.)

> +
> +       Overwrite paths in the working tree by replacing with the
> +       contents in the index or in the <tree-ish> (most often a
> +       commit).  When a <tree-ish> is given, the paths that
> +       match the <pathspec> are updated both in the index and in
> +       the working tree.

Is that the default we really want for this command?  Why do we
automatically assume these files are ready for commit?  I understand
that it's what checkout did, but I'd find it more natural to only
affect the working tree by default.  We can give it an option for
affecting the index instead (or perhaps in addition).

> ++
> +The index may contain unmerged entries because of a previous failed merge.
> +By default, if you try to check out such an entry from the index, the
> +checkout operation will fail and nothing will be checked out.
> +Using `-f` will ignore these unmerged entries.  The contents from a
> +specific side of the merge can be checked out of the index by
> +using `--ours` or `--theirs`.  With `-m`, changes made to the working tree
> +file can be discarded to re-create the original conflicted merge result.
> +
> +'git restore-files' (-p|--patch) [--from=<tree-ish>] [<pathspec>...]::
> +       This is similar to the "check out paths to the working tree
> +       from either the index or from a tree-ish" mode described
> +       above, but lets you use the interactive interface to show
> +       the "diff" output and choose which hunks to use in the
> +       result.  See below for the description of `--patch` option.
> +
> +OPTIONS
> +-------
> +-q::
> +--quiet::
> +       Quiet, suppress feedback messages.
> +
> +--[no-]progress::
> +       Progress status is reported on the standard error stream
> +       by default when it is attached to a terminal, unless `--quiet`
> +       is specified. This flag enables progress reporting even if not
> +       attached to a terminal, regardless of `--quiet`.
> +
> +-f::
> +--force::
> +       Do not fail upon unmerged entries; instead, unmerged entries
> +       are ignored.

You just copied this from the checkout manpage, but this is ambiguous
and/or wrong; using git-checkout (since your patch-series doesn't
apply cleanly to either master or next for me):

$ sha1sum counting
c0ed0e34b0fbef4274ef59480e0a0a1cb2776870  counting
$ git checkout counting; echo $?
error: path 'counting' is unmerged
1
$ git checkout -f counting; echo $?
warning: path 'counting' is unmerged
0
$ sha1sum counting
c0ed0e34b0fbef4274ef59480e0a0a1cb2776870  counting

Maybe printing a warning counts as "ignored", though it doesn't seem
like it.  However, even worse is:

$ git checkout -f HEAD~1 counting
$ sha1sum counting
612ca68d0305c821750a452e9d5bf050e915824f  counting

Now the unmerged entry wasn't ignored; it was updated in the working
tree and overwritten in the index.


Perhaps -f and --from should be incompatible and throw an error if
both are specified?  Also does "printed a warning message for"
actually count as "ignored" or should the documentation for this
option be updated?

> +--ours::
> +--theirs::
> +       Check out stage #2 ('ours') or #3 ('theirs') for unmerged
> +       paths.
> ++
> +Note that during `git rebase` and `git pull --rebase`, 'ours' and
> +'theirs' may appear swapped; `--ours` gives the version from the
> +branch the changes are rebased onto, while `--theirs` gives the
> +version from the branch that holds your work that is being rebased.
> ++
> +This is because `rebase` is used in a workflow that treats the
> +history at the remote as the shared canonical one, and treats the
> +work done on the branch you are rebasing as the third-party work to
> +be integrated, and you are temporarily assuming the role of the
> +keeper of the canonical history during the rebase.  As the keeper of
> +the canonical history, you need to view the history from the remote
> +as `ours` (i.e. "our shared canonical history"), while what you did
> +on your side branch as `theirs` (i.e. "one contributor's work on top
> +of it").

Total aside because I'm not sure what you could change here, but man
do I hate this.

> +
> +--ignore-skip-worktree-bits::
> +       In sparse checkout mode, update only entries matched by
> +       <paths> and sparse patterns in
> +       $GIT_DIR/info/sparse-checkout. This option ignores the sparse
> +       patterns and adds back any files in <paths>.

This doesn't make any sense now that you've removed the "`git checkout
-- <paths>` would" from the original.

> +
> +-m::
> +--merge::
> +       When checking out paths from the index, this option lets you
> +       recreate the conflicted merge in the specified paths.
> +
> +--conflict=<style>::
> +       The same as --merge option above, but changes the way the
> +       conflicting hunks are presented, overriding the
> +       merge.conflictStyle configuration variable.  Possible values are
> +       "merge" (default) and "diff3" (in addition to what is shown by
> +       "merge" style, shows the original contents).
> +
> +-p::
> +--patch::
> +       Interactively select hunks in the difference between the
> +       <tree-ish> (or the index, if unspecified) and the working
> +       tree.  The chosen hunks are then applied in reverse to the
> +       working tree (and if a <tree-ish> was specified, the index).
> ++
> +This means that you can use `git restore-files -p` to selectively
> +discard edits from your current working tree. See the ``Interactive
> +Mode'' section of linkgit:git-add[1] to learn how to operate the
> +`--patch` mode.
> +
> +--[no-]recurse-submodules::
> +       Using --recurse-submodules will update the content of all initialized
> +       submodules according to the commit recorded in the superproject. If
> +       local modifications in a submodule would be overwritten the checkout
> +       will fail unless `-f` is used. If nothing (or --no-recurse-submodules)
> +       is used, the work trees of submodules will not be updated.
> +       Just like linkgit:git-submodule[1], this will detach the
> +       submodules HEAD.
> +
> +<tree-ish>::
> +       Tree to checkout from (when paths are given). If not specified,
> +       the index will be used.

Again, I'd really rather use <revision> here.

> +
> +EXAMPLES
> +--------
> +
> +. The following sequence checks out the `master` branch, reverts
> +the `Makefile` to two revisions back, deletes hello.c by
> +mistake, and gets it back from the index.
> ++
> +------------
> +$ git switch-branch master                    <1>
> +$ git restore-files --from master~2 Makefile  <2>
> +$ rm -f hello.c
> +$ git restore-files hello.c                   <3>
> +------------
> ++
> +<1> switch branch
> +<2> take a file out of another commit
> +<3> restore hello.c from the index
> ++

Why is the switch-branch command labelled but not the rm command?  It
made sense in the original checkout manpage to label all checkout
commands, but here since only restore-files is being discussed it
seems the switch-branch should lose its label.

> +If you want to check out _all_ C source files out of the index,
> +you can say
> ++
> +------------
> +$ git restore-files '*.c'
> +------------
> ++
> +Note the quotes around `*.c`.  The file `hello.c` will also be
> +checked out, even though it is no longer in the working tree,
> +because the file globbing is used to match entries in the index
> +(not in the working tree by the shell).

Sounds good.

> ++
> +If you have an unfortunate branch that is named `hello.c`, this
> +step would be confused as an instruction to switch to that branch.
> +You should instead write:
> ++
> +------------
> +$ git restore-files hello.c
> +------------

Isn't the point of this command to allow us to remove paragraphs like
this last one?

> +
> +SEE ALSO
> +--------
> +linkgit:git-checkout[1]
> +
> +GIT
> +---
> +Part of the linkgit:git[1] suite


My single biggest worry about this whole series is that I'm worried
you're perpetuating and even further ingraining one of the biggest
usability problems with checkout: people suggest and use it for
reverting/restoring paths to a previous version, but it doesn't do
that:

git restore-files --from master~10 Documentation/
<edit some non-documentation files>
git add -u
git commit -m "Rationale for changing files including reverting Documentation/"

In particular, now you have a mixture of files in Documentation/ from
master~10 (er, now likely master~11) and HEAD~1; any files and
sub-directories that existed in HEAD~1 still remain and are mixed with
all other files in Documentation/ from the older commit.

You may think this is a special case, but this particular issue
results in some pretty bad surprises.  Also, it was pretty surprising
to me just how difficult it was to implement an svn-like revert in
EasyGit, in large part because of this 'oversight' in git.  git
checkout -- <paths> to me has always been fundamentally wrong, but I
just wasn't sure if I wanted to fight the backward compatibility
battle and suggest changing it.  With a new command, we definitely
shouldn't be reinforcing this error.  (And maybe we should consider
taking the time to fix git checkout too.)


> diff --git a/Documentation/git-switch-branch.txt b/Documentation/git-switch-branch.txt
> new file mode 100644
> index 0000000000..d5bf5cb37d
> --- /dev/null
> +++ b/Documentation/git-switch-branch.txt
> @@ -0,0 +1,289 @@
> +git-switch-branch(1)
> +====================
> +
> +NAME
> +----
> +git-switch-branch - Switch branches
> +
> +SYNOPSIS
> +--------
> +[verse]
> +'git switch-branch' [-q] [-f] [-m] <branch>
> +'git switch-branch' [-q] [-f] [-m] --detach [<commit>]
> +'git switch-branch' [-q] [-f] [-m] [[-b|-B|--orphan] <new_branch>] [<start_point>]

You label the options as -b/-B here, but -c/-C below.  Should be consistent.

> +
> +DESCRIPTION
> +-----------
> +Switch to a specified branch and update files in the working tree to
> +match it.
> +
> +'git switch-branch' <branch>::
> +       To prepare for working on <branch>, switch to it by updating
> +       the index and the files in the working tree. Local
> +       modifications to the files in the working tree are kept, so
> +       that they can be committed to the <branch>.
> ++
> +If <branch> is not found but there does exist a tracking branch in
> +exactly one remote (call it <remote>) with a matching name, treat as
> +equivalent to
> ++
> +------------
> +$ git switch-branch -b <branch> --track <remote>/<branch>
> +------------

If we're making --detach explicit (which I think is good), shouldn't
this also be made explicit?  I think I saw Junio arguing for this in
another thread.

Also, this is another case where you used -b instead of -c.  Finally,
--track wasn't mentioned in the synopsis but it is shown the first
time -b or -c is used?

> ++
> +If the branch exists in multiple remotes and one of them is named by
> +the `checkout.defaultRemote` configuration variable, we'll use that
> +one for the purposes of disambiguation, even if the `<branch>` isn't
> +unique across all remotes. Set it to
> +e.g. `checkout.defaultRemote=origin` to always checkout remote
> +branches from there if `<branch>` is ambiguous but exists on the
> +'origin' remote. See also `checkout.defaultRemote` in
> +linkgit:git-config[1].

So switch-branch will be controlled by checkout.* config variables?
That probably makes the most sense, but it does dilute the advantage
of adding these simpler commands.

Also, the fact that we're trying to make a simpler command makes me
think that removing the auto-vivify behavior from the default and
adding a simple flag which users can pass to request will allow this
part of the documentation to be hidden behind the appropriate flag,
which may make it easier for users to compartmentalize the command and
it's options, enabling them to learn as they go.

> +
> +'git switch-branch' -c|-C <new_branch> [<start_point>]::
> +
> +       Specifying `-c` causes a new branch to be created as if
> +       linkgit:git-branch[1] were called and then switched to. In
> +       this case you can use the `--track` or `--no-track` options,
> +       which will be passed to 'git branch'.  As a convenience,
> +       `--track` without `-c` implies branch creation; see the
> +       description of `--track` below.

Can we get rid of --track/--no-track and just provide a flag (which
takes no arguments) for the user to use?  Problems with --track:
  * it's not even in your synopsis
  * user has to repeat themselves (e.g. 'next' in two places from '-c
next --track origin/next'); this repetition is BOTH laborious AND
error-prone
  * it's rather inconsistent: --track is the default due to
auto-vivify when the user specifies nothing but a branch name that
doesn't exist yet, but when the user realizes the branch doesn't exist
yet and asks to have it created then suddenly tracking is not the
default??


I'm not sure what's best, but here's some food for thought:


   git switch-branch <branch>
switches to <branch>, if it exists.  Error cases:
  * If <branch> isn't actually a branch but a <tag> or
<remote-tracking-branch> or <revision>, error out and suggest using
--detach.
  * If <branch> isn't actually a branch but there is a similarly named
<remote-tracking-branch> (e.g. origin/<branch>), then suggest using
-c.

  git switch-branch -c <branch>
creates <branch> and, if a relevant-remote-tracking branch exists,
base the branch on that revision and set the new branch up to track
it.  Error cases:
  * If <branch> already exists, error out, suggesting -C or using a
non-conflicting name instead.

Other cases:
  * user wants a branch named 'master' that tracks 'origin/next'?  Use
git branch instead, don't support that in switch-branch.
  * user wants a branch named 'master' that doesn't track
'origin/master' despite 'origin/master' existing?  Use git branch
instead; don't support that in switch-branch.

> ++
> +If `-C` is given, <new_branch> is created if it doesn't exist;
> +otherwise, it is reset. This is the transactional equivalent of
> ++
> +------------
> +$ git branch -f <branch> [<start_point>]
> +$ git switch-branch <branch>
> +------------
> ++
> +that is to say, the branch is not reset/created unless "git
> +switch-branch" is successful.

...and when exactly would it fail?  Reading this, it looks like the
only possible error condition was removed due saying we'll reset the
branch if it already exists, so it's rather confusing.

> +
> +'git switch-branch' --detach [<commit>]::
> +
> +       Prepare to work on a unnamed branch on top of <commit> (see
> +       "DETACHED HEAD" section), and updating the index and the files
> +       in the working tree.  Local modifications to the files in the
> +       working tree are kept, so that the resulting working tree will
> +       be the state recorded in the commit plus the local
> +       modifications.
> ++
> +When the <commit> argument is a branch name, the `--detach` option can
> +be used to detach HEAD at the tip of the branch (`git switch-branch
> +<branch>` would check out that branch without detaching HEAD).
> ++
> +Omitting <commit> detaches HEAD at the tip of the current branch.
> +
> +OPTIONS
> +-------
> +-q::
> +--quiet::
> +       Quiet, suppress feedback messages.
> +
> +--[no-]progress::
> +       Progress status is reported on the standard error stream
> +       by default when it is attached to a terminal, unless `--quiet`
> +       is specified. This flag enables progress reporting even if not
> +       attached to a terminal, regardless of `--quiet`.
> +
> +-f::
> +--force::
> +       Proceed even if the index or the working tree differs from
> +       HEAD.  This is used to throw away local changes.

Haven't thought through this thoroughly, but do we really need an
option for that instead of telling users to 'git reset --hard HEAD'
before switching branches if they want their stuff thrown away?

> +-c <new_branch>::
> +--create <new_branch>::
> +       Create a new branch named <new_branch> and start it at
> +       <start_point>; see linkgit:git-branch[1] for details.
> +
> +-C <new_branch>::
> +--force-create <new_branch>::
> +       Creates the branch <new_branch> and start it at <start_point>;
> +       if it already exists, then reset it to <start_point>. This is
> +       equivalent to running "git branch" with "-f"; see
> +       linkgit:git-branch[1] for details.

Makes sense, but let's get the -b/-B vs. -c/-C consistent.

> +
> +-t::
> +--track::
> +       When creating a new branch, set up "upstream" configuration. See
> +       "--track" in linkgit:git-branch[1] for details.
> ++
> +If no `-c` option is given, the name of the new branch will be derived
> +from the remote-tracking branch, by looking at the local part of the
> +refspec configured for the corresponding remote, and then stripping
> +the initial part up to the "*".
> +This would tell us to use "hack" as the local branch when branching
> +off of "origin/hack" (or "remotes/origin/hack", or even
> +"refs/remotes/origin/hack").  If the given name has no slash, or the above
> +guessing results in an empty name, the guessing is aborted.  You can
> +explicitly give a name with `-c` in such a case.
> +
> +--no-track::
> +       Do not set up "upstream" configuration, even if the
> +       branch.autoSetupMerge configuration variable is true.

These two options and the intervening paragraph, while they make sense
to me, seem like the kind of stuff we'd want to throw out -- or at
least rework -- when trying to introduce a new command to simplify.
But I already discussed that above.

> +-l::
> +       Create the new branch's reflog; see linkgit:git-branch[1] for
> +       details.

??  Jettison this.

> +
> +--detach::
> +       Rather than checking out a branch to work on it, check out a
> +       commit for inspection and discardable experiments.
> +       This is the default behavior of "git checkout <commit>" when
> +       <commit> is not a branch name.  See the "DETACHED HEAD" section
> +       below for details.
> +
> +--orphan <new_branch>::
> +       Create a new 'orphan' branch, named <new_branch>, started from
> +       <start_point> and switch to it.  The first commit made on this

What??  started from <start_point>?  The whole point of --orphan is
you have no parent, i.e. no start point.  Also, why does the
explanation reference an argument that wasn't in the immediately
preceding synopsis?

> +       new branch will have no parents and it will be the root of a new
> +       history totally disconnected from all the other branches and
> +       commits.
> ++
> +The index and the working tree are adjusted as if you had previously run
> +"git checkout <start_point>".  This allows you to start a new history
> +that records a set of paths similar to <start_point> by easily running
> +"git commit -a" to make the root commit.
> ++
> +This can be useful when you want to publish the tree from a commit
> +without exposing its full history. You might want to do this to publish
> +an open source branch of a project whose current tree is "clean", but
> +whose full history contains proprietary or otherwise encumbered bits of
> +code.
> ++
> +If you want to start a disconnected history that records a set of paths
> +that is totally different from the one of <start_point>, then you should
> +clear the index and the working tree right after creating the orphan
> +branch by running "git rm -rf ." from the top level of the working tree.
> +Afterwards you will be ready to prepare your new files, repopulating the
> +working tree, by copying them from elsewhere, extracting a tarball, etc.

Ick.  Seems overly complex.  I'd rather that --orphan defaulted to
clearing the index and working tree, and that one would need to pass
HEAD for <start_point> if you wanted to start out with all those other
files.  That would certainly make the explanation a little clearer to
users, and more natural when they start experimenting with it.

However, --orphan is pretty special case.  Do we perhaps want to leave
it out of this new command and only include it in checkout?

> +-m::
> +--merge::
> +       If you have local modifications to one or more files that are
> +       different between the current branch and the branch to which
> +       you are switching, the command refuses to switch branches in
> +       order to preserve your modifications in context.  However,
> +       with this option, a three-way merge between the current
> +       branch, your working tree contents, and the new branch is
> +       done, and you will be on the new branch.
> ++
> +When a merge conflict happens, the index entries for conflicting
> +paths are left unmerged, and you need to resolve the conflicts
> +and mark the resolved paths with `git add` (or `git rm` if the merge
> +should result in deletion of the path).
> +
> +--conflict=<style>::
> +       The same as --merge option above, but changes the way the
> +       conflicting hunks are presented, overriding the
> +       merge.conflictStyle configuration variable.  Possible values are
> +       "merge" (default) and "diff3" (in addition to what is shown by
> +       "merge" style, shows the original contents).
> +
> +--ignore-other-worktrees::
> +       `git switch-branch` refuses when the wanted ref is already
> +       checked out by another worktree. This option makes it check
> +       the ref out anyway. In other words, the ref can be held by
> +       more than one worktree.

seems rather dangerous...is the goal to be an easier-to-use suggestion
for new users while checkout continues to exist, or is this command
meant to handle all branch switching functionality that checkout has?

> +
> +--[no-]recurse-submodules::
> +       Using --recurse-submodules will update the content of all initialized
> +       submodules according to the commit recorded in the superproject. If
> +       local modifications in a submodule would be overwritten the checkout
> +       will fail unless `-f` is used. If nothing (or --no-recurse-submodules)
> +       is used, the work trees of submodules will not be updated.
> +       Just like linkgit:git-submodule[1], this will detach the
> +       submodules HEAD.
> +
> +<branch>::
> +       Branch to checkout; if it refers to a branch (i.e., a name that,
> +       when prepended with "refs/heads/", is a valid ref), then that
> +       branch is checked out. Otherwise, if it refers to a valid
> +       commit, your HEAD becomes "detached" and you are no longer on
> +       any branch (see below for details).

I thought we requiring --detach in order to detach.  Does this
paragraph need updating?  Also, if we require --detach when we'll be
detaching HEAD, then this paragraph gets a LOT simpler.

> ++
> +You can use the `"@{-N}"` syntax to refer to the N-th last
> +branch/commit checked out using "git checkout" operation. You may
> +also specify `-` which is synonymous to `"@{-1}`.
> ++
> +As a special case, you may use `"A...B"` as a shortcut for the
> +merge base of `A` and `B` if there is exactly one merge base. You can
> +leave out at most one of `A` and `B`, in which case it defaults to `HEAD`.

I actually didn't know about the A...B special case for checkout.
Interesting...but I'm starting to wonder if this is too much info for
a "simplified command".

> +
> +<new_branch>::
> +       Name for the new branch.
> +
> +<start_point>::
> +       The name of a commit at which to start the new branch; see
> +       linkgit:git-branch[1] for details. Defaults to HEAD.

Erm, so if <start_point> is given, and there is an associated remote
tracking branch for the given branch name, perhaps we don't set up the
automatic tracking in contrast to what I mentioned above?  Hmm...

> +
> +DETACHED HEAD
> +-------------
> +include::detach-head.txt[]
> +
> +EXAMPLES
> +--------
> +
> +. The following sequence checks out the `master` branch.
> ++
> +------------
> +$ git switch-branch master
> +------------
> ++
> +
> +. After working in the wrong branch, switching to the correct
> +branch would be done using:
> ++
> +------------
> +$ git switch-branch mytopic
> +------------
> ++
> +However, your "wrong" branch and correct "mytopic" branch may
> +differ in files that you have modified locally, in which case
> +the above checkout would fail like this:
> ++
> +------------
> +$ git switch-branch mytopic
> +error: You have local changes to 'frotz'; not switching branches.
> +------------
> ++
> +You can give the `-m` flag to the command, which would try a
> +three-way merge:
> ++
> +------------
> +$ git switch-branch -m mytopic
> +Auto-merging frotz
> +------------
> ++
> +After this three-way merge, the local modifications are _not_
> +registered in your index file, so `git diff` would show you what
> +changes you made since the tip of the new branch.
> +
> +. When a merge conflict happens during switching branches with
> +the `-m` option, you would see something like this:
> ++
> +------------
> +$ git switch-branch -m mytopic
> +Auto-merging frotz
> +ERROR: Merge conflict in frotz
> +fatal: merge program failed
> +------------
> ++
> +At this point, `git diff` shows the changes cleanly merged as in
> +the previous example, as well as the changes in the conflicted
> +files.  Edit and resolve the conflict and mark it resolved with
> +`git add` as usual:
> ++
> +------------
> +$ edit frotz
> +$ git add frotz
> +------------
> +
> +SEE ALSO
> +--------
> +linkgit:git-checkout[1]
> +
> +GIT
> +---
> +Part of the linkgit:git[1] suite
> diff --git a/Makefile b/Makefile
> index 1a44c811aa..f035dbab9e 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -777,9 +777,11 @@ BUILT_INS += git-format-patch$X
>  BUILT_INS += git-fsck-objects$X
>  BUILT_INS += git-init$X
>  BUILT_INS += git-merge-subtree$X
> +BUILT_INS += git-restore-files$X
>  BUILT_INS += git-show$X
>  BUILT_INS += git-stage$X
>  BUILT_INS += git-status$X
> +BUILT_INS += git-switch-branch$X
>  BUILT_INS += git-whatchanged$X
>
>  # what 'all' will build and 'install' will install in gitexecdir,
> diff --git a/builtin.h b/builtin.h
> index 6538932e99..01ed43ea69 100644
> --- a/builtin.h
> +++ b/builtin.h
> @@ -214,6 +214,7 @@ extern int cmd_remote_fd(int argc, const char **argv, const char *prefix);
>  extern int cmd_repack(int argc, const char **argv, const char *prefix);
>  extern int cmd_rerere(int argc, const char **argv, const char *prefix);
>  extern int cmd_reset(int argc, const char **argv, const char *prefix);
> +extern int cmd_restore_files(int argc, const char **argv, const char *prefix);
>  extern int cmd_rev_list(int argc, const char **argv, const char *prefix);
>  extern int cmd_rev_parse(int argc, const char **argv, const char *prefix);
>  extern int cmd_revert(int argc, const char **argv, const char *prefix);
> @@ -227,6 +228,7 @@ extern int cmd_show_index(int argc, const char **argv, const char *prefix);
>  extern int cmd_status(int argc, const char **argv, const char *prefix);
>  extern int cmd_stripspace(int argc, const char **argv, const char *prefix);
>  extern int cmd_submodule__helper(int argc, const char **argv, const char *prefix);
> +extern int cmd_switch_branch(int argc, const char **argv, const char *prefix);
>  extern int cmd_symbolic_ref(int argc, const char **argv, const char *prefix);
>  extern int cmd_tag(int argc, const char **argv, const char *prefix);
>  extern int cmd_tar_tree(int argc, const char **argv, const char *prefix);
> diff --git a/builtin/checkout.c b/builtin/checkout.c
> index 764e1a83a1..7dc0f4d3f3 100644
> --- a/builtin/checkout.c
> +++ b/builtin/checkout.c
> @@ -33,6 +33,16 @@ static const char * const checkout_usage[] = {
>         NULL,
>  };
>
> +static const char * const switch_branch_usage[] = {
> +       N_("git switch-branch [<options>] [<branch>]"),
> +       NULL,
> +};
> +
> +static const char * const restore_files_usage[] = {
> +       N_("git restore-files [<options>] [<branch>] -- <file>..."),
> +       NULL,
> +};
> +
>  struct checkout_opts {
>         int patch_mode;
>         int quiet;
> @@ -1302,31 +1312,23 @@ static struct option *add_checkout_path_options(struct checkout_opts *opts,
>         return newopts;
>  }
>
> -int cmd_checkout(int argc, const char **argv, const char *prefix)
> +static int checkout_main(int argc, const char **argv, const char *prefix,
> +                        struct checkout_opts *opts, struct option *options,
> +                        const char * const usagestr[])
>  {
> -       struct checkout_opts real_opts;
> -       struct checkout_opts *opts = &real_opts;
>         struct branch_info new_branch_info;
>         int dwim_remotes_matched = 0;
> -       struct option *options = NULL;
>
> -       memset(opts, 0, sizeof(*opts));
>         memset(&new_branch_info, 0, sizeof(new_branch_info));
>         opts->overwrite_ignore = 1;
>         opts->prefix = prefix;
>         opts->show_progress = -1;
> -       opts->dwim_new_local_branch = 1;
>
>         git_config(git_checkout_config, opts);
>
>         opts->track = BRANCH_TRACK_UNSPECIFIED;
>
> -       options = parse_options_dup(options);
> -       options = add_common_options(opts, options);
> -       options = add_switch_branch_options(opts, options);
> -       options = add_checkout_path_options(opts, options);
> -
> -       argc = parse_options(argc, argv, prefix, options, checkout_usage,
> +       argc = parse_options(argc, argv, prefix, options, usagestr,
>                              PARSE_OPT_KEEP_DASHDASH);
>
>         if (opts->show_progress < 0) {
> @@ -1455,3 +1457,61 @@ int cmd_checkout(int argc, const char **argv, const char *prefix)
>                 return checkout_branch(opts, &new_branch_info);
>         }
>  }
> +
> +int cmd_checkout(int argc, const char **argv, const char *prefix)
> +{
> +       struct checkout_opts opts;
> +       struct option *options = NULL;
> +       int ret;
> +
> +       memset(&opts, 0, sizeof(opts));
> +       opts.dwim_new_local_branch = 1;
> +
> +       options = parse_options_dup(options);
> +       options = add_common_options(&opts, options);
> +       options = add_switch_branch_options(&opts, options);
> +       options = add_checkout_path_options(&opts, options);
> +
> +       ret = checkout_main(argc, argv, prefix, &opts,
> +                           options, checkout_usage);
> +       FREE_AND_NULL(options);
> +       return ret;
> +}
> +
> +int cmd_switch_branch(int argc, const char **argv, const char *prefix)
> +{
> +       struct checkout_opts opts;
> +       struct option *options = NULL;
> +       int ret;
> +
> +       memset(&opts, 0, sizeof(opts));
> +       opts.dwim_new_local_branch = 1;
> +
> +       options = parse_options_dup(options);
> +       options = add_common_options(&opts, options);
> +       options = add_switch_branch_options(&opts, options);
> +
> +       ret = checkout_main(argc, argv, prefix, &opts,
> +                           options, switch_branch_usage);
> +       FREE_AND_NULL(options);
> +       return ret;
> +}
> +
> +int cmd_restore_files(int argc, const char **argv, const char *prefix)
> +{
> +       struct checkout_opts opts;
> +       struct option *options = NULL;
> +       int ret;
> +
> +       memset(&opts, 0, sizeof(opts));
> +       opts.dwim_new_local_branch = 1;
> +
> +       options = parse_options_dup(options);
> +       options = add_common_options(&opts, options);
> +       options = add_checkout_path_options(&opts, options);
> +
> +       ret = checkout_main(argc, argv, prefix, &opts,
> +                           options, restore_files_usage);
> +       FREE_AND_NULL(options);
> +       return ret;
> +}
> diff --git a/command-list.txt b/command-list.txt
> index 3a9af104b5..4638802754 100644
> --- a/command-list.txt
> +++ b/command-list.txt
> @@ -151,6 +151,7 @@ git-replace                             ancillarymanipulators           complete
>  git-request-pull                        foreignscminterface             complete
>  git-rerere                              ancillaryinterrogators
>  git-reset                               mainporcelain           worktree
> +git-restore-files                       mainporcelain           worktree
>  git-revert                              mainporcelain
>  git-rev-list                            plumbinginterrogators
>  git-rev-parse                           plumbinginterrogators
> @@ -171,6 +172,7 @@ git-status                              mainporcelain           info
>  git-stripspace                          purehelpers
>  git-submodule                           mainporcelain
>  git-svn                                 foreignscminterface
> +git-switch-branch                       mainporcelain           history
>  git-symbolic-ref                        plumbingmanipulators
>  git-tag                                 mainporcelain           history
>  git-unpack-file                         plumbinginterrogators
> diff --git a/git.c b/git.c
> index 2f604a41ea..a2be6c3eb5 100644
> --- a/git.c
> +++ b/git.c
> @@ -542,6 +542,7 @@ static struct cmd_struct commands[] = {
>         { "replace", cmd_replace, RUN_SETUP },
>         { "rerere", cmd_rerere, RUN_SETUP },
>         { "reset", cmd_reset, RUN_SETUP },
> +       { "restore-files", cmd_restore_files, RUN_SETUP | NEED_WORK_TREE },
>         { "rev-list", cmd_rev_list, RUN_SETUP | NO_PARSEOPT },
>         { "rev-parse", cmd_rev_parse, NO_PARSEOPT },
>         { "revert", cmd_revert, RUN_SETUP | NEED_WORK_TREE },
> @@ -557,6 +558,7 @@ static struct cmd_struct commands[] = {
>         { "status", cmd_status, RUN_SETUP | NEED_WORK_TREE },
>         { "stripspace", cmd_stripspace },
>         { "submodule--helper", cmd_submodule__helper, RUN_SETUP | SUPPORT_SUPER_PREFIX | NO_PARSEOPT },
> +       { "switch-branch", cmd_switch_branch, RUN_SETUP | NEED_WORK_TREE },
>         { "symbolic-ref", cmd_symbolic_ref, RUN_SETUP },
>         { "tag", cmd_tag, RUN_SETUP | DELAY_PAGER_CONFIG },
>         { "unpack-file", cmd_unpack_file, RUN_SETUP | NO_PARSEOPT },
> --
> 2.20.0.rc1.380.g3eb999425c.dirty
>

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Elijah Newren]

On Thu, Nov 29, 2018 at 2:01 PM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
>
> v3 sees switch-branch go back to switch-branch (in v2 it was
> checkout-branch). checkout-files is also renamed restore-files (v1 was
> restore-paths). Hopefully we won't see another rename.

I started reading through the patches.  I also tried to apply them
locally, but they had conflicts or missing base file version on both
master and next.  What version did you base it on?

I stopped at 07/14, and dropped my comments all there.  I didn't read
any further yet, and may wait for your post-2.20 reroll.

> I'll try to summarize the differences between the new commands and
> 'git checkout' down here, but you're welcome to just head to 07/14 and
> read the new man pages.
>
> 'git switch-branch'
>
> - does not "do nothing", you have to either switch branch, create a
>   new branch, or detach. "git switch-branch" with no arguments is
>   rejected.
>
> - implicit detaching is rejected. If you need to detach, you need to
>   give --detach. Or stick to 'git checkout'.
>
> - -b/-B is renamed to -c/-C with long option names
>
> - of course does not accept pathspec
>
> 'git restore-files'
>
> - takes a ref from --from argument, not as a free ref. As a result,
>   '--' is no longer needed. All non-option arguments are pathspec
>
> - pathspec is mandatory, you can't do "git restore-files" without any
>   pathspec.
>
> - I just remember -p which is allowed to take no pathspec :( I'll fix
>   it later.

This all looks good.  I commented elsewhere but please remember that
pathspec implies directories as a possibility and we really need to
fix the broken behavior of checkout when given a directory.

> - Two more fancy features (the "git checkout --index" being the
>   default mode and the backup log for accidental overwrites) are of
>   course still missing. But they are coming.
>
> I did not go replace "detached HEAD" with "unnamed branch" (or "no
> branch") everywhere because I think a unique term is still good to
> refer to this concept. Or maybe "no branch" is good enough. I dunno.

I personally like "unnamed branch", but "no branch" would still be
better than "detached HEAD".

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Junio C Hamano]

Elijah Newren <newren@gmail.com> writes:

>> +Updates files in the working tree to match the version in the index
>> +or the specified tree.
>> +
>> +'git restore-files' [--from=<tree-ish>] <pathspec>...::
>
> <tree-ish> and <pathspec>?  I understand <commit-ish> and <pathspec>,
> or <tree-ish> but have no clue why it'd be okay to specify <tree-ish>
> and <pathspec> together.  What does that even mean?

I have this tree object v2.6.11-tree that is not enclosed in a
commit object.  I want to take the top-level Makefile out of that
tree, stuff it in the index and overwrite the working tree file.

	$ git checkout v2.6.11-tree Makefile
	$ git restore-files --from=v2.6.11-tree Makefile

>> +       Overwrite paths in the working tree by replacing with the
>> +       contents in the index or in the <tree-ish> (most often a
>> +       commit).  When a <tree-ish> is given, the paths that
>> +       match the <pathspec> are updated both in the index and in
>> +       the working tree.
>
> Is that the default we really want for this command?  Why do we
> automatically assume these files are ready for commit?  I understand
> that it's what checkout did, but I'd find it more natural to only
> affect the working tree by default.  We can give it an option for
> affecting the index instead (or perhaps in addition).

Oooah.  Now this is getting juicy.  

I do think supporting "--index" (which would make it more in line
with what Duy wrote), with optionally "--cached" as well, and making
the "working tree only" mode as default may not be a bad idea.  I am
offhand not sure how the "working tree only" mode (similar to the
default mode of "git apply" that mimics the way "patch -p1" works)
should interact with the non-overlay mode of the command, but other
than that, I tend to agree with the idea that restore-files is only
a part of making the contents into committable shape, not exactly
ready for it yet.

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Duy Nguyen]

Thanks for all the comments! There are still some I haven't replied
(either I'll agree and do it anyway, or I'll need some more time to
digest)

On Tue, Dec 4, 2018 at 1:45 AM Elijah Newren <newren@gmail.com> wrote:
> > +'git restore-files' [--from=<tree-ish>] <pathspec>...
>
> Isn't this already inferred by the previous line?  Or was the
> inclusion of --from on the previous line in error?  Looking at the
> git-checkout manpage, it looks like you may have just been copying an
> existing weirdness, but it needs to be fixed.  ;-)

Hehe.

> > +'git restore-files' (-p|--patch) [--from=<tree-ish>] [<pathspec>...]
> > +
> > +DESCRIPTION
> > +-----------
> > +Updates files in the working tree to match the version in the index
> > +or the specified tree.
> > +
> > +'git restore-files' [--from=<tree-ish>] <pathspec>...::
>
> <tree-ish> and <pathspec>?  I understand <commit-ish> and <pathspec>,
> or <tree-ish> but have no clue why it'd be okay to specify <tree-ish>
> and <pathspec> together.  What does that even mean?
>
> Also, rather than fixing from <tree-ish> to <commit-ish> or <commit>,
> perhaps we should just use <revision> here?  (I'm thinking of git
> rev-parse's "Specifying revisions", which suggests "revisions" as a
> good substitute for "commit-ish" that isn't quite so weird for new
> users.)

tree-ish is technically more accurate. But I'm ok with just
<revision>. If you give it a blob oid then you should get a nice
explanation what you're doing wrong anyway.


> > +       Overwrite paths in the working tree by replacing with the
> > +       contents in the index or in the <tree-ish> (most often a
> > +       commit).  When a <tree-ish> is given, the paths that
> > +       match the <pathspec> are updated both in the index and in
> > +       the working tree.
>
> Is that the default we really want for this command?  Why do we
> automatically assume these files are ready for commit?  I understand
> that it's what checkout did, but I'd find it more natural to only
> affect the working tree by default.  We can give it an option for
> affecting the index instead (or perhaps in addition).

Yeah, that behavior of updating the index always bothers me when I use
it but I seemed to forget when working on this.

> > +--ours::
> > +--theirs::
> > +       Check out stage #2 ('ours') or #3 ('theirs') for unmerged
> > +       paths.
> > ++
> > +Note that during `git rebase` and `git pull --rebase`, 'ours' and
> > +'theirs' may appear swapped; `--ours` gives the version from the
> > +branch the changes are rebased onto, while `--theirs` gives the
> > +version from the branch that holds your work that is being rebased.
> > ++
> > +This is because `rebase` is used in a workflow that treats the
> > +history at the remote as the shared canonical one, and treats the
> > +work done on the branch you are rebasing as the third-party work to
> > +be integrated, and you are temporarily assuming the role of the
> > +keeper of the canonical history during the rebase.  As the keeper of
> > +the canonical history, you need to view the history from the remote
> > +as `ours` (i.e. "our shared canonical history"), while what you did
> > +on your side branch as `theirs` (i.e. "one contributor's work on top
> > +of it").
>
> Total aside because I'm not sure what you could change here, but man
> do I hate this.

Uh it's actually documented? I'm always confused by this too. --ours
and --theirs at this point are pretty much tied to stage 2 and 3.
Nothing I can do about it. But if you could come up with some other
option names, then we could make "new ours" to be stage 3 during
rebase, for example.

> > +Part of the linkgit:git[1] suite
>
>
> My single biggest worry about this whole series is that I'm worried
> you're perpetuating and even further ingraining one of the biggest
> usability problems with checkout: people suggest and use it for
> reverting/restoring paths to a previous version, but it doesn't do
> that:
>
> git restore-files --from master~10 Documentation/
> <edit some non-documentation files>
> git add -u
> git commit -m "Rationale for changing files including reverting Documentation/"
>
> In particular, now you have a mixture of files in Documentation/ from
> master~10 (er, now likely master~11) and HEAD~1; any files and
> sub-directories that existed in HEAD~1 still remain and are mixed with
> all other files in Documentation/ from the older commit.
>
> You may think this is a special case, but this particular issue
> results in some pretty bad surprises.  Also, it was pretty surprising
> to me just how difficult it was to implement an svn-like revert in
> EasyGit, in large part because of this 'oversight' in git.  git
> checkout -- <paths> to me has always been fundamentally wrong, but I
> just wasn't sure if I wanted to fight the backward compatibility
> battle and suggest changing it.  With a new command, we definitely
> shouldn't be reinforcing this error.  (And maybe we should consider
> taking the time to fix git checkout too.)

What would be the right behavior for

 git restore-files --from=master~10 Documentation/

then? Consider it an error? I often use "git checkout HEAD" and "git
checkout HEAD^" (usually with -p) but not very far back like
master~10.

> > +If the branch exists in multiple remotes and one of them is named by
> > +the `checkout.defaultRemote` configuration variable, we'll use that
> > +one for the purposes of disambiguation, even if the `<branch>` isn't
> > +unique across all remotes. Set it to
> > +e.g. `checkout.defaultRemote=origin` to always checkout remote
> > +branches from there if `<branch>` is ambiguous but exists on the
> > +'origin' remote. See also `checkout.defaultRemote` in
> > +linkgit:git-config[1].
>
> So switch-branch will be controlled by checkout.* config variables?
> That probably makes the most sense, but it does dilute the advantage
> of adding these simpler commands.
>
> Also, the fact that we're trying to make a simpler command makes me
> think that removing the auto-vivify behavior from the default and
> adding a simple flag which users can pass to request will allow this
> part of the documentation to be hidden behind the appropriate flag,
> which may make it easier for users to compartmentalize the command and
> it's options, enabling them to learn as they go.

Sounds good. I don't know a good name for this new option though so
unless anybody comes up with some suggestion, I'll just disable
checkout.defaultRemote in switch-branch. If it comes back as a new
option, it can always be added later.

> > +'git switch-branch' -c|-C <new_branch> [<start_point>]::
> > +
> > +       Specifying `-c` causes a new branch to be created as if
> > +       linkgit:git-branch[1] were called and then switched to. In
> > +       this case you can use the `--track` or `--no-track` options,
> > +       which will be passed to 'git branch'.  As a convenience,
> > +       `--track` without `-c` implies branch creation; see the
> > +       description of `--track` below.
>
> Can we get rid of --track/--no-track and just provide a flag (which
> takes no arguments) for the user to use?  Problems with --track:
>   * it's not even in your synopsis
>   * user has to repeat themselves (e.g. 'next' in two places from '-c
> next --track origin/next'); this repetition is BOTH laborious AND
> error-prone
>   * it's rather inconsistent: --track is the default due to
> auto-vivify when the user specifies nothing but a branch name that
> doesn't exist yet, but when the user realizes the branch doesn't exist
> yet and asks to have it created then suddenly tracking is not the
> default??

I don't think --track is default anymore (maybe I haven't updated the
man page correctly). The dwim behavior is only activated in
switch-branch when you specify --guess to reduce the amount of magic
we throw at the user. With that in mind, do we still hide
--track/--no-track from switch-branch?

> I'm not sure what's best, but here's some food for thought:
>
>
>    git switch-branch <branch>
> switches to <branch>, if it exists.  Error cases:
>   * If <branch> isn't actually a branch but a <tag> or
> <remote-tracking-branch> or <revision>, error out and suggest using
> --detach.
>   * If <branch> isn't actually a branch but there is a similarly named
> <remote-tracking-branch> (e.g. origin/<branch>), then suggest using
> -c.

I would make these advice so I can hide them. Or if I manage to make
all these hints one line then I'll make it unconditional.

>   git switch-branch -c <branch>
> creates <branch> and, if a relevant-remote-tracking branch exists,
> base the branch on that revision and set the new branch up to track

Hmm.. this is a bit magical and could be surprising. If I create (and
switch to) a new branch foo, I don't necessarily mean tracking
origin/foo (I may not even think about origin/foo when I type the
command). So tentatively no.

> > +If `-C` is given, <new_branch> is created if it doesn't exist;
> > +otherwise, it is reset. This is the transactional equivalent of
> > ++
> > +------------
> > +$ git branch -f <branch> [<start_point>]
> > +$ git switch-branch <branch>
> > +------------
> > ++
> > +that is to say, the branch is not reset/created unless "git
> > +switch-branch" is successful.
>
> ...and when exactly would it fail?  Reading this, it looks like the
> only possible error condition was removed due saying we'll reset the
> branch if it already exists, so it's rather confusing.

Yeah probably just scrape it. The atomic nature is not worth highlighting.


> > +'git switch-branch' --detach [<commit>]::
> > +
> > +       Prepare to work on a unnamed branch on top of <commit> (see
> > +       "DETACHED HEAD" section), and updating the index and the files
> > +       in the working tree.  Local modifications to the files in the
> > +       working tree are kept, so that the resulting working tree will
> > +       be the state recorded in the commit plus the local
> > +       modifications.
> > ++
> > +When the <commit> argument is a branch name, the `--detach` option can
> > +be used to detach HEAD at the tip of the branch (`git switch-branch
> > +<branch>` would check out that branch without detaching HEAD).
> > ++
> > +Omitting <commit> detaches HEAD at the tip of the current branch.
> > +
> > +OPTIONS
> > +-------
> > +-q::
> > +--quiet::
> > +       Quiet, suppress feedback messages.
> > +
> > +--[no-]progress::
> > +       Progress status is reported on the standard error stream
> > +       by default when it is attached to a terminal, unless `--quiet`
> > +       is specified. This flag enables progress reporting even if not
> > +       attached to a terminal, regardless of `--quiet`.
> > +
> > +-f::
> > +--force::
> > +       Proceed even if the index or the working tree differs from
> > +       HEAD.  This is used to throw away local changes.
>
> Haven't thought through this thoroughly, but do we really need an
> option for that instead of telling users to 'git reset --hard HEAD'
> before switching branches if they want their stuff thrown away?

For me it's just a bit more convenient. Hit an error when switching
branch? Recall the command from bash history, stick -f in it and run.
Elsewhere I think both Junio and Thomas (or maybe only Junio) suggests
moving the "git reset" functionality without moving HEAD to one of
these commands, which goes the opposite direction...

> > +-c <new_branch>::
> > +--create <new_branch>::
> > +       Create a new branch named <new_branch> and start it at
> > +       <start_point>; see linkgit:git-branch[1] for details.
> > +
> > +-C <new_branch>::
> > +--force-create <new_branch>::
> > +       Creates the branch <new_branch> and start it at <start_point>;
> > +       if it already exists, then reset it to <start_point>. This is
> > +       equivalent to running "git branch" with "-f"; see
> > +       linkgit:git-branch[1] for details.
>
> Makes sense, but let's get the -b/-B vs. -c/-C consistent.

Another option I'm considering is -n/-N (for _new_ branch). Maybe
-c/-C is good enough.

> > +-l::
> > +       Create the new branch's reflog; see linkgit:git-branch[1] for
> > +       details.
>
> ??  Jettison this.

Yep. It looks weird to me too. reflog is just behind the scene these
days. Nobody need to explicitly ask for reflog anymore.

> > +--orphan <new_branch>::
> > +       Create a new 'orphan' branch, named <new_branch>, started from
> > +       <start_point> and switch to it.  The first commit made on this
>
> What??  started from <start_point>?  The whole point of --orphan is
> you have no parent, i.e. no start point.  Also, why does the
> explanation reference an argument that wasn't in the immediately
> preceding synopsis?

I guess bad phrasing. It should be "switch to <start_point> first,
then prepare the worktree so that the first commit will have no
parent". Or something along that line.

You should really review git-checkout.txt btw ;-)

> > +       new branch will have no parents and it will be the root of a new
> > +       history totally disconnected from all the other branches and
> > +       commits.
> > ++
> > +The index and the working tree are adjusted as if you had previously run
> > +"git checkout <start_point>".  This allows you to start a new history
> > +that records a set of paths similar to <start_point> by easily running
> > +"git commit -a" to make the root commit.
> > ++
> > +This can be useful when you want to publish the tree from a commit
> > +without exposing its full history. You might want to do this to publish
> > +an open source branch of a project whose current tree is "clean", but
> > +whose full history contains proprietary or otherwise encumbered bits of
> > +code.
> > ++
> > +If you want to start a disconnected history that records a set of paths
> > +that is totally different from the one of <start_point>, then you should
> > +clear the index and the working tree right after creating the orphan
> > +branch by running "git rm -rf ." from the top level of the working tree.
> > +Afterwards you will be ready to prepare your new files, repopulating the
> > +working tree, by copying them from elsewhere, extracting a tarball, etc.
>
> Ick.  Seems overly complex.  I'd rather that --orphan defaulted to
> clearing the index and working tree, and that one would need to pass
> HEAD for <start_point> if you wanted to start out with all those other
> files.  That would certainly make the explanation a little clearer to
> users, and more natural when they start experimenting with it.
>
> However, --orphan is pretty special case.  Do we perhaps want to leave
> it out of this new command and only include it in checkout?

I started this by simply splitting git-checkout in two commands that,
combined, can do everything git-checkout can. Then suggestions to have
better default came in and I think we started to drift further to
_removing_ options and falling back to git-checkout.

I think we could still keep "complicated" options as long as they are
clearly described and don't surprise users until they figure them out.
That way I don't have to go back to git-checkout and deal with all the
ambiguation it creates.

> > +-m::
> > +--merge::
> > +       If you have local modifications to one or more files that are
> > +       different between the current branch and the branch to which
> > +       you are switching, the command refuses to switch branches in
> > +       order to preserve your modifications in context.  However,
> > +       with this option, a three-way merge between the current
> > +       branch, your working tree contents, and the new branch is
> > +       done, and you will be on the new branch.
> > ++
> > +When a merge conflict happens, the index entries for conflicting
> > +paths are left unmerged, and you need to resolve the conflicts
> > +and mark the resolved paths with `git add` (or `git rm` if the merge
> > +should result in deletion of the path).
> > +
> > +--conflict=<style>::
> > +       The same as --merge option above, but changes the way the
> > +       conflicting hunks are presented, overriding the
> > +       merge.conflictStyle configuration variable.  Possible values are
> > +       "merge" (default) and "diff3" (in addition to what is shown by
> > +       "merge" style, shows the original contents).
> > +
> > +--ignore-other-worktrees::
> > +       `git switch-branch` refuses when the wanted ref is already
> > +       checked out by another worktree. This option makes it check
> > +       the ref out anyway. In other words, the ref can be held by
> > +       more than one worktree.
>
> seems rather dangerous...is the goal to be an easier-to-use suggestion
> for new users while checkout continues to exist, or is this command
> meant to handle all branch switching functionality that checkout has?

As explained above. I'm still thinking the latter, but with fewer
surprises and confusion. Though I guess I could be convinced to go
with the former (the problem with the former is, even as a
no-longer-new user, I still find git-checkout not that pleasant to use
and want a better replacement)

> > +<branch>::
> > +       Branch to checkout; if it refers to a branch (i.e., a name that,
> > +       when prepended with "refs/heads/", is a valid ref), then that
> > +       branch is checked out. Otherwise, if it refers to a valid
> > +       commit, your HEAD becomes "detached" and you are no longer on
> > +       any branch (see below for details).
>
> I thought we requiring --detach in order to detach.  Does this
> paragraph need updating?  Also, if we require --detach when we'll be
> detaching HEAD, then this paragraph gets a LOT simpler.

Yep. I really need to read through the document and update all of it.

> > +You can use the `"@{-N}"` syntax to refer to the N-th last
> > +branch/commit checked out using "git checkout" operation. You may
> > +also specify `-` which is synonymous to `"@{-1}`.
> > ++
> > +As a special case, you may use `"A...B"` as a shortcut for the
> > +merge base of `A` and `B` if there is exactly one merge base. You can
> > +leave out at most one of `A` and `B`, in which case it defaults to `HEAD`.
>
> I actually didn't know about the A...B special case for checkout.
> Interesting...but I'm starting to wonder if this is too much info for
> a "simplified command".

I could just hint about A...B and send the user to git-checkout.txt if
they need to know more. They can learn about git-checkout that way
too.
-- 
Duy

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Duy Nguyen]

On Tue, Dec 4, 2018 at 2:29 AM Elijah Newren <newren@gmail.com> wrote:
>
> On Thu, Nov 29, 2018 at 2:01 PM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
> >
> > v3 sees switch-branch go back to switch-branch (in v2 it was
> > checkout-branch). checkout-files is also renamed restore-files (v1 was
> > restore-paths). Hopefully we won't see another rename.
>
> I started reading through the patches.  I also tried to apply them
> locally, but they had conflicts or missing base file version on both
> master and next.  What version did you base it on?

I think nd/checkout-dwim-fix because of a non-trivial conflict there
(but I don't remember when I noticed it and rebased on that). Anyway
you can get the whole series at

https://gitlab.com/pclouds/git/tree/switch-branch-and-checkout-files

It fixes some of your comments already, a couple of bug fixes here and
there and in a good-enough shape that I start actually using it.

> > - Two more fancy features (the "git checkout --index" being the
> >   default mode and the backup log for accidental overwrites) are of
> >   course still missing. But they are coming.
> >
> > I did not go replace "detached HEAD" with "unnamed branch" (or "no
> > branch") everywhere because I think a unique term is still good to
> > refer to this concept. Or maybe "no branch" is good enough. I dunno.
>
> I personally like "unnamed branch", but "no branch" would still be
> better than "detached HEAD".

Haven't really worked on killing the term "detached HEAD" yet. But I
noticed the other day that git-branch reports

* (HEAD detached from 703266f6e4)

and I didn't know how to rephrase that. I guess "unnamed branch from
703266f6e4" is probably good enough but my old-timer brain screams no.
-- 
Duy

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Elijah Newren]

On Tue, Dec 4, 2018 at 8:22 AM Duy Nguyen <pclouds@gmail.com> wrote:
>
> Thanks for all the comments! There are still some I haven't replied
> (either I'll agree and do it anyway, or I'll need some more time to
> digest)
>
> On Tue, Dec 4, 2018 at 1:45 AM Elijah Newren <newren@gmail.com> wrote:
> > > +'git restore-files' [--from=<tree-ish>] <pathspec>...
> >
> > Isn't this already inferred by the previous line?  Or was the
> > inclusion of --from on the previous line in error?  Looking at the
> > git-checkout manpage, it looks like you may have just been copying an
> > existing weirdness, but it needs to be fixed.  ;-)
>
> Hehe.
>
> > > +'git restore-files' (-p|--patch) [--from=<tree-ish>] [<pathspec>...]
> > > +
> > > +DESCRIPTION
> > > +-----------
> > > +Updates files in the working tree to match the version in the index
> > > +or the specified tree.
> > > +
> > > +'git restore-files' [--from=<tree-ish>] <pathspec>...::
> >
> > <tree-ish> and <pathspec>?  I understand <commit-ish> and <pathspec>,
> > or <tree-ish> but have no clue why it'd be okay to specify <tree-ish>
> > and <pathspec> together.  What does that even mean?
> >
> > Also, rather than fixing from <tree-ish> to <commit-ish> or <commit>,
> > perhaps we should just use <revision> here?  (I'm thinking of git
> > rev-parse's "Specifying revisions", which suggests "revisions" as a
> > good substitute for "commit-ish" that isn't quite so weird for new
> > users.)
>
> tree-ish is technically more accurate. But I'm ok with just
> <revision>. If you give it a blob oid then you should get a nice
> explanation what you're doing wrong anyway.

Documenting as <revision> but having it be more general under the hood
and actually accept <tree-ish> sounds good to me.  I just think the
pain of trying to explain <tree-ish> is too much of a hurdle for
users, especially as I expect it to be very unlikely that users will
ever take advantage of it.

> > > +       Overwrite paths in the working tree by replacing with the
> > > +       contents in the index or in the <tree-ish> (most often a
> > > +       commit).  When a <tree-ish> is given, the paths that
> > > +       match the <pathspec> are updated both in the index and in
> > > +       the working tree.
> >
> > Is that the default we really want for this command?  Why do we
> > automatically assume these files are ready for commit?  I understand
> > that it's what checkout did, but I'd find it more natural to only
> > affect the working tree by default.  We can give it an option for
> > affecting the index instead (or perhaps in addition).
>
> Yeah, that behavior of updating the index always bothers me when I use
> it but I seemed to forget when working on this.
>
> > > +--ours::
> > > +--theirs::
> > > +       Check out stage #2 ('ours') or #3 ('theirs') for unmerged
> > > +       paths.
> > > ++
> > > +Note that during `git rebase` and `git pull --rebase`, 'ours' and
> > > +'theirs' may appear swapped; `--ours` gives the version from the
> > > +branch the changes are rebased onto, while `--theirs` gives the
> > > +version from the branch that holds your work that is being rebased.
> > > ++
> > > +This is because `rebase` is used in a workflow that treats the
> > > +history at the remote as the shared canonical one, and treats the
> > > +work done on the branch you are rebasing as the third-party work to
> > > +be integrated, and you are temporarily assuming the role of the
> > > +keeper of the canonical history during the rebase.  As the keeper of
> > > +the canonical history, you need to view the history from the remote
> > > +as `ours` (i.e. "our shared canonical history"), while what you did
> > > +on your side branch as `theirs` (i.e. "one contributor's work on top
> > > +of it").
> >
> > Total aside because I'm not sure what you could change here, but man
> > do I hate this.
>
> Uh it's actually documented? I'm always confused by this too. --ours
> and --theirs at this point are pretty much tied to stage 2 and 3.
> Nothing I can do about it. But if you could come up with some other
> option names, then we could make "new ours" to be stage 3 during
> rebase, for example.

I don't think it's a naming issue, personally.  Years ago we could
have defined --ours and --theirs differently based on which kind of
operation we were in the middle of, but you are probably right that
they are now tied to stage 2 and 3.  But there's another place that we
might still be able to address this; I think the brain-damage here may
have just been due to the fact that the recursive merge machinery was
rather inflexible and required HEAD to be stage 2.  If it were a
little more flexible, then we might just be able to make this problem
go away.  Maybe it can still be fixed (I haven't dug too deeply into
it), but if so, the only fix needed here would be to remove this long
explanation about why the tool gets things totally backward.

> > > +Part of the linkgit:git[1] suite
> >
> >
> > My single biggest worry about this whole series is that I'm worried
> > you're perpetuating and even further ingraining one of the biggest
> > usability problems with checkout: people suggest and use it for
> > reverting/restoring paths to a previous version, but it doesn't do
> > that:
> >
> > git restore-files --from master~10 Documentation/
> > <edit some non-documentation files>
> > git add -u
> > git commit -m "Rationale for changing files including reverting Documentation/"
> >
> > In particular, now you have a mixture of files in Documentation/ from
> > master~10 (er, now likely master~11) and HEAD~1; any files and
> > sub-directories that existed in HEAD~1 still remain and are mixed with
> > all other files in Documentation/ from the older commit.
> >
> > You may think this is a special case, but this particular issue
> > results in some pretty bad surprises.  Also, it was pretty surprising
> > to me just how difficult it was to implement an svn-like revert in
> > EasyGit, in large part because of this 'oversight' in git.  git
> > checkout -- <paths> to me has always been fundamentally wrong, but I
> > just wasn't sure if I wanted to fight the backward compatibility
> > battle and suggest changing it.  With a new command, we definitely
> > shouldn't be reinforcing this error.  (And maybe we should consider
> > taking the time to fix git checkout too.)
>
> What would be the right behavior for
>
>  git restore-files --from=master~10 Documentation/
>
> then? Consider it an error? I often use "git checkout HEAD" and "git
> checkout HEAD^" (usually with -p) but not very far back like
> master~10.

Well, when you use a file rather than a directory:
  git restore-files --from=master~10 foo.c
then you expect
  git diff master~10 foo.c
to come back empty afterward.  I expect the same if I give a directory
rather than a file.  (Even if it does make 'restore-files' feel
slightly mis-named.)

> > > +If the branch exists in multiple remotes and one of them is named by
> > > +the `checkout.defaultRemote` configuration variable, we'll use that
> > > +one for the purposes of disambiguation, even if the `<branch>` isn't
> > > +unique across all remotes. Set it to
> > > +e.g. `checkout.defaultRemote=origin` to always checkout remote
> > > +branches from there if `<branch>` is ambiguous but exists on the
> > > +'origin' remote. See also `checkout.defaultRemote` in
> > > +linkgit:git-config[1].
> >
> > So switch-branch will be controlled by checkout.* config variables?
> > That probably makes the most sense, but it does dilute the advantage
> > of adding these simpler commands.
> >
> > Also, the fact that we're trying to make a simpler command makes me
> > think that removing the auto-vivify behavior from the default and
> > adding a simple flag which users can pass to request will allow this
> > part of the documentation to be hidden behind the appropriate flag,
> > which may make it easier for users to compartmentalize the command and
> > it's options, enabling them to learn as they go.
>
> Sounds good. I don't know a good name for this new option though so
> unless anybody comes up with some suggestion, I'll just disable
> checkout.defaultRemote in switch-branch. If it comes back as a new
> option, it can always be added later.
>
> > > +'git switch-branch' -c|-C <new_branch> [<start_point>]::
> > > +
> > > +       Specifying `-c` causes a new branch to be created as if
> > > +       linkgit:git-branch[1] were called and then switched to. In
> > > +       this case you can use the `--track` or `--no-track` options,
> > > +       which will be passed to 'git branch'.  As a convenience,
> > > +       `--track` without `-c` implies branch creation; see the
> > > +       description of `--track` below.
> >
> > Can we get rid of --track/--no-track and just provide a flag (which
> > takes no arguments) for the user to use?  Problems with --track:
> >   * it's not even in your synopsis
> >   * user has to repeat themselves (e.g. 'next' in two places from '-c
> > next --track origin/next'); this repetition is BOTH laborious AND
> > error-prone
> >   * it's rather inconsistent: --track is the default due to
> > auto-vivify when the user specifies nothing but a branch name that
> > doesn't exist yet, but when the user realizes the branch doesn't exist
> > yet and asks to have it created then suddenly tracking is not the
> > default??
>
> I don't think --track is default anymore (maybe I haven't updated the
> man page correctly). The dwim behavior is only activated in
> switch-branch when you specify --guess to reduce the amount of magic
> we throw at the user. With that in mind, do we still hide
> --track/--no-track from switch-branch?

Ooh, you're adding --guess?  Cool, that addresses my concerns, just in
a different manner.

Personally, I'd leave --track/--no-track out.  It's extra mental
overhead, git branch has options for setting those if they need some
special non-default setup, and if there is enough demand for it we can
add it later.  Removing options once published is much harder.

> > I'm not sure what's best, but here's some food for thought:
> >
> >
> >    git switch-branch <branch>
> > switches to <branch>, if it exists.  Error cases:
> >   * If <branch> isn't actually a branch but a <tag> or
> > <remote-tracking-branch> or <revision>, error out and suggest using
> > --detach.
> >   * If <branch> isn't actually a branch but there is a similarly named
> > <remote-tracking-branch> (e.g. origin/<branch>), then suggest using
> > -c.
>
> I would make these advice so I can hide them. Or if I manage to make
> all these hints one line then I'll make it unconditional.
>
> >   git switch-branch -c <branch>
> > creates <branch> and, if a relevant-remote-tracking branch exists,
> > base the branch on that revision and set the new branch up to track
>
> Hmm.. this is a bit magical and could be surprising. If I create (and
> switch to) a new branch foo, I don't necessarily mean tracking
> origin/foo (I may not even think about origin/foo when I type the
> command). So tentatively no.

Yeah, if you're adding --guess then I'm happy.  I do think, though,
that if the user runs switch-branch to a branch that doesn't exist, we
should check if there is an associated remote-tracking branch so that
we can provide a better error message and help users learn about
--guess.  (Also, will there be a short -g form?)

>
> > > +If `-C` is given, <new_branch> is created if it doesn't exist;
> > > +otherwise, it is reset. This is the transactional equivalent of
> > > ++
> > > +------------
> > > +$ git branch -f <branch> [<start_point>]
> > > +$ git switch-branch <branch>
> > > +------------
> > > ++
> > > +that is to say, the branch is not reset/created unless "git
> > > +switch-branch" is successful.
> >
> > ...and when exactly would it fail?  Reading this, it looks like the
> > only possible error condition was removed due saying we'll reset the
> > branch if it already exists, so it's rather confusing.
>
> Yeah probably just scrape it. The atomic nature is not worth highlighting.
>
>
> > > +'git switch-branch' --detach [<commit>]::
> > > +
> > > +       Prepare to work on a unnamed branch on top of <commit> (see
> > > +       "DETACHED HEAD" section), and updating the index and the files
> > > +       in the working tree.  Local modifications to the files in the
> > > +       working tree are kept, so that the resulting working tree will
> > > +       be the state recorded in the commit plus the local
> > > +       modifications.
> > > ++
> > > +When the <commit> argument is a branch name, the `--detach` option can
> > > +be used to detach HEAD at the tip of the branch (`git switch-branch
> > > +<branch>` would check out that branch without detaching HEAD).
> > > ++
> > > +Omitting <commit> detaches HEAD at the tip of the current branch.
> > > +
> > > +OPTIONS
> > > +-------
> > > +-q::
> > > +--quiet::
> > > +       Quiet, suppress feedback messages.
> > > +
> > > +--[no-]progress::
> > > +       Progress status is reported on the standard error stream
> > > +       by default when it is attached to a terminal, unless `--quiet`
> > > +       is specified. This flag enables progress reporting even if not
> > > +       attached to a terminal, regardless of `--quiet`.
> > > +
> > > +-f::
> > > +--force::
> > > +       Proceed even if the index or the working tree differs from
> > > +       HEAD.  This is used to throw away local changes.
> >
> > Haven't thought through this thoroughly, but do we really need an
> > option for that instead of telling users to 'git reset --hard HEAD'
> > before switching branches if they want their stuff thrown away?
>
> For me it's just a bit more convenient. Hit an error when switching
> branch? Recall the command from bash history, stick -f in it and run.
> Elsewhere I think both Junio and Thomas (or maybe only Junio) suggests
> moving the "git reset" functionality without moving HEAD to one of
> these commands, which goes the opposite direction...

Fair enough.

> > > +-c <new_branch>::
> > > +--create <new_branch>::
> > > +       Create a new branch named <new_branch> and start it at
> > > +       <start_point>; see linkgit:git-branch[1] for details.
> > > +
> > > +-C <new_branch>::
> > > +--force-create <new_branch>::
> > > +       Creates the branch <new_branch> and start it at <start_point>;
> > > +       if it already exists, then reset it to <start_point>. This is
> > > +       equivalent to running "git branch" with "-f"; see
> > > +       linkgit:git-branch[1] for details.
> >
> > Makes sense, but let's get the -b/-B vs. -c/-C consistent.
>
> Another option I'm considering is -n/-N (for _new_ branch). Maybe
> -c/-C is good enough.

Actually, -n/-N seems like a good idea, especially since --guess also
creates a branch.  If we do stick with -c/-C, then we may need to
document --guess as implying -c to avoid confusion (and then make sure
we get the synopsis correct to show which flags can be used together).

> > > +-l::
> > > +       Create the new branch's reflog; see linkgit:git-branch[1] for
> > > +       details.
> >
> > ??  Jettison this.
>
> Yep. It looks weird to me too. reflog is just behind the scene these
> days. Nobody need to explicitly ask for reflog anymore.
>
> > > +--orphan <new_branch>::
> > > +       Create a new 'orphan' branch, named <new_branch>, started from
> > > +       <start_point> and switch to it.  The first commit made on this
> >
> > What??  started from <start_point>?  The whole point of --orphan is
> > you have no parent, i.e. no start point.  Also, why does the
> > explanation reference an argument that wasn't in the immediately
> > preceding synopsis?
>
> I guess bad phrasing. It should be "switch to <start_point> first,
> then prepare the worktree so that the first commit will have no
> parent". Or something along that line.
>
> You should really review git-checkout.txt btw ;-)

I did after writing several of these comments, and yeah, it really
needs a clean up.  Seems like something someone would do when writing
a (partial) replacement or simplified alternative.  ;-)

To be fair though, I suspect anyone familiar enough with git that
looks at git-checkout.txt again is probably going to miss at least one
thing that is bad for new users no matter how closely they look, just
because they've grown accustomed to the documentation as it is.  I was
trying to help point out possible issues that I spotted, but I suspect
others may be able to point out more.

> > > +       new branch will have no parents and it will be the root of a new
> > > +       history totally disconnected from all the other branches and
> > > +       commits.
> > > ++
> > > +The index and the working tree are adjusted as if you had previously run
> > > +"git checkout <start_point>".  This allows you to start a new history
> > > +that records a set of paths similar to <start_point> by easily running
> > > +"git commit -a" to make the root commit.
> > > ++
> > > +This can be useful when you want to publish the tree from a commit
> > > +without exposing its full history. You might want to do this to publish
> > > +an open source branch of a project whose current tree is "clean", but
> > > +whose full history contains proprietary or otherwise encumbered bits of
> > > +code.
> > > ++
> > > +If you want to start a disconnected history that records a set of paths
> > > +that is totally different from the one of <start_point>, then you should
> > > +clear the index and the working tree right after creating the orphan
> > > +branch by running "git rm -rf ." from the top level of the working tree.
> > > +Afterwards you will be ready to prepare your new files, repopulating the
> > > +working tree, by copying them from elsewhere, extracting a tarball, etc.
> >
> > Ick.  Seems overly complex.  I'd rather that --orphan defaulted to
> > clearing the index and working tree, and that one would need to pass
> > HEAD for <start_point> if you wanted to start out with all those other
> > files.  That would certainly make the explanation a little clearer to
> > users, and more natural when they start experimenting with it.
> >
> > However, --orphan is pretty special case.  Do we perhaps want to leave
> > it out of this new command and only include it in checkout?
>
> I started this by simply splitting git-checkout in two commands that,
> combined, can do everything git-checkout can. Then suggestions to have
> better default came in and I think we started to drift further to
> _removing_ options and falling back to git-checkout.
>
> I think we could still keep "complicated" options as long as they are
> clearly described and don't surprise users until they figure them out.
> That way I don't have to go back to git-checkout and deal with all the
> ambiguation it creates.

Fair enough...though I think it may make sense to also review the
complicated options and determine if they are overly complicated.  I
think --orphan qualifies (I stumbled with it a bit for years the
occasional time I needed to use it), and my small suggestion above
would simplify both it and its description.  We should probably also
consider just removing <start_point> as an acceptable argument to
--orphan; if people want files from some revision after creating an
orphan branch that's a simple extra command.

> > > +-m::
> > > +--merge::
> > > +       If you have local modifications to one or more files that are
> > > +       different between the current branch and the branch to which
> > > +       you are switching, the command refuses to switch branches in
> > > +       order to preserve your modifications in context.  However,
> > > +       with this option, a three-way merge between the current
> > > +       branch, your working tree contents, and the new branch is
> > > +       done, and you will be on the new branch.
> > > ++
> > > +When a merge conflict happens, the index entries for conflicting
> > > +paths are left unmerged, and you need to resolve the conflicts
> > > +and mark the resolved paths with `git add` (or `git rm` if the merge
> > > +should result in deletion of the path).
> > > +
> > > +--conflict=<style>::
> > > +       The same as --merge option above, but changes the way the
> > > +       conflicting hunks are presented, overriding the
> > > +       merge.conflictStyle configuration variable.  Possible values are
> > > +       "merge" (default) and "diff3" (in addition to what is shown by
> > > +       "merge" style, shows the original contents).
> > > +
> > > +--ignore-other-worktrees::
> > > +       `git switch-branch` refuses when the wanted ref is already
> > > +       checked out by another worktree. This option makes it check
> > > +       the ref out anyway. In other words, the ref can be held by
> > > +       more than one worktree.
> >
> > seems rather dangerous...is the goal to be an easier-to-use suggestion
> > for new users while checkout continues to exist, or is this command
> > meant to handle all branch switching functionality that checkout has?
>
> As explained above. I'm still thinking the latter, but with fewer
> surprises and confusion. Though I guess I could be convinced to go
> with the former (the problem with the former is, even as a
> no-longer-new user, I still find git-checkout not that pleasant to use
> and want a better replacement)
>
> > > +<branch>::
> > > +       Branch to checkout; if it refers to a branch (i.e., a name that,
> > > +       when prepended with "refs/heads/", is a valid ref), then that
> > > +       branch is checked out. Otherwise, if it refers to a valid
> > > +       commit, your HEAD becomes "detached" and you are no longer on
> > > +       any branch (see below for details).
> >
> > I thought we requiring --detach in order to detach.  Does this
> > paragraph need updating?  Also, if we require --detach when we'll be
> > detaching HEAD, then this paragraph gets a LOT simpler.
>
> Yep. I really need to read through the document and update all of it.
>
> > > +You can use the `"@{-N}"` syntax to refer to the N-th last
> > > +branch/commit checked out using "git checkout" operation. You may
> > > +also specify `-` which is synonymous to `"@{-1}`.
> > > ++
> > > +As a special case, you may use `"A...B"` as a shortcut for the
> > > +merge base of `A` and `B` if there is exactly one merge base. You can
> > > +leave out at most one of `A` and `B`, in which case it defaults to `HEAD`.
> >
> > I actually didn't know about the A...B special case for checkout.
> > Interesting...but I'm starting to wonder if this is too much info for
> > a "simplified command".
>
> I could just hint about A...B and send the user to git-checkout.txt if
> they need to know more. They can learn about git-checkout that way
> too.

It may be fine to stay.  Lots of my comments were just "let's try to
note any weirdness or complication that might seem excessive so we at
least have a conversation about it" (because it's easy to gloss over
since we've looked at these documents so many times) rather than a
"this definitely needs to go".

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Elijah Newren]

On Tue, Dec 4, 2018 at 8:28 AM Duy Nguyen <pclouds@gmail.com> wrote:
>
> On Tue, Dec 4, 2018 at 2:29 AM Elijah Newren <newren@gmail.com> wrote:
> >
> > On Thu, Nov 29, 2018 at 2:01 PM Nguyễn Thái Ngọc Duy <pclouds@gmail.com> wrote:
> > >
> > > v3 sees switch-branch go back to switch-branch (in v2 it was
> > > checkout-branch). checkout-files is also renamed restore-files (v1 was
> > > restore-paths). Hopefully we won't see another rename.
> >
> > I started reading through the patches.  I also tried to apply them
> > locally, but they had conflicts or missing base file version on both
> > master and next.  What version did you base it on?
>
> I think nd/checkout-dwim-fix because of a non-trivial conflict there
> (but I don't remember when I noticed it and rebased on that). Anyway
> you can get the whole series at
>
> https://gitlab.com/pclouds/git/tree/switch-branch-and-checkout-files
>
> It fixes some of your comments already, a couple of bug fixes here and
> there and in a good-enough shape that I start actually using it.

Cool.

> > > - Two more fancy features (the "git checkout --index" being the
> > >   default mode and the backup log for accidental overwrites) are of
> > >   course still missing. But they are coming.
> > >
> > > I did not go replace "detached HEAD" with "unnamed branch" (or "no
> > > branch") everywhere because I think a unique term is still good to
> > > refer to this concept. Or maybe "no branch" is good enough. I dunno.
> >
> > I personally like "unnamed branch", but "no branch" would still be
> > better than "detached HEAD".
>
> Haven't really worked on killing the term "detached HEAD" yet. But I
> noticed the other day that git-branch reports
>
> * (HEAD detached from 703266f6e4)
>
> and I didn't know how to rephrase that. I guess "unnamed branch from
> 703266f6e4" is probably good enough but my old-timer brain screams no.

Perhaps "* (On an unnamed branch, at 703266f6e4)"?

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Duy Nguyen]

On Tue, Dec 4, 2018 at 6:43 PM Elijah Newren <newren@gmail.com> wrote:
> > > > +--ours::
> > > > +--theirs::
> > > > +       Check out stage #2 ('ours') or #3 ('theirs') for unmerged
> > > > +       paths.
> > > > ++
> > > > +Note that during `git rebase` and `git pull --rebase`, 'ours' and
> > > > +'theirs' may appear swapped; `--ours` gives the version from the
> > > > +branch the changes are rebased onto, while `--theirs` gives the
> > > > +version from the branch that holds your work that is being rebased.
> > > > ++
> > > > +This is because `rebase` is used in a workflow that treats the
> > > > +history at the remote as the shared canonical one, and treats the
> > > > +work done on the branch you are rebasing as the third-party work to
> > > > +be integrated, and you are temporarily assuming the role of the
> > > > +keeper of the canonical history during the rebase.  As the keeper of
> > > > +the canonical history, you need to view the history from the remote
> > > > +as `ours` (i.e. "our shared canonical history"), while what you did
> > > > +on your side branch as `theirs` (i.e. "one contributor's work on top
> > > > +of it").
> > >
> > > Total aside because I'm not sure what you could change here, but man
> > > do I hate this.
> >
> > Uh it's actually documented? I'm always confused by this too. --ours
> > and --theirs at this point are pretty much tied to stage 2 and 3.
> > Nothing I can do about it. But if you could come up with some other
> > option names, then we could make "new ours" to be stage 3 during
> > rebase, for example.
>
> I don't think it's a naming issue, personally.  Years ago we could
> have defined --ours and --theirs differently based on which kind of
> operation we were in the middle of, but you are probably right that
> they are now tied to stage 2 and 3.  But there's another place that we
> might still be able to address this; I think the brain-damage here may
> have just been due to the fact that the recursive merge machinery was
> rather inflexible and required HEAD to be stage 2.  If it were a
> little more flexible, then we might just be able to make this problem
> go away.  Maybe it can still be fixed (I haven't dug too deeply into
> it), but if so, the only fix needed here would be to remove this long
> explanation about why the tool gets things totally backward.

Aha. I' not really deep in this merge business to know if stages 2 and
3 can be swapped. This is right up your alley. I'll just leave it to
you.

> > > > +'git switch-branch' -c|-C <new_branch> [<start_point>]::
> > > > +
> > > > +       Specifying `-c` causes a new branch to be created as if
> > > > +       linkgit:git-branch[1] were called and then switched to. In
> > > > +       this case you can use the `--track` or `--no-track` options,
> > > > +       which will be passed to 'git branch'.  As a convenience,
> > > > +       `--track` without `-c` implies branch creation; see the
> > > > +       description of `--track` below.
> > >
> > > Can we get rid of --track/--no-track and just provide a flag (which
> > > takes no arguments) for the user to use?  Problems with --track:
> > >   * it's not even in your synopsis
> > >   * user has to repeat themselves (e.g. 'next' in two places from '-c
> > > next --track origin/next'); this repetition is BOTH laborious AND
> > > error-prone
> > >   * it's rather inconsistent: --track is the default due to
> > > auto-vivify when the user specifies nothing but a branch name that
> > > doesn't exist yet, but when the user realizes the branch doesn't exist
> > > yet and asks to have it created then suddenly tracking is not the
> > > default??
> >
> > I don't think --track is default anymore (maybe I haven't updated the
> > man page correctly). The dwim behavior is only activated in
> > switch-branch when you specify --guess to reduce the amount of magic
> > we throw at the user. With that in mind, do we still hide
> > --track/--no-track from switch-branch?
>
> Ooh, you're adding --guess?  Cool, that addresses my concerns, just in
> a different manner.

No it's always there even in git-checkout, just hidden.

> Personally, I'd leave --track/--no-track out.  It's extra mental
> overhead, git branch has options for setting those if they need some
> special non-default setup, and if there is enough demand for it we can
> add it later.  Removing options once published is much harder.

Slightly less convenient (you would need a combination of git-branch
and git-switch-branch, if you avoid git-checkout). But since I don't
think I have ever used this option, I'm fine with leaving it out and
considering adding it back later.

> > > I'm not sure what's best, but here's some food for thought:
> > >
> > >
> > >    git switch-branch <branch>
> > > switches to <branch>, if it exists.  Error cases:
> > >   * If <branch> isn't actually a branch but a <tag> or
> > > <remote-tracking-branch> or <revision>, error out and suggest using
> > > --detach.
> > >   * If <branch> isn't actually a branch but there is a similarly named
> > > <remote-tracking-branch> (e.g. origin/<branch>), then suggest using
> > > -c.
> >
> > I would make these advice so I can hide them. Or if I manage to make
> > all these hints one line then I'll make it unconditional.
> >
> > >   git switch-branch -c <branch>
> > > creates <branch> and, if a relevant-remote-tracking branch exists,
> > > base the branch on that revision and set the new branch up to track
> >
> > Hmm.. this is a bit magical and could be surprising. If I create (and
> > switch to) a new branch foo, I don't necessarily mean tracking
> > origin/foo (I may not even think about origin/foo when I type the
> > command). So tentatively no.
>
> Yeah, if you're adding --guess then I'm happy.  I do think, though,
> that if the user runs switch-branch to a branch that doesn't exist, we
> should check if there is an associated remote-tracking branch so that
> we can provide a better error message and help users learn about
> --guess.  (Also, will there be a short -g form?)

Yeah better error and suggestion is a good idea. And yes the short
form -g is already added (I did try to use it and find --guess too
time consuming even with bash completion support).

> > > > +-f::
> > > > +--force::
> > > > +       Proceed even if the index or the working tree differs from
> > > > +       HEAD.  This is used to throw away local changes.
> > >
> > > Haven't thought through this thoroughly, but do we really need an
> > > option for that instead of telling users to 'git reset --hard HEAD'
> > > before switching branches if they want their stuff thrown away?
> >
> > For me it's just a bit more convenient. Hit an error when switching
> > branch? Recall the command from bash history, stick -f in it and run.
> > Elsewhere I think both Junio and Thomas (or maybe only Junio) suggests
> > moving the "git reset" functionality without moving HEAD to one of
> > these commands, which goes the opposite direction...
>
> Fair enough.

I'm actually still not sure how to move it here (I guess 'here' is
restore-files since we won't move HEAD). All the --mixed, --merge and
--hard are confusing. But maybe we could just make 'git restore-files
--from HEAD -f :/" behave just like "git reset --hard HEAD" (but with
some safety net) But we can leave it for discussion in the next round.

> > > > +--orphan <new_branch>::
> > > > +       Create a new 'orphan' branch, named <new_branch>, started from
> > > > +       <start_point> and switch to it.  The first commit made on this
> > >
> > > What??  started from <start_point>?  The whole point of --orphan is
> > > you have no parent, i.e. no start point.  Also, why does the
> > > explanation reference an argument that wasn't in the immediately
> > > preceding synopsis?
> >
> > I guess bad phrasing. It should be "switch to <start_point> first,
> > then prepare the worktree so that the first commit will have no
> > parent". Or something along that line.
> >
> > You should really review git-checkout.txt btw ;-)
>
> I did after writing several of these comments, and yeah, it really
> needs a clean up.  Seems like something someone would do when writing
> a (partial) replacement or simplified alternative.  ;-)

Heh ;-) Fine I'll do it. I have to read and re-read git-checkout.txt
anyway and already queued up a couple small fixes.

> > > > +       new branch will have no parents and it will be the root of a new
> > > > +       history totally disconnected from all the other branches and
> > > > +       commits.
> > > > ++
> > > > +The index and the working tree are adjusted as if you had previously run
> > > > +"git checkout <start_point>".  This allows you to start a new history
> > > > +that records a set of paths similar to <start_point> by easily running
> > > > +"git commit -a" to make the root commit.
> > > > ++
> > > > +This can be useful when you want to publish the tree from a commit
> > > > +without exposing its full history. You might want to do this to publish
> > > > +an open source branch of a project whose current tree is "clean", but
> > > > +whose full history contains proprietary or otherwise encumbered bits of
> > > > +code.
> > > > ++
> > > > +If you want to start a disconnected history that records a set of paths
> > > > +that is totally different from the one of <start_point>, then you should
> > > > +clear the index and the working tree right after creating the orphan
> > > > +branch by running "git rm -rf ." from the top level of the working tree.
> > > > +Afterwards you will be ready to prepare your new files, repopulating the
> > > > +working tree, by copying them from elsewhere, extracting a tarball, etc.
> > >
> > > Ick.  Seems overly complex.  I'd rather that --orphan defaulted to
> > > clearing the index and working tree, and that one would need to pass
> > > HEAD for <start_point> if you wanted to start out with all those other
> > > files.  That would certainly make the explanation a little clearer to
> > > users, and more natural when they start experimenting with it.
> > >
> > > However, --orphan is pretty special case.  Do we perhaps want to leave
> > > it out of this new command and only include it in checkout?
> >
> > I started this by simply splitting git-checkout in two commands that,
> > combined, can do everything git-checkout can. Then suggestions to have
> > better default came in and I think we started to drift further to
> > _removing_ options and falling back to git-checkout.
> >
> > I think we could still keep "complicated" options as long as they are
> > clearly described and don't surprise users until they figure them out.
> > That way I don't have to go back to git-checkout and deal with all the
> > ambiguation it creates.
>
> Fair enough...though I think it may make sense to also review the
> complicated options and determine if they are overly complicated.  I
> think --orphan qualifies (I stumbled with it a bit for years the
> occasional time I needed to use it), and my small suggestion above
> would simplify both it and its description.  We should probably also
> consider just removing <start_point> as an acceptable argument to
> --orphan; if people want files from some revision after creating an
> orphan branch that's a simple extra command.

It is good that you pointed it out though. I still have to digest this
option before I make more comments, but generally if there's a simpler
(even if new) way to achieve the same thing, I'm all for it.
-- 
Duy

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Duy Nguyen]

On Tue, Dec 4, 2018 at 6:45 PM Elijah Newren <newren@gmail.com> wrote:
> > > > - Two more fancy features (the "git checkout --index" being the
> > > >   default mode and the backup log for accidental overwrites) are of
> > > >   course still missing. But they are coming.
> > > >
> > > > I did not go replace "detached HEAD" with "unnamed branch" (or "no
> > > > branch") everywhere because I think a unique term is still good to
> > > > refer to this concept. Or maybe "no branch" is good enough. I dunno.
> > >
> > > I personally like "unnamed branch", but "no branch" would still be
> > > better than "detached HEAD".
> >
> > Haven't really worked on killing the term "detached HEAD" yet. But I
> > noticed the other day that git-branch reports
> >
> > * (HEAD detached from 703266f6e4)
> >
> > and I didn't know how to rephrase that. I guess "unnamed branch from
> > 703266f6e4" is probably good enough but my old-timer brain screams no.
>
> Perhaps "* (On an unnamed branch, at 703266f6e4)"?

This 703266f6e4 is the fork point. Once you start adding more commits
on top of this unnamed branch, I find it hard to define it "at"
703266f6e4 anymore. "forked from 703266f6e4" (or even starting/growing
from...) is probably clearest but also a bit longer.
-- 
Duy

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Elijah Newren]

On Tue, Dec 4, 2018 at 10:22 AM Duy Nguyen <pclouds@gmail.com> wrote:
>
> On Tue, Dec 4, 2018 at 6:45 PM Elijah Newren <newren@gmail.com> wrote:
> > > > > - Two more fancy features (the "git checkout --index" being the
> > > > >   default mode and the backup log for accidental overwrites) are of
> > > > >   course still missing. But they are coming.
> > > > >
> > > > > I did not go replace "detached HEAD" with "unnamed branch" (or "no
> > > > > branch") everywhere because I think a unique term is still good to
> > > > > refer to this concept. Or maybe "no branch" is good enough. I dunno.
> > > >
> > > > I personally like "unnamed branch", but "no branch" would still be
> > > > better than "detached HEAD".
> > >
> > > Haven't really worked on killing the term "detached HEAD" yet. But I
> > > noticed the other day that git-branch reports
> > >
> > > * (HEAD detached from 703266f6e4)
> > >
> > > and I didn't know how to rephrase that. I guess "unnamed branch from
> > > 703266f6e4" is probably good enough but my old-timer brain screams no.
> >
> > Perhaps "* (On an unnamed branch, at 703266f6e4)"?
>
> This 703266f6e4 is the fork point. Once you start adding more commits
> on top of this unnamed branch, I find it hard to define it "at"
> 703266f6e4 anymore. "forked from 703266f6e4" (or even starting/growing
> from...) is probably clearest but also a bit longer.

It reports the fork point rather than the commit HEAD points to?  Ah,
I guess I never payed that close of attention before.  I actually
think "on an unnamed branch" is good enough, but if others gain value
from the extra info, then I understand the conundrum.  I'm not sure
what the use or rationale is for the fork point, though, so I feel
slightly at a loss to try to describe this extra piece of info.

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Duy Nguyen]

On Tue, Dec 4, 2018 at 7:31 PM Elijah Newren <newren@gmail.com> wrote:
>
> On Tue, Dec 4, 2018 at 10:22 AM Duy Nguyen <pclouds@gmail.com> wrote:
> >
> > On Tue, Dec 4, 2018 at 6:45 PM Elijah Newren <newren@gmail.com> wrote:
> > > > > > - Two more fancy features (the "git checkout --index" being the
> > > > > >   default mode and the backup log for accidental overwrites) are of
> > > > > >   course still missing. But they are coming.
> > > > > >
> > > > > > I did not go replace "detached HEAD" with "unnamed branch" (or "no
> > > > > > branch") everywhere because I think a unique term is still good to
> > > > > > refer to this concept. Or maybe "no branch" is good enough. I dunno.
> > > > >
> > > > > I personally like "unnamed branch", but "no branch" would still be
> > > > > better than "detached HEAD".
> > > >
> > > > Haven't really worked on killing the term "detached HEAD" yet. But I
> > > > noticed the other day that git-branch reports
> > > >
> > > > * (HEAD detached from 703266f6e4)
> > > >
> > > > and I didn't know how to rephrase that. I guess "unnamed branch from
> > > > 703266f6e4" is probably good enough but my old-timer brain screams no.
> > >
> > > Perhaps "* (On an unnamed branch, at 703266f6e4)"?
> >
> > This 703266f6e4 is the fork point. Once you start adding more commits
> > on top of this unnamed branch, I find it hard to define it "at"
> > 703266f6e4 anymore. "forked from 703266f6e4" (or even starting/growing
> > from...) is probably clearest but also a bit longer.
>
> It reports the fork point rather than the commit HEAD points to?  Ah,
> I guess I never payed that close of attention before.  I actually
> think "on an unnamed branch" is good enough, but if others gain value
> from the extra info, then I understand the conundrum.  I'm not sure
> what the use or rationale is for the fork point, though, so I feel
> slightly at a loss to try to describe this extra piece of info.

It's probably a corner case. This is a better example

* (HEAD detached at pclouds/backup-log)

It does help see i'm working on top of some branch (or tag)
-- 
Duy

Re: [PATCH/RFC v3 00/14] Introduce new commands switch-branch and restore-files

[Eric Sunshine]

On Tue, Dec 4, 2018 at 11:28 AM Duy Nguyen <pclouds@gmail.com> wrote:
> Haven't really worked on killing the term "detached HEAD" yet. But I
> noticed the other day that git-branch reports
>
> * (HEAD detached from 703266f6e4)
>
> and I didn't know how to rephrase that. I guess "unnamed branch from
> 703266f6e4" is probably good enough but my old-timer brain screams no.

"git worktree add" and "git worktree show" also report similar messages.

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Junio C Hamano]

Duy Nguyen <pclouds@gmail.com> writes:

>> My single biggest worry about this whole series is that I'm worried
>> you're perpetuating and even further ingraining one of the biggest
>> usability problems with checkout: people suggest and use it for
>> reverting/restoring paths to a previous version, but it doesn't do
>> that:
>
> ...
>
>  git restore-files --from=master~10 Documentation/

The "single biggest worry" could be due to Elijah not being aware of
other recent discussions.  My understanding of the plan is

 - "git checkout" will learn a new "--[no-]overlay" option, where
   the current behaviour, i.e. "take paths in master~10 that match
   pathspec Documentation/, and overlay them on top of what is in
   the index and the working tree", is explained as "the overlay
   mode" and stays to be the default.  With "checkout --no-overlay
   master~10 Documentation/", the command will become "replace paths
   in the current index and the working tree that match the pathspec
   Documentation/ with paths in master~10 that match pathspec
   Documentation/".

 - "git restore-files --from=<tree> <pathspec>" by default will use
   "--no-overlay" semantics, but the users can still use "--overlay"
   from the command line as an option.

So "restore-files" would become truly "restore the state of
Documentation/ to match that of master~10", I would think.

>> Also, the fact that we're trying to make a simpler command makes me
>> think that removing the auto-vivify behavior from the default and
>> adding a simple flag which users can pass to request will allow this
>> part of the documentation to be hidden behind the appropriate flag,
>> which may make it easier for users to compartmentalize the command and
>> it's options, enabling them to learn as they go.
>
> Sounds good. I don't know a good name for this new option though so
> unless anybody comes up with some suggestion, I'll just disable
> checkout.defaultRemote in switch-branch. If it comes back as a new
> option, it can always be added later.

Are you two discussing the "checkout --guess" option?  I am somewhat
lost here.

>> > +-f::
>> > +--force::
>> > +       Proceed even if the index or the working tree differs from
>> > +       HEAD.  This is used to throw away local changes.
>>
>> Haven't thought through this thoroughly, but do we really need an
>> option for that instead of telling users to 'git reset --hard HEAD'
>> before switching branches if they want their stuff thrown away?
>
> For me it's just a bit more convenient. Hit an error when switching
> branch? Recall the command from bash history, stick -f in it and run.
> Elsewhere I think both Junio and Thomas (or maybe only Junio) suggests
> moving the "git reset" functionality without moving HEAD to one of
> these commands, which goes the opposite direction...

Isn't there a huge difference?  "checkout --force <other-branch>"
needs to clobber only the changes that are involved in the switch,
i.e. if your README.txt is the same between master and maint while
Makefile is different, after editing both files while on master, you
can not "switch-branch" to maint without doing something to Makefile
(i.e. either discard your local change or wiggle your local change
to the context of 'maint' with "checkout -m").  But you can carry
the changes to README.txt while checking out 'maint' branch.
Running "git reset --hard HEAD" would mean that you will lose the
changes to README.txt as well.

>> > +--orphan <new_branch>::
>> > +       Create a new 'orphan' branch, named <new_branch>, started from
>> > +       <start_point> and switch to it.  The first commit made on this
>>
>> What??  started from <start_point>?  The whole point of --orphan is
>> you have no parent, i.e. no start point.  Also, why does the
>> explanation reference an argument that wasn't in the immediately
>> preceding synopsis?
>
> I guess bad phrasing. It should be "switch to <start_point> first,
> then prepare the worktree so that the first commit will have no
> parent". Or something along that line.

It should be a <tree-ish>, no?  It is not a "point" in history, but
is "start with this tree".

I may have more comments on this message but that's it from me for
now.

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Junio C Hamano]

Duy Nguyen <pclouds@gmail.com> writes:

> On Tue, Dec 4, 2018 at 6:43 PM Elijah Newren <newren@gmail.com> wrote:
>> > > > +--ours::
>> > > > +--theirs::
>> ...
>> go away.  Maybe it can still be fixed (I haven't dug too deeply into
>> it), but if so, the only fix needed here would be to remove this long
>> explanation about why the tool gets things totally backward.
>
> Aha. I' not really deep in this merge business to know if stages 2 and
> 3 can be swapped. This is right up your alley. I'll just leave it to
> you.

Please don't show stage#2 and stage#3 swapped to the end user,
unless that is protected behind an option (not per-repo config).
It is pretty much ingrained that stage#2 is what came from the
commit that will become the first parent of the commit being
prepared, and changing it without an explicit command line option
will break tools.

> I'm actually still not sure how to move it here (I guess 'here' is
> restore-files since we won't move HEAD). All the --mixed, --merge and
> --hard are confusing. But maybe we could just make 'git restore-files
> --from HEAD -f :/" behave just like "git reset --hard HEAD" (but with
> some safety net) But we can leave it for discussion in the next round.

Perhaps you two should pay a bit closer attention to what Thomas
Gummerer is working on.  I've touched above in my earlier comments,
too, e.g.  <xmqqefb3mhrs.fsf@gitster-ct.c.googlers.com>

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Elijah Newren]

On Tue, Dec 4, 2018 at 6:14 PM Junio C Hamano <gitster@pobox.com> wrote:
>
> Duy Nguyen <pclouds@gmail.com> writes:
>
> >> My single biggest worry about this whole series is that I'm worried
> >> you're perpetuating and even further ingraining one of the biggest
> >> usability problems with checkout: people suggest and use it for
> >> reverting/restoring paths to a previous version, but it doesn't do
> >> that:
> >
> > ...
> >
> >  git restore-files --from=master~10 Documentation/
>
> The "single biggest worry" could be due to Elijah not being aware of
> other recent discussions.  My understanding of the plan is
>
>  - "git checkout" will learn a new "--[no-]overlay" option, where
>    the current behaviour, i.e. "take paths in master~10 that match
>    pathspec Documentation/, and overlay them on top of what is in
>    the index and the working tree", is explained as "the overlay
>    mode" and stays to be the default.  With "checkout --no-overlay
>    master~10 Documentation/", the command will become "replace paths
>    in the current index and the working tree that match the pathspec
>    Documentation/ with paths in master~10 that match pathspec
>    Documentation/".
>
>  - "git restore-files --from=<tree> <pathspec>" by default will use
>    "--no-overlay" semantics, but the users can still use "--overlay"
>    from the command line as an option.
>
> So "restore-files" would become truly "restore the state of
> Documentation/ to match that of master~10", I would think.

Oh, sweet, that's awesome.  I saw some of the other emails as I was
scanning through and looked at the --overlay stuff since it sounded
relevant but something in the explanation made me think it was about
whether the command would write to the index or the working tree,
rather than being about adding+overwriting vs. just overwriting.

> >> Also, the fact that we're trying to make a simpler command makes me
> >> think that removing the auto-vivify behavior from the default and
> >> adding a simple flag which users can pass to request will allow this
> >> part of the documentation to be hidden behind the appropriate flag,
> >> which may make it easier for users to compartmentalize the command and
> >> it's options, enabling them to learn as they go.
> >
> > Sounds good. I don't know a good name for this new option though so
> > unless anybody comes up with some suggestion, I'll just disable
> > checkout.defaultRemote in switch-branch. If it comes back as a new
> > option, it can always be added later.
>
> Are you two discussing the "checkout --guess" option?  I am somewhat
> lost here.

Generally what was being discussed was just that this manpage was
rather complicated for the standard base-case due to the need to
explain all the behavior associated with --guess since that option is
on by default in checkout.  And since --guess is controlled by
checkout.defaultRemote, that was part of the extra complexity that had
to be learned in the basic explanation, rather than letting users
learn it when they learn a new flag.

The critical part you were missing was part of the original text just
before the quoted part was:

>>> So switch-branch will be controlled by checkout.* config variables?
>>> That probably makes the most sense, but it does dilute the advantage
>>> of adding these simpler commands.

Duy is responding to that even if it wasn't included in his quoting.

> >> > +-f::
> >> > +--force::
> >> > +       Proceed even if the index or the working tree differs from
> >> > +       HEAD.  This is used to throw away local changes.
> >>
> >> Haven't thought through this thoroughly, but do we really need an
> >> option for that instead of telling users to 'git reset --hard HEAD'
> >> before switching branches if they want their stuff thrown away?
> >
> > For me it's just a bit more convenient. Hit an error when switching
> > branch? Recall the command from bash history, stick -f in it and run.
> > Elsewhere I think both Junio and Thomas (or maybe only Junio) suggests
> > moving the "git reset" functionality without moving HEAD to one of
> > these commands, which goes the opposite direction...
>
> Isn't there a huge difference?  "checkout --force <other-branch>"
> needs to clobber only the changes that are involved in the switch,
> i.e. if your README.txt is the same between master and maint while
> Makefile is different, after editing both files while on master, you
> can not "switch-branch" to maint without doing something to Makefile
> (i.e. either discard your local change or wiggle your local change
> to the context of 'maint' with "checkout -m").  But you can carry
> the changes to README.txt while checking out 'maint' branch.
> Running "git reset --hard HEAD" would mean that you will lose the
> changes to README.txt as well.

Ah, indeed.  Thanks for pointing out what I missed here.

> >> > +--orphan <new_branch>::
> >> > +       Create a new 'orphan' branch, named <new_branch>, started from
> >> > +       <start_point> and switch to it.  The first commit made on this
> >>
> >> What??  started from <start_point>?  The whole point of --orphan is
> >> you have no parent, i.e. no start point.  Also, why does the
> >> explanation reference an argument that wasn't in the immediately
> >> preceding synopsis?
> >
> > I guess bad phrasing. It should be "switch to <start_point> first,
> > then prepare the worktree so that the first commit will have no
> > parent". Or something along that line.
>
> It should be a <tree-ish>, no?  It is not a "point" in history, but
> is "start with this tree".

Are you saying that referring to it as a tree will lessen the
confusion users face when they think it's a commit that serves as a
parent and get confused with the fact that this option is named
"--orphan"?  Or are you making some other orthogonal point here?

> I may have more comments on this message but that's it from me for
> now.

Fair enough.  Sorry if I've distracted from RC stuff with all my responses.

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Elijah Newren]

On Tue, Dec 4, 2018 at 6:26 PM Junio C Hamano <gitster@pobox.com> wrote:
>
> Duy Nguyen <pclouds@gmail.com> writes:
>
> > On Tue, Dec 4, 2018 at 6:43 PM Elijah Newren <newren@gmail.com> wrote:
> >> > > > +--ours::
> >> > > > +--theirs::
> >> ...
> >> go away.  Maybe it can still be fixed (I haven't dug too deeply into
> >> it), but if so, the only fix needed here would be to remove this long
> >> explanation about why the tool gets things totally backward.
> >
> > Aha. I' not really deep in this merge business to know if stages 2 and
> > 3 can be swapped. This is right up your alley. I'll just leave it to
> > you.
>
> Please don't show stage#2 and stage#3 swapped to the end user,
> unless that is protected behind an option (not per-repo config).
> It is pretty much ingrained that stage#2 is what came from the
> commit that will become the first parent of the commit being
> prepared, and changing it without an explicit command line option
> will break tools.

What depends on stage#2 coming from the commit that will become the
first parent?  I wasn't thinking in terms of modifying
checkout/restore-files/diff/etc in a way that would make them show
things different than what was recorded in the index, I was rather
musing on whether it was feasible to have rebase tell the merge
machinery to treat HEAD as stage #3 and the other commit as stage #2
so that it was swapping what was actually recorded in the index.

I know the merge machinery implicitly assumes HEAD == stage #2 in
multiple places, and it'd obviously need a fair amount of fixing to
handle this.  I wasn't immediately aware of other things that would
break.  If you know of some, I'm happy to hear.  Otherwise, I might go
and learn the hard way (after I get around to the merge rewrite) why
my idea is crazy.  :-)

> > I'm actually still not sure how to move it here (I guess 'here' is
> > restore-files since we won't move HEAD). All the --mixed, --merge and
> > --hard are confusing. But maybe we could just make 'git restore-files
> > --from HEAD -f :/" behave just like "git reset --hard HEAD" (but with
> > some safety net) But we can leave it for discussion in the next round.
>
> Perhaps you two should pay a bit closer attention to what Thomas
> Gummerer is working on.  I've touched above in my earlier comments,
> too, e.g.  <xmqqefb3mhrs.fsf@gitster-ct.c.googlers.com>

Indeed, I'm excited about his changes now; I'll keep an eye out.

Re: [PATCH v3 07/14] checkout: split into switch-branch and restore-files

[Junio C Hamano]

Elijah Newren <newren@gmail.com> writes:

> What depends on stage#2 coming from the commit that will become the
> first parent?

How about "git diff --cc" for a starter?  What came from HEAD's
ancestry should appear first and then what came from the side branch
that is merged into.

Re: [PATCH v2] checkout: print something when checking out paths

[Junio C Hamano]

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

> One of the problems with "git checkout" is that it does so many
> different things and could confuse people specially when we fail to
> handle ambiguation correctly.
>
> One way to help with that is tell the user what sort of operation is
> actually carried out. When switching branches, we always print
> something unless --quiet, either
>
>  - "HEAD is now at ..."
>  - "Reset branch ..."
>  - "Already on ..."
>  - "Switched to and reset ..."
>  - "Switched to a new branch ..."
>  - "Switched to branch ..."
>
> Checking out paths however is silent. Print something so that if we
> got the user intention wrong, they won't waste too much time to find
> that out. For the remaining cases of checkout we now print either
>
>  - "Checked out ... paths out of the index"
>  - "Checked out ... paths out of <abbrev hash>"
>
> Since the purpose of printing this is to help disambiguate. Only do it
> when "--" is missing.
>
> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
> ---
>  v2 updates the messages a bit but it does not check isatty or add
>  --count-paths, for consistency reason with how messages are printed
>  in the branch switching case.
>  
>  Consistency is not always a good reason to follow. But I haven't
>  seen a strong reason to go against it.

One small bug I saw since this was merged is that the message that
is given when unmerging, i.e.

	git merge other-branch   ;# conflicts
	git checkout --m <path>

is misleading.  It gives the same "checked out ... out of the index",
but it should be made a lot more distinct, perhaps "unmerged N paths".

Re: [PATCH v2] checkout: print something when checking out paths

[Duy Nguyen]

On Tue, Jan 29, 2019 at 4:58 AM Junio C Hamano <gitster@pobox.com> wrote:
> One small bug I saw since this was merged is that the message that
> is given when unmerging, i.e.
>
>         git merge other-branch   ;# conflicts
>         git checkout --m <path>
>
> is misleading.  It gives the same "checked out ... out of the index",
> but it should be made a lot more distinct, perhaps "unmerged N paths".

Right, I missed this. How about "re-created %d merge conflicts"? It's
a bit clearer, I think.
-- 
Duy

[PATCH 0/2] nd/checkout-noisy updates

[Nguyễn Thái Ngọc Duy]

This fixes the misleading message from "git checkout -m". I also
rephrased the original messages a bit so that when
tg/checkout-no-overlay is merged, fixing counting file deletion is one
line change.

Nguyễn Thái Ngọc Duy (2):
  checkout: update count-checkouts messages
  checkout: count and print -m paths separately

 builtin/checkout.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

-- 
2.20.1.682.gd5861c6d90

[PATCH 1/2] checkout: update count-checkouts messages

[Nguyễn Thái Ngọc Duy]

Commit 0f086e6dca [1] counts the number of files updated by "git
checkout -- <paths>" command and prints it. Later on 536ec1839d [2]
adds the ability to remove files in "git checkout -- <paths>". This is
still an update on worktree and should be reported to the user.

To prepare for such an update since that commit is on track to
'master' now, the messages are rephrased to avoid "checked out" which
does not imply file deletion.

[1] 0f086e6dca (checkout: print something when checking out paths -
    2018-11-13)
[2] 536ec1839d (entry: support CE_WT_REMOVE flag in checkout_entry -
    2018-12-20)

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 9f8f3466f6..6e33850d9f 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -393,15 +393,15 @@ static int checkout_paths(const struct checkout_opts *opts,
 
 	if (opts->count_checkout_paths) {
 		if (opts->source_tree)
-			fprintf_ln(stderr, Q_("Checked out %d path out of %s",
-					      "Checked out %d paths out of %s",
+			fprintf_ln(stderr, Q_("Updated %d path from %s",
+					      "Updated %d paths from %s",
 					      nr_checkouts),
 				   nr_checkouts,
 				   find_unique_abbrev(&opts->source_tree->object.oid,
 						      DEFAULT_ABBREV));
 		else
-			fprintf_ln(stderr, Q_("Checked out %d path out of the index",
-					      "Checked out %d paths out of the index",
+			fprintf_ln(stderr, Q_("Updated %d path from the index",
+					      "Updated %d paths from the index",
 					      nr_checkouts),
 				   nr_checkouts);
 	}
-- 
2.20.1.682.gd5861c6d90

[PATCH 2/2] checkout: count and print -m paths separately

[Nguyễn Thái Ngọc Duy]

Since 0f086e6dca (checkout: print something when checking out paths -
2018-11-13), this command reports how many paths have been updated
from what source (either from a tree, or from the index). I forget
that there's a third source: when -m is used, the merge conflict is
re-created (granted, also from the index, but it's not a straight copy
from the index).

Count and report unmerged paths separately. There's a bit more update
to avoid reporting:

   Recreated X merge conflicts
   Updated 0 paths from the index

The second line is unnecessary. Though if there's no conflict
recreation, we still report

   Updated 0 paths from the index

to make it clear we're not really doing anything.

Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
---
 builtin/checkout.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)

diff --git a/builtin/checkout.c b/builtin/checkout.c
index 6e33850d9f..e762b53542 100644
--- a/builtin/checkout.c
+++ b/builtin/checkout.c
@@ -260,7 +260,7 @@ static int checkout_paths(const struct checkout_opts *opts,
 	struct commit *head;
 	int errs = 0;
 	struct lock_file lock_file = LOCK_INIT;
-	int nr_checkouts = 0;
+	int nr_checkouts = 0, nr_unmerged = 0;
 
 	if (opts->track != BRANCH_TRACK_UNSPECIFIED)
 		die(_("'%s' cannot be used with updating paths"), "--track");
@@ -385,13 +385,18 @@ static int checkout_paths(const struct checkout_opts *opts,
 						       &state, &nr_checkouts);
 			else if (opts->merge)
 				errs |= checkout_merged(pos, &state,
-							&nr_checkouts);
+							&nr_unmerged);
 			pos = skip_same_name(ce, pos) - 1;
 		}
 	}
 	errs |= finish_delayed_checkout(&state, &nr_checkouts);
 
 	if (opts->count_checkout_paths) {
+		if (nr_unmerged)
+			fprintf_ln(stderr, Q_("Recreated %d merge conflict",
+					      "Recreated %d merge conflicts",
+					      nr_unmerged),
+				   nr_unmerged);
 		if (opts->source_tree)
 			fprintf_ln(stderr, Q_("Updated %d path from %s",
 					      "Updated %d paths from %s",
@@ -399,7 +404,7 @@ static int checkout_paths(const struct checkout_opts *opts,
 				   nr_checkouts,
 				   find_unique_abbrev(&opts->source_tree->object.oid,
 						      DEFAULT_ABBREV));
-		else
+		else if (!nr_unmerged || nr_checkouts)
 			fprintf_ln(stderr, Q_("Updated %d path from the index",
 					      "Updated %d paths from the index",
 					      nr_checkouts),
-- 
2.20.1.682.gd5861c6d90

Re: [PATCH v3 11/14] switch-branch: only allow explicit detached HEAD

[Eckhard Maaß]

On Thu, Nov 29, 2018 at 10:58:46PM +0100, Nguyễn Thái Ngọc Duy wrote:
> +	if (!opts->implicit_detach &&
> +	    !opts->new_branch &&
> +	    !opts->new_branch_force &&
> +	    new_branch_info->name &&
> +	    !new_branch_info->path)
> +		die(_("a branch is expected, got %s"), new_branch_info->name);

Wouldn't it be nice to give more context here, for example the symbolic
reference that the name actually points to? When expereimenting with the
feature and trying to switch to a tag, it refuses with
"a branch is expected, got v1.2.0". I personally would prefer something
more like "a branch is expected, got v1.2.0 that resolved to
refs/tags/v1.2.0", so I get "oh, yeah, that is actually a tag ...". Does
this seem worthwhile to dig deeper into? A quick glance left me a bit
puzzled, I admit.

Greetings,
Eckhard

Re: [PATCH v3 11/14] switch-branch: only allow explicit detached HEAD

[Duy Nguyen]

On Mon, Mar 11, 2019 at 2:32 AM Eckhard Maaß
<eckhard.s.maass@googlemail.com> wrote:
>
> On Thu, Nov 29, 2018 at 10:58:46PM +0100, Nguyễn Thái Ngọc Duy wrote:
> > +     if (!opts->implicit_detach &&
> > +         !opts->new_branch &&
> > +         !opts->new_branch_force &&
> > +         new_branch_info->name &&
> > +         !new_branch_info->path)
> > +             die(_("a branch is expected, got %s"), new_branch_info->name);
>
> Wouldn't it be nice to give more context here, for example the symbolic
> reference that the name actually points to? When expereimenting with the
> feature and trying to switch to a tag, it refuses with
> "a branch is expected, got v1.2.0". I personally would prefer something
> more like "a branch is expected, got v1.2.0 that resolved to
> refs/tags/v1.2.0", so I get "oh, yeah, that is actually a tag ...". Does
> this seem worthwhile to dig deeper into? A quick glance left me a bit
> puzzled, I admit.

Good suggestion. I'll try to report one of the following

a branch is expected, got tag 'v1.2.0'
a branch is expected, got remote branch 'origin/master'
a branch is expected, got 'refs/foo/bar'
a branch is expected, got commit 'HEAD^'

It's a bit more code, but I think it definitely helps.
-- 
Duy