git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
* [PATCH] pull: fix 'git pull --all' when current branch is tracking remote that is not last in the list of remotes
@ 2010-02-23 22:55 Michael Lukashov
  2010-02-23 23:02 ` Junio C Hamano
  0 siblings, 1 reply; 7+ messages in thread
From: Michael Lukashov @ 2010-02-23 22:55 UTC (permalink / raw
  To: git; +Cc: Michael Lukashov

Steps to reproduce the bug:

	1. Create repository and add more than one remote
	2. Make sure current branch is tracking branch from the remote AND this remote
	   is not last in the list of remotes
	3. 'git pull --all' exits with error message:

You asked to pull from the remote '--all', but did not specify
a branch. Because this is not the default configured remote
for your current branch, you must specify a branch on the command line.

A minimal test case is added that reproduces the problem.
Tested under Windows and Debian GNU/Linux.

Signed-off-by: Michael Lukashov <michael.lukashov@gmail.com>
---
 builtin-fetch.c         |    6 +++++-
 git-pull.sh             |    6 +++++-
 t/t5521-pull-options.sh |   39 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 49 insertions(+), 2 deletions(-)

diff --git a/builtin-fetch.c b/builtin-fetch.c
index d3b9d8a..8e54c5a 100644
--- a/builtin-fetch.c
+++ b/builtin-fetch.c
@@ -784,13 +784,17 @@ static int add_remote_or_group(const char *name, struct string_list *list)
 static int fetch_multiple(struct string_list *list)
 {
 	int i, result = 0;
-	const char *argv[] = { "fetch", NULL, NULL, NULL, NULL, NULL, NULL };
+	const char *argv[] = { "fetch", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL };
 	int argc = 1;
 
 	if (dry_run)
 		argv[argc++] = "--dry-run";
 	if (prune)
 		argv[argc++] = "--prune";
+	if (append)
+		argv[argc++] = "--append";
+	if (update_head_ok)
+		argv[argc++] = "--update-head-ok";
 	if (verbosity >= 2)
 		argv[argc++] = "-v";
 	if (verbosity >= 1)
diff --git a/git-pull.sh b/git-pull.sh
index 38331a8..fcde096 100755
--- a/git-pull.sh
+++ b/git-pull.sh
@@ -214,7 +214,11 @@ test true = "$rebase" && {
 	done
 }
 orig_head=$(git rev-parse -q --verify HEAD)
-git fetch $verbosity --update-head-ok "$@" || exit 1
+if test -e "$GIT_DIR"/FETCH_HEAD
+then
+	rm "$GIT_DIR"/FETCH_HEAD 2>/dev/null
+fi
+git fetch $verbosity --update-head-ok --append "$@" || exit 1
 
 curr_head=$(git rev-parse -q --verify HEAD)
 if test -n "$orig_head" && test "$curr_head" != "$orig_head"
diff --git a/t/t5521-pull-options.sh b/t/t5521-pull-options.sh
index 83e2e8a..2665caa 100755
--- a/t/t5521-pull-options.sh
+++ b/t/t5521-pull-options.sh
@@ -4,6 +4,17 @@ test_description='pull options'
 
 . ./test-lib.sh
 
+setup_repository () {
+	mkdir "$1" && (
+	cd "$1" &&
+	git init &&
+	>file &&
+	git add file &&
+	test_tick &&
+	git commit -m "Initial"
+	)
+}
+
 D=`pwd`
 
 test_expect_success 'setup' '
@@ -57,4 +68,32 @@ test_expect_success 'git pull -q -v' '
 	test -s out
 '
 
+cd "$D"
+
+test_expect_success 'git pull --all' '
+	mkdir pullall &&
+	cd pullall &&
+	setup_repository remote1 &&
+	setup_repository remote2 &&
+	mkdir test &&
+	cd test &&
+	git init &&
+	git remote add remote1 "$D/pullall/remote1" &&
+	git remote add remote2 "$D/pullall/remote2" &&
+	(
+		# "git pull remote1" should print error message
+		# because there is no local branch that is tracking remote repo
+		git pull remote1
+		test $? = 1
+	) &&
+	(
+		# "git pull --all" should not print error message
+		# when current branch is tracking remote repo and that remote
+		# is not last in the list of remotes
+		git checkout -b remote1master remote1/master
+		git pull --all
+		test $? = 0
+	)
+'
+
 test_done
-- 
1.7.0.1706.g00cdbe

^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH] pull: fix 'git pull --all' when current branch is tracking remote that is not last in the list of remotes
  2010-02-23 22:55 [PATCH] pull: fix 'git pull --all' when current branch is tracking remote that is not last in the list of remotes Michael Lukashov
@ 2010-02-23 23:02 ` Junio C Hamano
  2010-02-23 23:44   ` Michael Lukashov
  0 siblings, 1 reply; 7+ messages in thread
From: Junio C Hamano @ 2010-02-23 23:02 UTC (permalink / raw
  To: Michael Lukashov; +Cc: git

Michael Lukashov <michael.lukashov@gmail.com> writes:

> diff --git a/git-pull.sh b/git-pull.sh
> index 38331a8..fcde096 100755
> --- a/git-pull.sh
> +++ b/git-pull.sh
> @@ -214,7 +214,11 @@ test true = "$rebase" && {
>  	done
>  }
>  orig_head=$(git rev-parse -q --verify HEAD)
> -git fetch $verbosity --update-head-ok "$@" || exit 1
> +if test -e "$GIT_DIR"/FETCH_HEAD
> +then
> +	rm "$GIT_DIR"/FETCH_HEAD 2>/dev/null
> +fi

When is it sane to ignore an error from this "rm", especially after you
made sure that it exists?

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH] pull: fix 'git pull --all' when current branch is  tracking remote that is not last in the list of remotes
  2010-02-23 23:02 ` Junio C Hamano
@ 2010-02-23 23:44   ` Michael Lukashov
  2010-02-24  0:22     ` Junio C Hamano
  0 siblings, 1 reply; 7+ messages in thread
From: Michael Lukashov @ 2010-02-23 23:44 UTC (permalink / raw
  To: Junio C Hamano; +Cc: git

On Wed, Feb 24, 2010 at 2:02 AM, Junio C Hamano <gitster@pobox.com> wrote:
> Michael Lukashov <michael.lukashov@gmail.com> writes:
>
>> diff --git a/git-pull.sh b/git-pull.sh
>> index 38331a8..fcde096 100755
>> --- a/git-pull.sh
>> +++ b/git-pull.sh
>> @@ -214,7 +214,11 @@ test true = "$rebase" && {
>>       done
>>  }
>>  orig_head=$(git rev-parse -q --verify HEAD)
>> -git fetch $verbosity --update-head-ok "$@" || exit 1
>> +if test -e "$GIT_DIR"/FETCH_HEAD
>> +then
>> +     rm "$GIT_DIR"/FETCH_HEAD 2>/dev/null
>> +fi
>
> When is it sane to ignore an error from this "rm", especially after you
> made sure that it exists?
>

The file "$GIT_DIR"/FETCH_HEAD is rewritten
in subsequent call to 'git fetch', thus it is safe to ignore all errors.

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH] pull: fix 'git pull --all' when current branch is  tracking remote that is not last in the list of remotes
  2010-02-23 23:44   ` Michael Lukashov
@ 2010-02-24  0:22     ` Junio C Hamano
  2010-02-24 13:07       ` [PATCH v2] " Michael Lukashov
  0 siblings, 1 reply; 7+ messages in thread
From: Junio C Hamano @ 2010-02-24  0:22 UTC (permalink / raw
  To: Michael Lukashov; +Cc: git

Michael Lukashov <michael.lukashov@gmail.com> writes:

> On Wed, Feb 24, 2010 at 2:02 AM, Junio C Hamano <gitster@pobox.com> wrote:
>> Michael Lukashov <michael.lukashov@gmail.com> writes:
>>
>>> diff --git a/git-pull.sh b/git-pull.sh
>>> index 38331a8..fcde096 100755
>>> --- a/git-pull.sh
>>> +++ b/git-pull.sh
>>> @@ -214,7 +214,11 @@ test true = "$rebase" && {
>>>       done
>>>  }
>>>  orig_head=$(git rev-parse -q --verify HEAD)
>>> -git fetch $verbosity --update-head-ok "$@" || exit 1
>>> +if test -e "$GIT_DIR"/FETCH_HEAD
>>> +then
>>> +     rm "$GIT_DIR"/FETCH_HEAD 2>/dev/null
>>> +fi
>>
>> When is it sane to ignore an error from this "rm", especially after you
>> made sure that it exists?
>
> The file "$GIT_DIR"/FETCH_HEAD is rewritten
> in subsequent call to 'git fetch', thus it is safe to ignore all errors.

You are not answering my question.

You found out that the thing exists, and you want to overwrite it later.
You _need_ that file to either not exist, or at least be empty, because
you will be _appending_ to it, unlike the earlier code.

Now, you expected you would be able to remove it, and that is why you
called "rm".  Suppose that removal has failed for some reason.  The file
stays.  It is not emptied, either.

Why is it sane to ignore that error and let fetch --append to run, as if
it is starting from either non-existing file or an empty one?  You already
diagnosed that the file is in some _funny_ state.  It is not sensible to
continue further at that point, knowing that there is something wrong.

If the new code you introduced were

	rm -f "$GIT_DIR/FETCH_HEAD" || exit

then I would understand it.  But your patch doesn't make sense to me;
neither your "thus it is safe".

^ permalink raw reply	[flat|nested] 7+ messages in thread

* [PATCH v2] pull: fix 'git pull --all' when current branch is tracking remote that is not last in the list of remotes
  2010-02-24  0:22     ` Junio C Hamano
@ 2010-02-24 13:07       ` Michael Lukashov
  2010-02-24 13:33         ` Paolo Bonzini
  2010-02-24 17:05         ` Junio C Hamano
  0 siblings, 2 replies; 7+ messages in thread
From: Michael Lukashov @ 2010-02-24 13:07 UTC (permalink / raw
  To: git; +Cc: Michael Lukashov

Steps to reproduce the bug:

	1. Create repository and add more than one remote
	2. Make sure current branch is tracking branch from the remote AND this remote
	   is not last in the list of remotes
	3. 'git pull --all' exits with error message:

		You asked to pull from the remote '--all', but did not specify
		a branch. Because this is not the default configured remote
		for your current branch, you must specify a branch on the command line.

After 'git pull --all' you need to run 'git pull' to update current branch.
This is annoying.

After this patch, 'git pull --all' does what it should do - fetches all changes
from all remotes and then updates current branch, if there were changes.

A minimal test case is added that reproduces the problem.
Tested under Windows and Debian GNU/Linux.

Signed-off-by: Michael Lukashov <michael.lukashov@gmail.com>
---
 builtin-fetch.c         |    6 +++++-
 git-pull.sh             |    6 +++++-
 t/t5521-pull-options.sh |   39 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 49 insertions(+), 2 deletions(-)

diff --git a/builtin-fetch.c b/builtin-fetch.c
index d3b9d8a..8e54c5a 100644
--- a/builtin-fetch.c
+++ b/builtin-fetch.c
@@ -784,13 +784,17 @@ static int add_remote_or_group(const char *name, struct string_list *list)
 static int fetch_multiple(struct string_list *list)
 {
 	int i, result = 0;
-	const char *argv[] = { "fetch", NULL, NULL, NULL, NULL, NULL, NULL };
+	const char *argv[] = { "fetch", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL };
 	int argc = 1;
 
 	if (dry_run)
 		argv[argc++] = "--dry-run";
 	if (prune)
 		argv[argc++] = "--prune";
+	if (append)
+		argv[argc++] = "--append";
+	if (update_head_ok)
+		argv[argc++] = "--update-head-ok";
 	if (verbosity >= 2)
 		argv[argc++] = "-v";
 	if (verbosity >= 1)
diff --git a/git-pull.sh b/git-pull.sh
index 38331a8..2fbee42 100755
--- a/git-pull.sh
+++ b/git-pull.sh
@@ -214,7 +214,11 @@ test true = "$rebase" && {
 	done
 }
 orig_head=$(git rev-parse -q --verify HEAD)
-git fetch $verbosity --update-head-ok "$@" || exit 1
+if test -e "$GIT_DIR"/FETCH_HEAD
+then
+	rm -f "$GIT_DIR"/FETCH_HEAD || exit
+fi
+git fetch $verbosity --update-head-ok --append "$@" || exit 1
 
 curr_head=$(git rev-parse -q --verify HEAD)
 if test -n "$orig_head" && test "$curr_head" != "$orig_head"
diff --git a/t/t5521-pull-options.sh b/t/t5521-pull-options.sh
index 83e2e8a..2665caa 100755
--- a/t/t5521-pull-options.sh
+++ b/t/t5521-pull-options.sh
@@ -4,6 +4,17 @@ test_description='pull options'
 
 . ./test-lib.sh
 
+setup_repository () {
+	mkdir "$1" && (
+	cd "$1" &&
+	git init &&
+	>file &&
+	git add file &&
+	test_tick &&
+	git commit -m "Initial"
+	)
+}
+
 D=`pwd`
 
 test_expect_success 'setup' '
@@ -57,4 +68,32 @@ test_expect_success 'git pull -q -v' '
 	test -s out
 '
 
+cd "$D"
+
+test_expect_success 'git pull --all' '
+	mkdir pullall &&
+	cd pullall &&
+	setup_repository remote1 &&
+	setup_repository remote2 &&
+	mkdir test &&
+	cd test &&
+	git init &&
+	git remote add remote1 "$D/pullall/remote1" &&
+	git remote add remote2 "$D/pullall/remote2" &&
+	(
+		# "git pull remote1" should print error message
+		# because there is no local branch that is tracking remote repo
+		git pull remote1
+		test $? = 1
+	) &&
+	(
+		# "git pull --all" should not print error message
+		# when current branch is tracking remote repo and that remote
+		# is not last in the list of remotes
+		git checkout -b remote1master remote1/master
+		git pull --all
+		test $? = 0
+	)
+'
+
 test_done
-- 
1.7.0.1706.g00cdbe

^ permalink raw reply related	[flat|nested] 7+ messages in thread

* Re: [PATCH v2] pull: fix 'git pull --all' when current branch is tracking remote that is not last in the list of remotes
  2010-02-24 13:07       ` [PATCH v2] " Michael Lukashov
@ 2010-02-24 13:33         ` Paolo Bonzini
  2010-02-24 17:05         ` Junio C Hamano
  1 sibling, 0 replies; 7+ messages in thread
From: Paolo Bonzini @ 2010-02-24 13:33 UTC (permalink / raw
  To: Michael Lukashov; +Cc: git

On 02/24/2010 02:07 PM, Michael Lukashov wrote:
> +if test -e "$GIT_DIR"/FETCH_HEAD
> +then
> +	rm -f "$GIT_DIR"/FETCH_HEAD || exit
> +fi

You do not need the if.

Paolo

^ permalink raw reply	[flat|nested] 7+ messages in thread

* Re: [PATCH v2] pull: fix 'git pull --all' when current branch is tracking remote that is not last in the list of remotes
  2010-02-24 13:07       ` [PATCH v2] " Michael Lukashov
  2010-02-24 13:33         ` Paolo Bonzini
@ 2010-02-24 17:05         ` Junio C Hamano
  1 sibling, 0 replies; 7+ messages in thread
From: Junio C Hamano @ 2010-02-24 17:05 UTC (permalink / raw
  To: Michael Lukashov; +Cc: git

Michael Lukashov <michael.lukashov@gmail.com> writes:

> Steps to reproduce the bug:
>
> 	1. Create repository and add more than one remote
> 	2. Make sure current branch is tracking branch from the remote AND this remote
> 	   is not last in the list of remotes
> 	3. 'git pull --all' exits with error message:
>
> 		You asked to pull from the remote '--all', but did not specify
> 		a branch. Because this is not the default configured remote
> 		for your current branch, you must specify a branch on the command line.
>
> After 'git pull --all' you need to run 'git pull' to update current branch.
> This is annoying.

I started to rewrite the commit log message to present the backstory and
the assumptions existing code makes, _why_ things break, and then explain
what the proposed solution is and why it is the right fix to the problem
(by the way, presenting the log message this way helps you to think things
through).  And I had to stop immediately after the description of _why_
things break:

    "git fetch" learned "--all" option and it has become tempting for
    users to say "git pull --all", even though it may not be absolutely
    necessary to pull from many remotes that are not involved in the merge
    about to happen to the current branch.

    "git fetch --all" however clears the list of fetched branches every
    time it contacts a different remote.  Unless the current branch is
    configured to merge with a branch from a remote that happens to be the
    last in the list of remotes "fetch --all" contacts with, "git pull
    --all" will not be able to find the branch it should be merging with.

Notice that presented this way, it becomes clear that it not a bug in
"pull --all" at all. "fetch --all" should be doing the clearing of
FETCH_HEAD at the very beginning (unless --append is given from the
command line, in which case it should just use FETCH_HEAD as-is), and then
run in the --append mode even when --append is not given.

So I think you identified a valid issue to address, but the patch is
solving it in a wrong way.  The commit log message I started above would
be concluded with something like the following explanation of what the
proposed solution is and why it is the right fix:

    Make "fetch --all" to clear FETCH_HEAD (unless --append is given) and
    then append the list of branches fetched to it (even when --apend is
    not given).  That way, "pull --all" will be able to find the data for
    the branch being merged in FETCH_HEAD no matter where the remote
    appears in the list of remotes to be contacted by "git fetch".

and the patch would touch fetch, not pull.

^ permalink raw reply	[flat|nested] 7+ messages in thread

end of thread, other threads:[~2010-02-24 17:05 UTC | newest]

Thread overview: 7+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2010-02-23 22:55 [PATCH] pull: fix 'git pull --all' when current branch is tracking remote that is not last in the list of remotes Michael Lukashov
2010-02-23 23:02 ` Junio C Hamano
2010-02-23 23:44   ` Michael Lukashov
2010-02-24  0:22     ` Junio C Hamano
2010-02-24 13:07       ` [PATCH v2] " Michael Lukashov
2010-02-24 13:33         ` Paolo Bonzini
2010-02-24 17:05         ` Junio C Hamano

Code repositories for project(s) associated with this public inbox

	https://80x24.org/mirrors/git.git

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).