[PATCH 0/5] Create 'feature.*' config area and some centralized config parsing

[PATCH 0/5] Create 'feature.*' config area and some centralized config parsing

[Derrick Stolee via GitGitGadget]

This is a brand-new thread to replace ds/early-access. The discussion on
that thread was very helpful [1].

With this in mind, I propose instead a set of "feature.*" config settings
that form groups of "community recommended" settings (with some caveats). In
the space below, I'll list a set of possible feature names and the implied
config options.

First, the main two categories we've discussed so far: many commits and many
files. These two feature sets are for when your repo is large in one of
these dimensions. Perhaps there are other settings to include in these?

feature.manyFiles:
    index.version = 4
    core.untrackedCache = true

feature.manyCommits:
    core.commitGraph = true
    gc.writeCommitGraph = true
    (future: fetch.writeSplitCommitGraph = true)

Note: the fetch.writeSplitCommitGraph does not exist yet, but could be
introduced in a later release to write a new commit-graph (with --split) on
fetch.

The other category that has been discussed already is that of "experimental
features that we generally think are helpful but change behavior slightly in
some cases".

feature.experimental:
    pack.useSparse = true
    merge.directoryRenames = true
    fetch.negotiationAlgorithm = skipping

Specifically, this setting is for config values we are not sure will ever be
on by default, but additional testing is needed to be sure. This is
different than a possible 'feature.preview' setting that would include
config settings that we are committed to updating the defaults in a future
release. There are many ways we can take this idea in the future (including
more additions to these categories).

Thanks, -Stolee

[1] https://public-inbox.org/git/pull.254.git.gitgitgadget@gmail.com/

Derrick Stolee (5):
  repo-settings: consolidate some config settings
  repo-settings: add feature.manyCommits setting
  repo-settings: parse core.untrackedCache
  repo-settings: create feature.manyFiles setting
  repo-settings: create feature.experimental setting

 Documentation/config.txt             |   2 +
 Documentation/config/core.txt        |   7 +-
 Documentation/config/feature.txt     |  42 +++++++++++
 Documentation/config/fetch.txt       |   3 +-
 Documentation/config/gc.txt          |   4 +-
 Documentation/config/index.txt       |   1 +
 Documentation/config/merge.txt       |   3 +-
 Documentation/config/pack.txt        |   3 +-
 Makefile                             |   1 +
 builtin/am.c                         |   4 +-
 builtin/gc.c                         |  13 ++--
 builtin/pack-objects.c               |   9 ++-
 builtin/update-index.c               |   7 +-
 commit-graph.c                       |   7 +-
 config.c                             |  24 ------
 fetch-negotiator.c                   |  26 ++++---
 fetch-negotiator.h                   |   5 +-
 fetch-pack.c                         |  11 ++-
 merge-recursive.c                    |  32 ++++----
 merge-recursive.h                    |   1 -
 read-cache.c                         |  31 ++++----
 repo-settings.c                      | 108 +++++++++++++++++++++++++++
 repo-settings.h                      |  29 +++++++
 repository.h                         |   3 +
 t/t1600-index.sh                     |  31 ++++++--
 t/t5552-skipping-fetch-negotiator.sh |  23 ------
 26 files changed, 303 insertions(+), 127 deletions(-)
 create mode 100644 Documentation/config/feature.txt
 create mode 100644 repo-settings.c
 create mode 100644 repo-settings.h


base-commit: 9c9b961d7eb15fb583a2a812088713a68a85f1c0
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-292%2Fderrickstolee%2Frepo-settings%2Fhead-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-292/derrickstolee/repo-settings/head-v1
Pull-Request: https://github.com/gitgitgadget/git/pull/292
-- 
gitgitgadget

[PATCH 1/5] repo-settings: consolidate some config settings

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

There are a few important config settings that are not loaded
during git_default_config. These are instead loaded on-demand.

Centralize these config options to a single scan, and store
all of the values in a repo_settings struct. The values for
each setting are initialized as negative to indicate "unset".

This centralization will be particularly important in a later
change to introduce "meta" config settings that change the
defaults for these config settings.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Makefile               |  1 +
 builtin/gc.c           | 13 ++++++-------
 builtin/pack-objects.c |  9 +++++----
 commit-graph.c         |  7 ++++---
 read-cache.c           | 12 +++++++-----
 repo-settings.c        | 44 ++++++++++++++++++++++++++++++++++++++++++
 repo-settings.h        | 15 ++++++++++++++
 repository.h           |  3 +++
 8 files changed, 85 insertions(+), 19 deletions(-)
 create mode 100644 repo-settings.c
 create mode 100644 repo-settings.h

diff --git a/Makefile b/Makefile
index 11ccea4071..032fe9b5f9 100644
--- a/Makefile
+++ b/Makefile
@@ -964,6 +964,7 @@ LIB_OBJS += refspec.o
 LIB_OBJS += ref-filter.o
 LIB_OBJS += remote.o
 LIB_OBJS += replace-object.o
+LIB_OBJS += repo-settings.o
 LIB_OBJS += repository.o
 LIB_OBJS += rerere.o
 LIB_OBJS += resolve-undo.o
diff --git a/builtin/gc.c b/builtin/gc.c
index c18efadda5..243be2907b 100644
--- a/builtin/gc.c
+++ b/builtin/gc.c
@@ -27,6 +27,7 @@
 #include "pack-objects.h"
 #include "blob.h"
 #include "tree.h"
+#include "repo-settings.h"
 
 #define FAILED_RUN "failed to run %s"
 
@@ -41,7 +42,6 @@ static int aggressive_depth = 50;
 static int aggressive_window = 250;
 static int gc_auto_threshold = 6700;
 static int gc_auto_pack_limit = 50;
-static int gc_write_commit_graph;
 static int detach_auto = 1;
 static timestamp_t gc_log_expire_time;
 static const char *gc_log_expire = "1.day.ago";
@@ -148,7 +148,6 @@ static void gc_config(void)
 	git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 	git_config_get_int("gc.auto", &gc_auto_threshold);
 	git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
-	git_config_get_bool("gc.writecommitgraph", &gc_write_commit_graph);
 	git_config_get_bool("gc.autodetach", &detach_auto);
 	git_config_get_expiry("gc.pruneexpire", &prune_expire);
 	git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
@@ -685,11 +684,11 @@ int cmd_gc(int argc, const char **argv, const char *prefix)
 		clean_pack_garbage();
 	}
 
-	if (gc_write_commit_graph &&
-	    write_commit_graph_reachable(get_object_directory(),
-					 !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
-					 NULL))
-		return 1;
+	prepare_repo_settings(the_repository);
+	if (the_repository->settings->gc_write_commit_graph == 1)
+		write_commit_graph_reachable(get_object_directory(),
+					     !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
+					     NULL);
 
 	if (auto_gc && too_many_loose_objects())
 		warning(_("There are too many unreachable loose objects; "
diff --git a/builtin/pack-objects.c b/builtin/pack-objects.c
index 267c562b1f..223038b309 100644
--- a/builtin/pack-objects.c
+++ b/builtin/pack-objects.c
@@ -34,6 +34,7 @@
 #include "dir.h"
 #include "midx.h"
 #include "trace2.h"
+#include "repo-settings.h"
 
 #define IN_PACK(obj) oe_in_pack(&to_pack, obj)
 #define SIZE(obj) oe_size(&to_pack, obj)
@@ -2709,10 +2710,6 @@ static int git_pack_config(const char *k, const char *v, void *cb)
 		use_bitmap_index_default = git_config_bool(k, v);
 		return 0;
 	}
-	if (!strcmp(k, "pack.usesparse")) {
-		sparse = git_config_bool(k, v);
-		return 0;
-	}
 	if (!strcmp(k, "pack.threads")) {
 		delta_search_threads = git_config_int(k, v);
 		if (delta_search_threads < 0)
@@ -3332,6 +3329,10 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 	read_replace_refs = 0;
 
 	sparse = git_env_bool("GIT_TEST_PACK_SPARSE", 0);
+	prepare_repo_settings(the_repository);
+	if (!sparse && the_repository->settings->pack_use_sparse != -1)
+		sparse = the_repository->settings->pack_use_sparse;
+
 	reset_pack_idx_option(&pack_idx_opts);
 	git_config(git_pack_config, NULL);
 
diff --git a/commit-graph.c b/commit-graph.c
index b3c4de79b6..48e03aa285 100644
--- a/commit-graph.c
+++ b/commit-graph.c
@@ -16,6 +16,7 @@
 #include "hashmap.h"
 #include "replace-object.h"
 #include "progress.h"
+#include "repo-settings.h"
 
 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
@@ -466,7 +467,6 @@ static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 static int prepare_commit_graph(struct repository *r)
 {
 	struct object_directory *odb;
-	int config_value;
 
 	if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
 		die("dying as requested by the '%s' variable on commit-graph load!",
@@ -476,9 +476,10 @@ static int prepare_commit_graph(struct repository *r)
 		return !!r->objects->commit_graph;
 	r->objects->commit_graph_attempted = 1;
 
+	prepare_repo_settings(r);
+
 	if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
-	    (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
-	    !config_value))
+	    r->settings->core_commit_graph != 1)
 		/*
 		 * This repository is not configured to use commit graphs, so
 		 * do not load one. (But report commit_graph_attempted anyway
diff --git a/read-cache.c b/read-cache.c
index c701f7f8b8..ee1aaa8917 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -25,6 +25,7 @@
 #include "fsmonitor.h"
 #include "thread-utils.h"
 #include "progress.h"
+#include "repo-settings.h"
 
 /* Mask for the name length in ce_flags in the on-disk index */
 
@@ -1599,16 +1600,17 @@ struct cache_entry *refresh_cache_entry(struct index_state *istate,
 
 #define INDEX_FORMAT_DEFAULT 3
 
-static unsigned int get_index_format_default(void)
+static unsigned int get_index_format_default(struct repository *r)
 {
 	char *envversion = getenv("GIT_INDEX_VERSION");
 	char *endp;
-	int value;
 	unsigned int version = INDEX_FORMAT_DEFAULT;
 
 	if (!envversion) {
-		if (!git_config_get_int("index.version", &value))
-			version = value;
+		prepare_repo_settings(r);
+
+		if (r->settings->index_version >= 0)
+			version = r->settings->index_version;
 		if (version < INDEX_FORMAT_LB || INDEX_FORMAT_UB < version) {
 			warning(_("index.version set, but the value is invalid.\n"
 				  "Using version %i"), INDEX_FORMAT_DEFAULT);
@@ -2765,7 +2767,7 @@ static int do_write_index(struct index_state *istate, struct tempfile *tempfile,
 	}
 
 	if (!istate->version) {
-		istate->version = get_index_format_default();
+		istate->version = get_index_format_default(the_repository);
 		if (git_env_bool("GIT_TEST_SPLIT_INDEX", 0))
 			init_split_index(istate);
 	}
diff --git a/repo-settings.c b/repo-settings.c
new file mode 100644
index 0000000000..13a9128f62
--- /dev/null
+++ b/repo-settings.c
@@ -0,0 +1,44 @@
+#include "cache.h"
+#include "repository.h"
+#include "config.h"
+#include "repo-settings.h"
+
+static int git_repo_config(const char *key, const char *value, void *cb)
+{
+	struct repo_settings *rs = (struct repo_settings *)cb;
+
+	if (!strcmp(key, "core.commitgraph")) {
+		rs->core_commit_graph = git_config_bool(key, value);
+		return 0;
+	}
+	if (!strcmp(key, "gc.writecommitgraph")) {
+		rs->gc_write_commit_graph = git_config_bool(key, value);
+		return 0;
+	}
+	if (!strcmp(key, "pack.usesparse")) {
+		rs->pack_use_sparse = git_config_bool(key, value);
+		return 0;
+	}
+	if (!strcmp(key, "index.version")) {
+		rs->index_version = git_config_int(key, value);
+		return 0;
+	}
+
+	return 1;
+}
+
+void prepare_repo_settings(struct repository *r)
+{
+	if (r->settings)
+		return;
+
+	r->settings = xmalloc(sizeof(*r->settings));
+
+	/* Defaults */
+	r->settings->core_commit_graph = -1;
+	r->settings->gc_write_commit_graph = -1;
+	r->settings->pack_use_sparse = -1;
+	r->settings->index_version = -1;
+
+	repo_config(r, git_repo_config, r->settings);
+}
diff --git a/repo-settings.h b/repo-settings.h
new file mode 100644
index 0000000000..1151c2193a
--- /dev/null
+++ b/repo-settings.h
@@ -0,0 +1,15 @@
+#ifndef REPO_SETTINGS_H
+#define REPO_SETTINGS_H
+
+struct repo_settings {
+	int core_commit_graph;
+	int gc_write_commit_graph;
+	int pack_use_sparse;
+	int index_version;
+};
+
+struct repository;
+
+void prepare_repo_settings(struct repository *r);
+
+#endif /* REPO_SETTINGS_H */
diff --git a/repository.h b/repository.h
index 4fb6a5885f..352afc9cd8 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct repo_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -72,6 +73,8 @@ struct repository {
 	 */
 	char *submodule_prefix;
 
+	struct repo_settings *settings;
+
 	/* Subsystems */
 	/*
 	 * Repository's config which contains key-value pairs from the usual
-- 
gitgitgadget

[PATCH 2/5] repo-settings: add feature.manyCommits setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

When a repo has many commits, it is helpful to write and read the
commit-graph file. Future changes to Git may include new config
settings that are benefitial in this scenario.

Create the 'feature.manyCommits' config setting that changes the
default values of 'core.commitGraph' and 'gc.writeCommitGraph' to
true.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config.txt         |  2 ++
 Documentation/config/core.txt    |  3 ++-
 Documentation/config/feature.txt | 15 +++++++++++++++
 Documentation/config/gc.txt      |  4 ++--
 repo-settings.c                  |  7 +++++++
 5 files changed, 28 insertions(+), 3 deletions(-)
 create mode 100644 Documentation/config/feature.txt

diff --git a/Documentation/config.txt b/Documentation/config.txt
index e3f5bc3396..77f3b1486b 100644
--- a/Documentation/config.txt
+++ b/Documentation/config.txt
@@ -345,6 +345,8 @@ include::config/difftool.txt[]
 
 include::config/fastimport.txt[]
 
+include::config/feature.txt[]
+
 include::config/fetch.txt[]
 
 include::config/format.txt[]
diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index 75538d27e7..d80162681a 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -577,7 +577,8 @@ the `GIT_NOTES_REF` environment variable.  See linkgit:git-notes[1].
 
 core.commitGraph::
 	If true, then git will read the commit-graph file (if it exists)
-	to parse the graph structure of commits. Defaults to false. See
+	to parse the graph structure of commits. Defaults to false, unless
+	`feature.manyCommits` is enabled. See
 	linkgit:git-commit-graph[1] for more information.
 
 core.useReplaceRefs::
diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
new file mode 100644
index 0000000000..f74314ae90
--- /dev/null
+++ b/Documentation/config/feature.txt
@@ -0,0 +1,15 @@
+feature.*::
+	The config settings that start with `feature.` modify the defaults of
+	a group of other config settings. These groups are created by the Git
+	developer community as recommended defaults and are subject to change.
+	In particular, new config options may be added with different defaults.
+
+feature.manyCommits::
+	Enable config options that optimize for repos with many commits. This
+	setting is recommended for repos with at least 100,000 commits. The
+	new default values are:
++
+* `core.commitGraph=true` enables reading the commit-graph file.
++
+* `gc.writeCommitGraph=true` enables writing the commit-graph file during
+garbage collection.
\ No newline at end of file
diff --git a/Documentation/config/gc.txt b/Documentation/config/gc.txt
index 02b92b18b5..31a5fc4f75 100644
--- a/Documentation/config/gc.txt
+++ b/Documentation/config/gc.txt
@@ -63,8 +63,8 @@ gc.writeCommitGraph::
 	If true, then gc will rewrite the commit-graph file when
 	linkgit:git-gc[1] is run. When using `git gc --auto`
 	the commit-graph will be updated if housekeeping is
-	required. Default is false. See linkgit:git-commit-graph[1]
-	for details.
+	required. Default is false, unless `feature.manyCommits`
+	is enabled. See linkgit:git-commit-graph[1] for details.
 
 gc.logExpiry::
 	If the file gc.log exists, then `git gc --auto` will print
diff --git a/repo-settings.c b/repo-settings.c
index 13a9128f62..f328602fd7 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -3,10 +3,17 @@
 #include "config.h"
 #include "repo-settings.h"
 
+#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
+
 static int git_repo_config(const char *key, const char *value, void *cb)
 {
 	struct repo_settings *rs = (struct repo_settings *)cb;
 
+	if (!strcmp(key, "feature.manycommits")) {
+		UPDATE_DEFAULT(rs->core_commit_graph, 1);
+		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
+		return 0;
+	}
 	if (!strcmp(key, "core.commitgraph")) {
 		rs->core_commit_graph = git_config_bool(key, value);
 		return 0;
-- 
gitgitgadget

[PATCH 3/5] repo-settings: parse core.untrackedCache

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The core.untrackedCache config setting is slightly complicated,
so clarify its use and centralize its parsing into the repo
settings.

The default value is "keep" (returned as -1), which persists the
untracked cache if it exists.

If the value is set as "false" (returned as 0), then remove the
untracked cache if it exists.

If the value is set as "true" (returned as 1), then write the
untracked cache and persist it.

Instead of relying on magic values of -1, 0, and 1, split these
options into bitflags CORE_UNTRACKED_CACHE_KEEP and
CORE_UNTRACKED_CACHE_WRITE. This allows the use of "-1" as a
default value. After parsing the config options, if the value is
unset we can initialize it to CORE_UNTRACKED_CACHE_KEEP.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 builtin/update-index.c |  7 +++++--
 config.c               | 24 ------------------------
 read-cache.c           | 19 +++++++++----------
 repo-settings.c        | 21 +++++++++++++++++++++
 repo-settings.h        |  4 ++++
 5 files changed, 39 insertions(+), 36 deletions(-)

diff --git a/builtin/update-index.c b/builtin/update-index.c
index dff2f4b837..6c26bbae80 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -18,6 +18,7 @@
 #include "dir.h"
 #include "split-index.h"
 #include "fsmonitor.h"
+#include "repo-settings.h"
 
 /*
  * Default to not allowing changes to the list of files. The
@@ -966,6 +967,7 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	struct parse_opt_ctx_t ctx;
 	strbuf_getline_fn getline_fn;
 	int parseopt_state = PARSE_OPT_UNKNOWN;
+	struct repository *r = the_repository;
 	struct option options[] = {
 		OPT_BIT('q', NULL, &refresh_args.flags,
 			N_("continue refresh even when index needs update"),
@@ -1180,11 +1182,12 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 		remove_split_index(&the_index);
 	}
 
+	prepare_repo_settings(r);
 	switch (untracked_cache) {
 	case UC_UNSPECIFIED:
 		break;
 	case UC_DISABLE:
-		if (git_config_get_untracked_cache() == 1)
+		if (r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_WRITE)
 			warning(_("core.untrackedCache is set to true; "
 				  "remove or change it, if you really want to "
 				  "disable the untracked cache"));
@@ -1196,7 +1199,7 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 		return !test_if_untracked_cache_is_supported();
 	case UC_ENABLE:
 	case UC_FORCE:
-		if (git_config_get_untracked_cache() == 0)
+		if (r->settings->core_untracked_cache == 0)
 			warning(_("core.untrackedCache is set to false; "
 				  "remove or change it, if you really want to "
 				  "enable the untracked cache"));
diff --git a/config.c b/config.c
index faa57e436c..3241dbc54d 100644
--- a/config.c
+++ b/config.c
@@ -2277,30 +2277,6 @@ int git_config_get_expiry_in_days(const char *key, timestamp_t *expiry, timestam
 	return -1; /* thing exists but cannot be parsed */
 }
 
-int git_config_get_untracked_cache(void)
-{
-	int val = -1;
-	const char *v;
-
-	/* Hack for test programs like test-dump-untracked-cache */
-	if (ignore_untracked_cache_config)
-		return -1;
-
-	if (!git_config_get_maybe_bool("core.untrackedcache", &val))
-		return val;
-
-	if (!git_config_get_value("core.untrackedcache", &v)) {
-		if (!strcasecmp(v, "keep"))
-			return -1;
-
-		error(_("unknown core.untrackedCache value '%s'; "
-			"using 'keep' default value"), v);
-		return -1;
-	}
-
-	return -1; /* default value */
-}
-
 int git_config_get_split_index(void)
 {
 	int val;
diff --git a/read-cache.c b/read-cache.c
index ee1aaa8917..e67e6f6e3e 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -1846,18 +1846,17 @@ static void check_ce_order(struct index_state *istate)
 
 static void tweak_untracked_cache(struct index_state *istate)
 {
-	switch (git_config_get_untracked_cache()) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
+	struct repository *r = the_repository;
+
+	prepare_repo_settings(r);
+
+	if (!(r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_KEEP)) {
 		remove_untracked_cache(istate);
-		break;
-	case 1: /* true */
-		add_untracked_cache(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
+		return;
 	}
+
+	if (r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_WRITE)
+		add_untracked_cache(istate);
 }
 
 static void tweak_split_index(struct index_state *istate)
diff --git a/repo-settings.c b/repo-settings.c
index f328602fd7..807c5a29d6 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -30,6 +30,20 @@ static int git_repo_config(const char *key, const char *value, void *cb)
 		rs->index_version = git_config_int(key, value);
 		return 0;
 	}
+	if (!strcmp(key, "core.untrackedcache")) {
+		int bool_value = git_parse_maybe_bool(value);
+		if (bool_value == 0)
+			rs->core_untracked_cache = 0;
+		else if (bool_value == 1)
+			rs->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP |
+						   CORE_UNTRACKED_CACHE_WRITE;
+		else if (!strcasecmp(value, "keep"))
+			rs->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
+		else
+			error(_("unknown core.untrackedCache value '%s'; "
+				"using 'keep' default value"), value);
+		return 0;
+	}
 
 	return 1;
 }
@@ -46,6 +60,13 @@ void prepare_repo_settings(struct repository *r)
 	r->settings->gc_write_commit_graph = -1;
 	r->settings->pack_use_sparse = -1;
 	r->settings->index_version = -1;
+	r->settings->core_untracked_cache = -1;
 
 	repo_config(r, git_repo_config, r->settings);
+
+	/* Hack for test programs like test-dump-untracked-cache */
+	if (ignore_untracked_cache_config)
+		r->settings->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
+	else
+		UPDATE_DEFAULT(r->settings->core_untracked_cache, CORE_UNTRACKED_CACHE_KEEP);
 }
diff --git a/repo-settings.h b/repo-settings.h
index 1151c2193a..bac9b87d49 100644
--- a/repo-settings.h
+++ b/repo-settings.h
@@ -1,11 +1,15 @@
 #ifndef REPO_SETTINGS_H
 #define REPO_SETTINGS_H
 
+#define CORE_UNTRACKED_CACHE_WRITE (1 << 0)
+#define CORE_UNTRACKED_CACHE_KEEP (1 << 1)
+
 struct repo_settings {
 	int core_commit_graph;
 	int gc_write_commit_graph;
 	int pack_use_sparse;
 	int index_version;
+	int core_untracked_cache;
 };
 
 struct repository;
-- 
gitgitgadget

[PATCH 4/5] repo-settings: create feature.manyFiles setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The feature.manyFiles setting is suitable for repos with many
files in the working directory. By setting index.version=4 and
core.untrackedCache=true, commands such as 'git status' should
improve.

While adding this setting, modify the index version precedence
tests to check how this setting overrides the default for
index.version is unset.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config/core.txt    |  4 +++-
 Documentation/config/feature.txt | 12 +++++++++++-
 Documentation/config/index.txt   |  1 +
 repo-settings.c                  |  6 ++++++
 t/t1600-index.sh                 | 31 ++++++++++++++++++++++++++-----
 5 files changed, 47 insertions(+), 7 deletions(-)

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index d80162681a..7a2a33bc8c 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -86,7 +86,9 @@ core.untrackedCache::
 	it will automatically be removed, if set to `false`. Before
 	setting it to `true`, you should check that mtime is working
 	properly on your system.
-	See linkgit:git-update-index[1]. `keep` by default.
+	See linkgit:git-update-index[1]. `keep` by default, unless
+	`feature.manyFiles` is enabled which sets this setting to
+	`true` by default.
 
 core.checkStat::
 	When missing or is set to `default`, many fields in the stat
diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
index f74314ae90..c2d9ef7473 100644
--- a/Documentation/config/feature.txt
+++ b/Documentation/config/feature.txt
@@ -12,4 +12,14 @@ feature.manyCommits::
 * `core.commitGraph=true` enables reading the commit-graph file.
 +
 * `gc.writeCommitGraph=true` enables writing the commit-graph file during
-garbage collection.
\ No newline at end of file
+garbage collection.
+
+feature.manyFiles::
+	Enable config options that optimize for repos with many files in the
+	working directory. With many files, commands such as `git status` and
+	`git checkout` may be slow and these new defaults improve performance:
++
+* `index.version=4` enables path-prefix compression in the index.
++
+* `core.untrackedCache=true` enables the untracked cache. This setting assumes
+that mtime is working on your machine.
\ No newline at end of file
diff --git a/Documentation/config/index.txt b/Documentation/config/index.txt
index f181503041..7cb50b37e9 100644
--- a/Documentation/config/index.txt
+++ b/Documentation/config/index.txt
@@ -24,3 +24,4 @@ index.threads::
 index.version::
 	Specify the version with which new index files should be
 	initialized.  This does not affect existing repositories.
+	If `feature.manyFiles` is enabled, then the default is 4.
diff --git a/repo-settings.c b/repo-settings.c
index 807c5a29d6..9e4b8e6268 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -14,6 +14,12 @@ static int git_repo_config(const char *key, const char *value, void *cb)
 		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
 		return 0;
 	}
+	if (!strcmp(key, "feature.manyfiles")) {
+		UPDATE_DEFAULT(rs->index_version, 4);
+		UPDATE_DEFAULT(rs->core_untracked_cache,
+			       CORE_UNTRACKED_CACHE_KEEP | CORE_UNTRACKED_CACHE_WRITE);
+		return 0;
+	}
 	if (!strcmp(key, "core.commitgraph")) {
 		rs->core_commit_graph = git_config_bool(key, value);
 		return 0;
diff --git a/t/t1600-index.sh b/t/t1600-index.sh
index 42962ed7d4..c77721b580 100755
--- a/t/t1600-index.sh
+++ b/t/t1600-index.sh
@@ -59,17 +59,38 @@ test_expect_success 'out of bounds index.version issues warning' '
 	)
 '
 
-test_expect_success 'GIT_INDEX_VERSION takes precedence over config' '
+test_index_version () {
+	INDEX_VERSION_CONFIG=$1 &&
+	FEATURE_MANY_FILES=$2 &&
+	ENV_VAR_VERSION=$3
+	EXPECTED_OUTPUT_VERSION=$4 &&
 	(
 		rm -f .git/index &&
-		GIT_INDEX_VERSION=4 &&
-		export GIT_INDEX_VERSION &&
-		git config --add index.version 2 &&
+		rm -f .git/config &&
+		if test "$INDEX_VERSION_CONFIG" -ne 0
+		then
+			git config --add index.version $INDEX_VERSION_CONFIG
+		fi &&
+		git config --add feature.manyFiles $FEATURE_MANY_FILES
+		if test "$ENV_VAR_VERSION" -ne 0
+		then
+			GIT_INDEX_VERSION=$ENV_VAR_VERSION &&
+			export GIT_INDEX_VERSION
+		else
+			unset GIT_INDEX_VERSION
+		fi &&
 		git add a 2>&1 &&
-		echo 4 >expect &&
+		echo $EXPECTED_OUTPUT_VERSION >expect &&
 		test-tool index-version <.git/index >actual &&
 		test_cmp expect actual
 	)
+}
+
+test_expect_success 'index version config precedence' '
+	test_index_version 2 false 4 4 &&
+	test_index_version 2 true 0 2 &&
+	test_index_version 0 true 0 4 &&
+	test_index_version 0 true 2 2
 '
 
 test_done
-- 
gitgitgadget

[PATCH 5/5] repo-settings: create feature.experimental setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The 'feature.experimental' setting includes config options that are
not committed to become defaults, but could use additional testing.

Update the following config settings to take new defaults, and to
use the repo_settings struct if not already using them:

* 'pack.useSparse=true'

* 'merge.directoryRenames=true'

* 'fetch.negotiationAlgorithm=skipping'

In the case of fetch.negotiationAlgorithm, the existing logic
would load the config option only when about to use the setting,
so had a die() statement on an unknown string value. This is
removed as now the config is parsed under prepare_repo_settings().
In general, this die() is probably misplaced and not valuable.
A test was removed that checked this die() statement executed.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config/feature.txt     | 17 +++++++++++++++
 Documentation/config/fetch.txt       |  3 ++-
 Documentation/config/merge.txt       |  3 ++-
 Documentation/config/pack.txt        |  3 ++-
 builtin/am.c                         |  4 +++-
 fetch-negotiator.c                   | 26 +++++++++++-----------
 fetch-negotiator.h                   |  5 +++--
 fetch-pack.c                         | 11 +++++-----
 merge-recursive.c                    | 32 +++++++++++++++-------------
 merge-recursive.h                    |  1 -
 repo-settings.c                      | 32 +++++++++++++++++++++++++++-
 repo-settings.h                      | 12 ++++++++++-
 t/t5552-skipping-fetch-negotiator.sh | 23 --------------------
 13 files changed, 107 insertions(+), 65 deletions(-)

diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
index c2d9ef7473..287621b1e8 100644
--- a/Documentation/config/feature.txt
+++ b/Documentation/config/feature.txt
@@ -4,6 +4,23 @@ feature.*::
 	developer community as recommended defaults and are subject to change.
 	In particular, new config options may be added with different defaults.
 
+feature.experimental::
+	Enable config options that are new to Git, and are being considered for
+	future defaults. Config settings included here may be added or removed
+	with each release, including minor version updates. These settings may
+	have unintended interactions since they are so new. Please enable this
+	setting if you are interested in providing feedback on experimental
+	features. The new default values are:
++
+* `pack.useSparse=true` uses a new algorithm when constructing a pack-file
+which can improve `git push` performance in repos with many files.
++
+* `merge.directoryRenames=true` uses a new algorithm for detecting renames by
+using entire directories at a time instead of single files at a time.
++
+* `fetch.negotiationAlgorithm=skipping` may improve fetch negotiation times by
+skipping more commits at a time, reducing the number of round trips.
+
 feature.manyCommits::
 	Enable config options that optimize for repos with many commits. This
 	setting is recommended for repos with at least 100,000 commits. The
diff --git a/Documentation/config/fetch.txt b/Documentation/config/fetch.txt
index ba890b5884..d402110638 100644
--- a/Documentation/config/fetch.txt
+++ b/Documentation/config/fetch.txt
@@ -59,7 +59,8 @@ fetch.negotiationAlgorithm::
 	effort to converge faster, but may result in a larger-than-necessary
 	packfile; The default is "default" which instructs Git to use the default algorithm
 	that never skips commits (unless the server has acknowledged it or one
-	of its descendants).
+	of its descendants). If `feature.experimental` is enabled, then this
+	setting defaults to "skipping".
 	Unknown values will cause 'git fetch' to error out.
 +
 See also the `--negotiation-tip` option for linkgit:git-fetch[1].
diff --git a/Documentation/config/merge.txt b/Documentation/config/merge.txt
index 6a313937f8..e8def2d63c 100644
--- a/Documentation/config/merge.txt
+++ b/Documentation/config/merge.txt
@@ -54,7 +54,8 @@ merge.directoryRenames::
 	moved into the new directory.  If set to "conflict", a conflict
 	will be reported for such paths.  If merge.renames is false,
 	merge.directoryRenames is ignored and treated as false.  Defaults
-	to "conflict".
+	to "conflict" unless `feature.experimental` is enabled and the
+	default is "true".
 
 merge.renormalize::
 	Tell Git that canonical representation of files in the
diff --git a/Documentation/config/pack.txt b/Documentation/config/pack.txt
index 9cdcfa7324..1d66f0c992 100644
--- a/Documentation/config/pack.txt
+++ b/Documentation/config/pack.txt
@@ -112,7 +112,8 @@ pack.useSparse::
 	objects. This can have significant performance benefits when
 	computing a pack to send a small change. However, it is possible
 	that extra objects are added to the pack-file if the included
-	commits contain certain types of direct renames.
+	commits contain certain types of direct renames. Default is `false`
+	unless `feature.experimental` is enabled.
 
 pack.writeBitmaps (deprecated)::
 	This is a deprecated synonym for `repack.writeBitmaps`.
diff --git a/builtin/am.c b/builtin/am.c
index 1aea657a7f..5f3ee3e6cd 100644
--- a/builtin/am.c
+++ b/builtin/am.c
@@ -34,6 +34,7 @@
 #include "string-list.h"
 #include "packfile.h"
 #include "repository.h"
+#include "repo-settings.h"
 
 /**
  * Returns the length of the first line of msg.
@@ -1538,7 +1539,8 @@ static int fall_back_threeway(const struct am_state *state, const char *index_pa
 	o.branch1 = "HEAD";
 	their_tree_name = xstrfmt("%.*s", linelen(state->msg), state->msg);
 	o.branch2 = their_tree_name;
-	o.detect_directory_renames = 0;
+	prepare_repo_settings(the_repository);
+	the_repository->settings->merge_directory_renames = 0;
 
 	if (state->quiet)
 		o.verbosity = 0;
diff --git a/fetch-negotiator.c b/fetch-negotiator.c
index d6d685cba0..e2990cf46e 100644
--- a/fetch-negotiator.c
+++ b/fetch-negotiator.c
@@ -2,19 +2,21 @@
 #include "fetch-negotiator.h"
 #include "negotiator/default.h"
 #include "negotiator/skipping.h"
+#include "repository.h"
+#include "repo-settings.h"
 
-void fetch_negotiator_init(struct fetch_negotiator *negotiator,
-			   const char *algorithm)
+void fetch_negotiator_init(struct repository *r,
+			   struct fetch_negotiator *negotiator)
 {
-	if (algorithm) {
-		if (!strcmp(algorithm, "skipping")) {
-			skipping_negotiator_init(negotiator);
-			return;
-		} else if (!strcmp(algorithm, "default")) {
-			/* Fall through to default initialization */
-		} else {
-			die("unknown fetch negotiation algorithm '%s'", algorithm);
-		}
+	prepare_repo_settings(r);
+	switch(r->settings->fetch_negotiation_algorithm) {
+	case FETCH_NEGOTIATION_SKIPPING:
+		skipping_negotiator_init(negotiator);
+		return;
+
+	case FETCH_NEGOTIATION_DEFAULT:
+	default:
+		default_negotiator_init(negotiator);
+		return;
 	}
-	default_negotiator_init(negotiator);
 }
diff --git a/fetch-negotiator.h b/fetch-negotiator.h
index 9e3967ce66..ea78868504 100644
--- a/fetch-negotiator.h
+++ b/fetch-negotiator.h
@@ -2,6 +2,7 @@
 #define FETCH_NEGOTIATOR_H
 
 struct commit;
+struct repository;
 
 /*
  * An object that supplies the information needed to negotiate the contents of
@@ -52,7 +53,7 @@ struct fetch_negotiator {
 	void *data;
 };
 
-void fetch_negotiator_init(struct fetch_negotiator *negotiator,
-			   const char *algorithm);
+void fetch_negotiator_init(struct repository *r,
+			   struct fetch_negotiator *negotiator);
 
 #endif
diff --git a/fetch-pack.c b/fetch-pack.c
index 65be043f2a..d81f47c07b 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -36,7 +36,6 @@ static int agent_supported;
 static int server_supports_filtering;
 static struct lock_file shallow_lock;
 static const char *alternate_shallow_file;
-static char *negotiation_algorithm;
 static struct strbuf fsck_msg_types = STRBUF_INIT;
 
 /* Remember to update object flag allocation in object.h */
@@ -892,12 +891,13 @@ static struct ref *do_fetch_pack(struct fetch_pack_args *args,
 				 struct shallow_info *si,
 				 char **pack_lockfile)
 {
+	struct repository *r = the_repository;
 	struct ref *ref = copy_ref_list(orig_ref);
 	struct object_id oid;
 	const char *agent_feature;
 	int agent_len;
 	struct fetch_negotiator negotiator;
-	fetch_negotiator_init(&negotiator, negotiation_algorithm);
+	fetch_negotiator_init(r, &negotiator);
 
 	sort_ref_list(&ref, ref_compare_name);
 	QSORT(sought, nr_sought, cmp_ref_by_name);
@@ -911,7 +911,7 @@ static struct ref *do_fetch_pack(struct fetch_pack_args *args,
 
 	if (server_supports("shallow"))
 		print_verbose(args, _("Server supports %s"), "shallow");
-	else if (args->depth > 0 || is_repository_shallow(the_repository))
+	else if (args->depth > 0 || is_repository_shallow(r))
 		die(_("Server does not support shallow clients"));
 	if (args->depth > 0 || args->deepen_since || args->deepen_not)
 		args->deepen = 1;
@@ -1379,6 +1379,7 @@ static struct ref *do_fetch_pack_v2(struct fetch_pack_args *args,
 				    struct shallow_info *si,
 				    char **pack_lockfile)
 {
+	struct repository *r = the_repository;
 	struct ref *ref = copy_ref_list(orig_ref);
 	enum fetch_state state = FETCH_CHECK_LOCAL;
 	struct oidset common = OIDSET_INIT;
@@ -1386,7 +1387,7 @@ static struct ref *do_fetch_pack_v2(struct fetch_pack_args *args,
 	int in_vain = 0;
 	int haves_to_send = INITIAL_FLUSH;
 	struct fetch_negotiator negotiator;
-	fetch_negotiator_init(&negotiator, negotiation_algorithm);
+	fetch_negotiator_init(r, &negotiator);
 	packet_reader_init(&reader, fd[0], NULL, 0,
 			   PACKET_READ_CHOMP_NEWLINE |
 			   PACKET_READ_DIE_ON_ERR_PACKET);
@@ -1505,8 +1506,6 @@ static void fetch_pack_config(void)
 	git_config_get_bool("repack.usedeltabaseoffset", &prefer_ofs_delta);
 	git_config_get_bool("fetch.fsckobjects", &fetch_fsck_objects);
 	git_config_get_bool("transfer.fsckobjects", &transfer_fsck_objects);
-	git_config_get_string("fetch.negotiationalgorithm",
-			      &negotiation_algorithm);
 
 	git_config(fetch_pack_config_cb, NULL);
 }
diff --git a/merge-recursive.c b/merge-recursive.c
index 12300131fc..162d5a4753 100644
--- a/merge-recursive.c
+++ b/merge-recursive.c
@@ -28,6 +28,7 @@
 #include "submodule.h"
 #include "revision.h"
 #include "commit-reach.h"
+#include "repo-settings.h"
 
 struct path_hashmap_entry {
 	struct hashmap_entry e;
@@ -1375,10 +1376,14 @@ static int handle_rename_via_dir(struct merge_options *opt,
 	 * there is no content merge to do; just move the file into the
 	 * desired final location.
 	 */
+	struct repository *r = the_repository;
 	const struct rename *ren = ci->ren1;
 	const struct diff_filespec *dest = ren->pair->two;
 	char *file_path = dest->path;
-	int mark_conflicted = (opt->detect_directory_renames == 1);
+	int mark_conflicted;
+
+	prepare_repo_settings(r);
+	mark_conflicted = (r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_CONFLICT);
 	assert(ren->dir_rename_original_dest);
 
 	if (!opt->call_depth && would_lose_untracked(opt, dest->path)) {
@@ -2850,6 +2855,7 @@ static int detect_and_process_renames(struct merge_options *opt,
 				      struct string_list *entries,
 				      struct rename_info *ri)
 {
+	struct repository *r = the_repository;
 	struct diff_queue_struct *head_pairs, *merge_pairs;
 	struct hashmap *dir_re_head, *dir_re_merge;
 	int clean = 1;
@@ -2863,7 +2869,8 @@ static int detect_and_process_renames(struct merge_options *opt,
 	head_pairs = get_diffpairs(opt, common, head);
 	merge_pairs = get_diffpairs(opt, common, merge);
 
-	if (opt->detect_directory_renames) {
+	prepare_repo_settings(r);
+	if (r->settings->merge_directory_renames) {
 		dir_re_head = get_directory_renames(head_pairs);
 		dir_re_merge = get_directory_renames(merge_pairs);
 
@@ -3112,6 +3119,7 @@ static int handle_rename_normal(struct merge_options *opt,
 				const struct diff_filespec *b,
 				struct rename_conflict_info *ci)
 {
+	struct repository *r = the_repository;
 	struct rename *ren = ci->ren1;
 	struct merge_file_info mfi;
 	int clean;
@@ -3121,7 +3129,9 @@ static int handle_rename_normal(struct merge_options *opt,
 	clean = handle_content_merge(&mfi, opt, path, was_dirty(opt, path),
 				     o, a, b, ci);
 
-	if (clean && opt->detect_directory_renames == 1 &&
+	prepare_repo_settings(r);
+	if (clean &&
+	    r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_CONFLICT &&
 	    ren->dir_rename_original_dest) {
 		if (update_stages(opt, path,
 				  NULL,
@@ -3155,6 +3165,7 @@ static void dir_rename_warning(const char *msg,
 static int warn_about_dir_renamed_entries(struct merge_options *opt,
 					  struct rename *ren)
 {
+	struct repository *r = the_repository;
 	const char *msg;
 	int clean = 1, is_add;
 
@@ -3166,12 +3177,13 @@ static int warn_about_dir_renamed_entries(struct merge_options *opt,
 		return clean;
 
 	/* Sanity checks */
-	assert(opt->detect_directory_renames > 0);
+	prepare_repo_settings(r);
+	assert(r->settings->merge_directory_renames > 0);
 	assert(ren->dir_rename_original_type == 'A' ||
 	       ren->dir_rename_original_type == 'R');
 
 	/* Check whether to treat directory renames as a conflict */
-	clean = (opt->detect_directory_renames == 2);
+	clean = (r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_TRUE);
 
 	is_add = (ren->dir_rename_original_type == 'A');
 	if (ren->dir_rename_original_type == 'A' && clean) {
@@ -3662,15 +3674,6 @@ static void merge_recursive_config(struct merge_options *opt)
 		opt->merge_detect_rename = git_config_rename("merge.renames", value);
 		free(value);
 	}
-	if (!git_config_get_string("merge.directoryrenames", &value)) {
-		int boolval = git_parse_maybe_bool(value);
-		if (0 <= boolval) {
-			opt->detect_directory_renames = boolval ? 2 : 0;
-		} else if (!strcasecmp(value, "conflict")) {
-			opt->detect_directory_renames = 1;
-		} /* avoid erroring on values from future versions of git */
-		free(value);
-	}
 	git_config(git_xmerge_config, NULL);
 }
 
@@ -3687,7 +3690,6 @@ void init_merge_options(struct merge_options *opt,
 	opt->renormalize = 0;
 	opt->diff_detect_rename = -1;
 	opt->merge_detect_rename = -1;
-	opt->detect_directory_renames = 1;
 	merge_recursive_config(opt);
 	merge_verbosity = getenv("GIT_MERGE_VERBOSITY");
 	if (merge_verbosity)
diff --git a/merge-recursive.h b/merge-recursive.h
index c2b7bb65c6..b8eba244ee 100644
--- a/merge-recursive.h
+++ b/merge-recursive.h
@@ -22,7 +22,6 @@ struct merge_options {
 	unsigned renormalize : 1;
 	long xdl_opts;
 	int verbosity;
-	int detect_directory_renames;
 	int diff_detect_rename;
 	int merge_detect_rename;
 	int diff_rename_limit;
diff --git a/repo-settings.c b/repo-settings.c
index 9e4b8e6268..5e9249c437 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -9,6 +9,12 @@ static int git_repo_config(const char *key, const char *value, void *cb)
 {
 	struct repo_settings *rs = (struct repo_settings *)cb;
 
+	if (!strcmp(key, "feature.experimental")) {
+		UPDATE_DEFAULT(rs->pack_use_sparse, 1);
+		UPDATE_DEFAULT(rs->merge_directory_renames, MERGE_DIRECTORY_RENAMES_TRUE);
+		UPDATE_DEFAULT(rs->fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
+		return 0;
+	}
 	if (!strcmp(key, "feature.manycommits")) {
 		UPDATE_DEFAULT(rs->core_commit_graph, 1);
 		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
@@ -50,6 +56,23 @@ static int git_repo_config(const char *key, const char *value, void *cb)
 				"using 'keep' default value"), value);
 		return 0;
 	}
+	if (!strcmp(key, "merge.directoryrenames")) {
+		int bool_value = git_parse_maybe_bool(value);
+		if (0 <= bool_value) {
+			rs->merge_directory_renames = bool_value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
+		} else if (!strcasecmp(value, "conflict")) {
+			rs->merge_directory_renames = MERGE_DIRECTORY_RENAMES_CONFLICT;
+		}
+		return 0;
+	}
+	if (!strcmp(key, "fetch.negotiationalgorithm"))	{
+		if (!strcasecmp(value, "skipping")) {
+			rs->fetch_negotiation_algorithm = FETCH_NEGOTIATION_SKIPPING;
+		} else {
+			rs->fetch_negotiation_algorithm = FETCH_NEGOTIATION_DEFAULT;
+		}
+		return 0;
+	}
 
 	return 1;
 }
@@ -64,10 +87,14 @@ void prepare_repo_settings(struct repository *r)
 	/* Defaults */
 	r->settings->core_commit_graph = -1;
 	r->settings->gc_write_commit_graph = -1;
-	r->settings->pack_use_sparse = -1;
+
 	r->settings->index_version = -1;
 	r->settings->core_untracked_cache = -1;
 
+	r->settings->pack_use_sparse = -1;
+	r->settings->merge_directory_renames = -1;
+	r->settings->fetch_negotiation_algorithm = -1;
+
 	repo_config(r, git_repo_config, r->settings);
 
 	/* Hack for test programs like test-dump-untracked-cache */
@@ -75,4 +102,7 @@ void prepare_repo_settings(struct repository *r)
 		r->settings->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
 	else
 		UPDATE_DEFAULT(r->settings->core_untracked_cache, CORE_UNTRACKED_CACHE_KEEP);
+
+	UPDATE_DEFAULT(r->settings->merge_directory_renames, MERGE_DIRECTORY_RENAMES_CONFLICT);
+	UPDATE_DEFAULT(r->settings->fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
 }
diff --git a/repo-settings.h b/repo-settings.h
index bac9b87d49..cecf7d0028 100644
--- a/repo-settings.h
+++ b/repo-settings.h
@@ -4,12 +4,22 @@
 #define CORE_UNTRACKED_CACHE_WRITE (1 << 0)
 #define CORE_UNTRACKED_CACHE_KEEP (1 << 1)
 
+#define MERGE_DIRECTORY_RENAMES_CONFLICT 1
+#define MERGE_DIRECTORY_RENAMES_TRUE 2
+
+#define FETCH_NEGOTIATION_DEFAULT 1
+#define FETCH_NEGOTIATION_SKIPPING 2
+
 struct repo_settings {
 	int core_commit_graph;
 	int gc_write_commit_graph;
-	int pack_use_sparse;
+
 	int index_version;
 	int core_untracked_cache;
+
+	int pack_use_sparse;
+	int merge_directory_renames;
+	int fetch_negotiation_algorithm;
 };
 
 struct repository;
diff --git a/t/t5552-skipping-fetch-negotiator.sh b/t/t5552-skipping-fetch-negotiator.sh
index 8a14be51a1..f70cbcc9ca 100755
--- a/t/t5552-skipping-fetch-negotiator.sh
+++ b/t/t5552-skipping-fetch-negotiator.sh
@@ -60,29 +60,6 @@ test_expect_success 'commits with no parents are sent regardless of skip distanc
 	have_not_sent c6 c4 c3
 '
 
-test_expect_success 'unknown fetch.negotiationAlgorithm values error out' '
-	rm -rf server client trace &&
-	git init server &&
-	test_commit -C server to_fetch &&
-
-	git init client &&
-	test_commit -C client on_client &&
-	git -C client checkout on_client &&
-
-	test_config -C client fetch.negotiationAlgorithm invalid &&
-	test_must_fail git -C client fetch "$(pwd)/server" 2>err &&
-	test_i18ngrep "unknown fetch negotiation algorithm" err &&
-
-	# Explicit "default" value
-	test_config -C client fetch.negotiationAlgorithm default &&
-	git -C client -c fetch.negotiationAlgorithm=default fetch "$(pwd)/server" &&
-
-	# Implementation detail: If there is nothing to fetch, we will not error out
-	test_config -C client fetch.negotiationAlgorithm invalid &&
-	git -C client fetch "$(pwd)/server" 2>err &&
-	test_i18ngrep ! "unknown fetch negotiation algorithm" err
-'
-
 test_expect_success 'when two skips collide, favor the larger one' '
 	rm -rf server client trace &&
 	git init server &&
-- 
gitgitgadget

Re: [PATCH 1/5] repo-settings: consolidate some config settings

[Johannes Schindelin]

Hi Stolee,

On Mon, 22 Jul 2019, Derrick Stolee via GitGitGadget wrote:

> diff --git a/builtin/gc.c b/builtin/gc.c
> index c18efadda5..243be2907b 100644
> --- a/builtin/gc.c
> +++ b/builtin/gc.c
> @@ -27,6 +27,7 @@
>  #include "pack-objects.h"
>  #include "blob.h"
>  #include "tree.h"
> +#include "repo-settings.h"
>
>  #define FAILED_RUN "failed to run %s"
>
> @@ -41,7 +42,6 @@ static int aggressive_depth = 50;
>  static int aggressive_window = 250;
>  static int gc_auto_threshold = 6700;
>  static int gc_auto_pack_limit = 50;
> -static int gc_write_commit_graph;

I _really_ like that direction. Anything that removes global state will
improve Git's source code.

> [...]
> diff --git a/read-cache.c b/read-cache.c
> index c701f7f8b8..ee1aaa8917 100644
> --- a/read-cache.c
> +++ b/read-cache.c
> [...]
> @@ -2765,7 +2767,7 @@ static int do_write_index(struct index_state *istate, struct tempfile *tempfile,
>  	}
>
>  	if (!istate->version) {
> -		istate->version = get_index_format_default();
> +		istate->version = get_index_format_default(the_repository);

It is too bad that `read-cache.h` is not `the_repository`-free at the
moment...

>  		if (git_env_bool("GIT_TEST_SPLIT_INDEX", 0))
>  			init_split_index(istate);
>  	}
> diff --git a/repo-settings.c b/repo-settings.c
> new file mode 100644
> index 0000000000..13a9128f62
> --- /dev/null
> +++ b/repo-settings.c
> @@ -0,0 +1,44 @@
> +#include "cache.h"
> +#include "repository.h"
> +#include "config.h"
> +#include "repo-settings.h"
> +
> +static int git_repo_config(const char *key, const char *value, void *cb)
> +{
> +	struct repo_settings *rs = (struct repo_settings *)cb;
> +
> +	if (!strcmp(key, "core.commitgraph")) {
> +		rs->core_commit_graph = git_config_bool(key, value);
> +		return 0;
> +	}
> +	if (!strcmp(key, "gc.writecommitgraph")) {
> +		rs->gc_write_commit_graph = git_config_bool(key, value);
> +		return 0;
> +	}
> +	if (!strcmp(key, "pack.usesparse")) {
> +		rs->pack_use_sparse = git_config_bool(key, value);
> +		return 0;
> +	}
> +	if (!strcmp(key, "index.version")) {
> +		rs->index_version = git_config_int(key, value);
> +		return 0;
> +	}

I would actually prefer to use the `repo_config_get_*()` family here.
That way, we really avoid re-parsing the config.

> +
> +	return 1;
> +}
> +
> +void prepare_repo_settings(struct repository *r)
> +{
> +	if (r->settings)
> +		return;
> +
> +	r->settings = xmalloc(sizeof(*r->settings));
> +
> +	/* Defaults */
> +	r->settings->core_commit_graph = -1;
> +	r->settings->gc_write_commit_graph = -1;
> +	r->settings->pack_use_sparse = -1;
> +	r->settings->index_version = -1;
> +
> +	repo_config(r, git_repo_config, r->settings);
> +}
> diff --git a/repo-settings.h b/repo-settings.h
> new file mode 100644
> index 0000000000..1151c2193a
> --- /dev/null
> +++ b/repo-settings.h
> @@ -0,0 +1,15 @@
> +#ifndef REPO_SETTINGS_H
> +#define REPO_SETTINGS_H
> +
> +struct repo_settings {
> +	int core_commit_graph;
> +	int gc_write_commit_graph;
> +	int pack_use_sparse;
> +	int index_version;
> +};
> +
> +struct repository;
> +
> +void prepare_repo_settings(struct repository *r);

Hmm. I can see that you wanted to encapsulate this, but I do not really
agree that this needs to be encapsulated away from `repository.h`. I'd
rather declare `struct repo_settings` in `repository.h` and then make
the `settings` a field of that type (as opposed to a pointer to that
type). In general, I like to avoid unnecessary `malloc()`s, and this
here instance is one of them.

Thanks,
Dscho

> +
> +#endif /* REPO_SETTINGS_H */
> diff --git a/repository.h b/repository.h
> index 4fb6a5885f..352afc9cd8 100644
> --- a/repository.h
> +++ b/repository.h
> @@ -4,6 +4,7 @@
>  #include "path.h"
>
>  struct config_set;
> +struct repo_settings;
>  struct git_hash_algo;
>  struct index_state;
>  struct lock_file;
> @@ -72,6 +73,8 @@ struct repository {
>  	 */
>  	char *submodule_prefix;
>
> +	struct repo_settings *settings;
> +
>  	/* Subsystems */
>  	/*
>  	 * Repository's config which contains key-value pairs from the usual
> --
> gitgitgadget
>
>

Re: [PATCH 2/5] repo-settings: add feature.manyCommits setting

[Johannes Schindelin]

Hi Stolee,

On Mon, 22 Jul 2019, Derrick Stolee via GitGitGadget wrote:

> From: Derrick Stolee <dstolee@microsoft.com>
>
> When a repo has many commits, it is helpful to write and read the
> commit-graph file. Future changes to Git may include new config
> settings that are benefitial in this scenario.

s/benefitial/beneficial/

>
> Create the 'feature.manyCommits' config setting that changes the
> default values of 'core.commitGraph' and 'gc.writeCommitGraph' to
> true.

Great!

> diff --git a/repo-settings.c b/repo-settings.c
> index 13a9128f62..f328602fd7 100644
> --- a/repo-settings.c
> +++ b/repo-settings.c
> @@ -3,10 +3,17 @@
>  #include "config.h"
>  #include "repo-settings.h"
>
> +#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
> +
>  static int git_repo_config(const char *key, const char *value, void *cb)
>  {
>  	struct repo_settings *rs = (struct repo_settings *)cb;
>
> +	if (!strcmp(key, "feature.manycommits")) {
> +		UPDATE_DEFAULT(rs->core_commit_graph, 1);
> +		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
> +		return 0;
> +	}
>  	if (!strcmp(key, "core.commitgraph")) {
>  		rs->core_commit_graph = git_config_bool(key, value);
>  		return 0;

Okay, this one is tricky. The behavior I would want is for
`feature.manycommits` to override the _default_. And if I set
`feature.manycommits = false` (e.g. via `git -c
feature.manycommits=false ...`), I would want the default to "go back".

So I'd really rather see this as

	if (!repo_config_get_bool(r, "feature.manycommits", &b) && b) {
		rs->core_commit_graph = 1;
		rs->gc_write_commit_graph = 1;
	}

	[...]

	repo_config_get_bool(r, "core.commitgraph", &rs->core_commit_graph);

I.e. override the default only if `feature.manyCommits` was true *in the
end*.

In any case, we really need to make sure to handle the `= false` case
correctly here. We might want to override the setting from the
command-line, or it might be set in `~/.gitconfig` and need to be
overridden in a local repository. We need to handle it.

Otherwise this looks good!
Dscho

Re: [PATCH 3/5] repo-settings: parse core.untrackedCache

[Johannes Schindelin]

Hi Stolee,

On Mon, 22 Jul 2019, Derrick Stolee via GitGitGadget wrote:

> From: Derrick Stolee <dstolee@microsoft.com>
>
> The core.untrackedCache config setting is slightly complicated,
> so clarify its use and centralize its parsing into the repo
> settings.
>
> The default value is "keep" (returned as -1), which persists the
> untracked cache if it exists.
>
> If the value is set as "false" (returned as 0), then remove the
> untracked cache if it exists.
>
> If the value is set as "true" (returned as 1), then write the
> untracked cache and persist it.
>
> Instead of relying on magic values of -1, 0, and 1, split these
> options into bitflags CORE_UNTRACKED_CACHE_KEEP and
> CORE_UNTRACKED_CACHE_WRITE. This allows the use of "-1" as a
> default value. After parsing the config options, if the value is
> unset we can initialize it to CORE_UNTRACKED_CACHE_KEEP.

Nice!

> [...]
> diff --git a/read-cache.c b/read-cache.c
> index ee1aaa8917..e67e6f6e3e 100644
> --- a/read-cache.c
> +++ b/read-cache.c
> @@ -1846,18 +1846,17 @@ static void check_ce_order(struct index_state *istate)
>
>  static void tweak_untracked_cache(struct index_state *istate)
>  {
> -	switch (git_config_get_untracked_cache()) {
> -	case -1: /* keep: do nothing */
> -		break;
> -	case 0: /* false */
> +	struct repository *r = the_repository;
> +
> +	prepare_repo_settings(r);
> +
> +	if (!(r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_KEEP)) {
>  		remove_untracked_cache(istate);
> -		break;
> -	case 1: /* true */
> -		add_untracked_cache(istate);
> -		break;
> -	default: /* unknown value: do nothing */
> -		break;
> +		return;
>  	}
> +
> +	if (r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_WRITE)
> +		add_untracked_cache(istate);

This changes the flow in a subtle way: in the
`CORE_UNTRACKED_CACHE_WRITE` case, we used to _not_ remove the untracked
cache, but now we do.

I _think_ what you would want to do is replace the `!(..._KEEP)`
condition by `..._REMOVE`.

>  }
>
>  static void tweak_split_index(struct index_state *istate)
> diff --git a/repo-settings.c b/repo-settings.c
> index f328602fd7..807c5a29d6 100644
> --- a/repo-settings.c
> +++ b/repo-settings.c
> @@ -30,6 +30,20 @@ static int git_repo_config(const char *key, const char *value, void *cb)
>  		rs->index_version = git_config_int(key, value);
>  		return 0;
>  	}
> +	if (!strcmp(key, "core.untrackedcache")) {
> +		int bool_value = git_parse_maybe_bool(value);
> +		if (bool_value == 0)
> +			rs->core_untracked_cache = 0;
> +		else if (bool_value == 1)
> +			rs->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP |
> +						   CORE_UNTRACKED_CACHE_WRITE;
> +		else if (!strcasecmp(value, "keep"))
> +			rs->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
> +		else
> +			error(_("unknown core.untrackedCache value '%s'; "
> +				"using 'keep' default value"), value);
> +		return 0;
> +	}
>
>  	return 1;
>  }
> @@ -46,6 +60,13 @@ void prepare_repo_settings(struct repository *r)
>  	r->settings->gc_write_commit_graph = -1;
>  	r->settings->pack_use_sparse = -1;
>  	r->settings->index_version = -1;
> +	r->settings->core_untracked_cache = -1;

At this point at the latest, I am starting to wonder whether it would
not make more sense to use `memset(..., -1, sizeof(struct
repo_settings)` instead.

>
>  	repo_config(r, git_repo_config, r->settings);
> +
> +	/* Hack for test programs like test-dump-untracked-cache */
> +	if (ignore_untracked_cache_config)
> +		r->settings->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
> +	else
> +		UPDATE_DEFAULT(r->settings->core_untracked_cache, CORE_UNTRACKED_CACHE_KEEP);
>  }
> diff --git a/repo-settings.h b/repo-settings.h
> index 1151c2193a..bac9b87d49 100644
> --- a/repo-settings.h
> +++ b/repo-settings.h
> @@ -1,11 +1,15 @@
>  #ifndef REPO_SETTINGS_H
>  #define REPO_SETTINGS_H
>
> +#define CORE_UNTRACKED_CACHE_WRITE (1 << 0)
> +#define CORE_UNTRACKED_CACHE_KEEP (1 << 1)

I think it would read even nicer as an enum. In any case, using `1<<1`
suggests that this is a bit field, but I don't think that is what we
actually want here. Instead, what `core_untracked_cache` seems to be (at
least from my point of view) is a mode, where any two modes are mutually
exclusive.

For example, what is the difference between `(_KEEP | _WRITE)` and
`(_WRITE)` supposed to be? That the latter writes a fresh untracked
cache without looking at the previous one? That's not how the existing
code behaves, though...

Ciao,
Dscho

> +
>  struct repo_settings {
>  	int core_commit_graph;
>  	int gc_write_commit_graph;
>  	int pack_use_sparse;
>  	int index_version;
> +	int core_untracked_cache;
>  };
>
>  struct repository;
> --
> gitgitgadget
>
>

Re: [PATCH 5/5] repo-settings: create feature.experimental setting

[Johannes Schindelin]

Hi Stolee,

On Mon, 22 Jul 2019, Derrick Stolee via GitGitGadget wrote:

> From: Derrick Stolee <dstolee@microsoft.com>
>
> The 'feature.experimental' setting includes config options that are
> not committed to become defaults, but could use additional testing.
>
> Update the following config settings to take new defaults, and to
> use the repo_settings struct if not already using them:
>
> * 'pack.useSparse=true'
>
> * 'merge.directoryRenames=true'
>
> * 'fetch.negotiationAlgorithm=skipping'
>
> In the case of fetch.negotiationAlgorithm, the existing logic
> would load the config option only when about to use the setting,
> so had a die() statement on an unknown string value. This is
> removed as now the config is parsed under prepare_repo_settings().
> In general, this die() is probably misplaced and not valuable.
> A test was removed that checked this die() statement executed.

Good.

> [...]
> diff --git a/merge-recursive.c b/merge-recursive.c
> index 12300131fc..162d5a4753 100644
> --- a/merge-recursive.c
> +++ b/merge-recursive.c
> @@ -28,6 +28,7 @@
>  #include "submodule.h"
>  #include "revision.h"
>  #include "commit-reach.h"
> +#include "repo-settings.h"
>
>  struct path_hashmap_entry {
>  	struct hashmap_entry e;
> @@ -1375,10 +1376,14 @@ static int handle_rename_via_dir(struct merge_options *opt,
>  	 * there is no content merge to do; just move the file into the
>  	 * desired final location.
>  	 */
> +	struct repository *r = the_repository;
>  	const struct rename *ren = ci->ren1;
>  	const struct diff_filespec *dest = ren->pair->two;
>  	char *file_path = dest->path;
> -	int mark_conflicted = (opt->detect_directory_renames == 1);

I actually don't think that we want to do that; the `opt` parameter is
passed to the merge recursive algorithm specifically so that it can be
overridden by the caller.

Instead, what we should do, I think, is to change `init_merge_options()`
(which already gets a parameter of type `struct repository *`) so that
it does not hard-code the `detect_directory_renames` value to `1` but to
query the repo settings instead.

> +	int mark_conflicted;
> +
> +	prepare_repo_settings(r);
> +	mark_conflicted = (r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_CONFLICT);
>  	assert(ren->dir_rename_original_dest);
>
>  	if (!opt->call_depth && would_lose_untracked(opt, dest->path)) {
> @@ -2850,6 +2855,7 @@ static int detect_and_process_renames(struct merge_options *opt,
>  				      struct string_list *entries,
>  				      struct rename_info *ri)
>  {
> +	struct repository *r = the_repository;
>  	struct diff_queue_struct *head_pairs, *merge_pairs;
>  	struct hashmap *dir_re_head, *dir_re_merge;
>  	int clean = 1;
> @@ -2863,7 +2869,8 @@ static int detect_and_process_renames(struct merge_options *opt,
>  	head_pairs = get_diffpairs(opt, common, head);
>  	merge_pairs = get_diffpairs(opt, common, merge);
>
> -	if (opt->detect_directory_renames) {
> +	prepare_repo_settings(r);
> +	if (r->settings->merge_directory_renames) {
>  		dir_re_head = get_directory_renames(head_pairs);
>  		dir_re_merge = get_directory_renames(merge_pairs);
>
> @@ -3112,6 +3119,7 @@ static int handle_rename_normal(struct merge_options *opt,
>  				const struct diff_filespec *b,
>  				struct rename_conflict_info *ci)
>  {
> +	struct repository *r = the_repository;
>  	struct rename *ren = ci->ren1;
>  	struct merge_file_info mfi;
>  	int clean;
> @@ -3121,7 +3129,9 @@ static int handle_rename_normal(struct merge_options *opt,
>  	clean = handle_content_merge(&mfi, opt, path, was_dirty(opt, path),
>  				     o, a, b, ci);
>
> -	if (clean && opt->detect_directory_renames == 1 &&
> +	prepare_repo_settings(r);
> +	if (clean &&
> +	    r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_CONFLICT &&
>  	    ren->dir_rename_original_dest) {
>  		if (update_stages(opt, path,
>  				  NULL,
> @@ -3155,6 +3165,7 @@ static void dir_rename_warning(const char *msg,
>  static int warn_about_dir_renamed_entries(struct merge_options *opt,
>  					  struct rename *ren)
>  {
> +	struct repository *r = the_repository;
>  	const char *msg;
>  	int clean = 1, is_add;
>
> @@ -3166,12 +3177,13 @@ static int warn_about_dir_renamed_entries(struct merge_options *opt,
>  		return clean;
>
>  	/* Sanity checks */
> -	assert(opt->detect_directory_renames > 0);
> +	prepare_repo_settings(r);
> +	assert(r->settings->merge_directory_renames > 0);
>  	assert(ren->dir_rename_original_type == 'A' ||
>  	       ren->dir_rename_original_type == 'R');
>
>  	/* Check whether to treat directory renames as a conflict */
> -	clean = (opt->detect_directory_renames == 2);
> +	clean = (r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_TRUE);
>
>  	is_add = (ren->dir_rename_original_type == 'A');
>  	if (ren->dir_rename_original_type == 'A' && clean) {
> @@ -3662,15 +3674,6 @@ static void merge_recursive_config(struct merge_options *opt)
>  		opt->merge_detect_rename = git_config_rename("merge.renames", value);
>  		free(value);
>  	}
> -	if (!git_config_get_string("merge.directoryrenames", &value)) {
> -		int boolval = git_parse_maybe_bool(value);
> -		if (0 <= boolval) {
> -			opt->detect_directory_renames = boolval ? 2 : 0;
> -		} else if (!strcasecmp(value, "conflict")) {
> -			opt->detect_directory_renames = 1;
> -		} /* avoid erroring on values from future versions of git */
> -		free(value);
> -	}
>  	git_config(git_xmerge_config, NULL);
>  }
>
> @@ -3687,7 +3690,6 @@ void init_merge_options(struct merge_options *opt,
>  	opt->renormalize = 0;
>  	opt->diff_detect_rename = -1;
>  	opt->merge_detect_rename = -1;
> -	opt->detect_directory_renames = 1;

In other words: here.

>  	merge_recursive_config(opt);
>  	merge_verbosity = getenv("GIT_MERGE_VERBOSITY");
>  	if (merge_verbosity)
> diff --git a/merge-recursive.h b/merge-recursive.h
> index c2b7bb65c6..b8eba244ee 100644
> --- a/merge-recursive.h
> +++ b/merge-recursive.h
> @@ -22,7 +22,6 @@ struct merge_options {
>  	unsigned renormalize : 1;
>  	long xdl_opts;
>  	int verbosity;
> -	int detect_directory_renames;
>  	int diff_detect_rename;
>  	int merge_detect_rename;
>  	int diff_rename_limit;
> diff --git a/repo-settings.c b/repo-settings.c
> index 9e4b8e6268..5e9249c437 100644
> --- a/repo-settings.c
> +++ b/repo-settings.c
> @@ -9,6 +9,12 @@ static int git_repo_config(const char *key, const char *value, void *cb)
>  {
>  	struct repo_settings *rs = (struct repo_settings *)cb;
>
> +	if (!strcmp(key, "feature.experimental")) {
> +		UPDATE_DEFAULT(rs->pack_use_sparse, 1);
> +		UPDATE_DEFAULT(rs->merge_directory_renames, MERGE_DIRECTORY_RENAMES_TRUE);
> +		UPDATE_DEFAULT(rs->fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
> +		return 0;
> +	}
>  	if (!strcmp(key, "feature.manycommits")) {
>  		UPDATE_DEFAULT(rs->core_commit_graph, 1);
>  		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
> @@ -50,6 +56,23 @@ static int git_repo_config(const char *key, const char *value, void *cb)
>  				"using 'keep' default value"), value);
>  		return 0;
>  	}
> +	if (!strcmp(key, "merge.directoryrenames")) {
> +		int bool_value = git_parse_maybe_bool(value);
> +		if (0 <= bool_value) {
> +			rs->merge_directory_renames = bool_value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
> +		} else if (!strcasecmp(value, "conflict")) {
> +			rs->merge_directory_renames = MERGE_DIRECTORY_RENAMES_CONFLICT;
> +		}
> +		return 0;
> +	}
> +	if (!strcmp(key, "fetch.negotiationalgorithm"))	{
> +		if (!strcasecmp(value, "skipping")) {
> +			rs->fetch_negotiation_algorithm = FETCH_NEGOTIATION_SKIPPING;
> +		} else {
> +			rs->fetch_negotiation_algorithm = FETCH_NEGOTIATION_DEFAULT;
> +		}
> +		return 0;
> +	}
>
>  	return 1;
>  }
> @@ -64,10 +87,14 @@ void prepare_repo_settings(struct repository *r)
>  	/* Defaults */
>  	r->settings->core_commit_graph = -1;
>  	r->settings->gc_write_commit_graph = -1;
> -	r->settings->pack_use_sparse = -1;
> +
>  	r->settings->index_version = -1;
>  	r->settings->core_untracked_cache = -1;
>
> +	r->settings->pack_use_sparse = -1;

This reordering at this stage in the patch series is a bit confusing.
Maybe add it at the location where you want it to end up to begin with?

Or use `memset(..., -1, ...)`.

> +	r->settings->merge_directory_renames = -1;
> +	r->settings->fetch_negotiation_algorithm = -1;
> +
>  	repo_config(r, git_repo_config, r->settings);
>
>  	/* Hack for test programs like test-dump-untracked-cache */
> @@ -75,4 +102,7 @@ void prepare_repo_settings(struct repository *r)
>  		r->settings->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
>  	else
>  		UPDATE_DEFAULT(r->settings->core_untracked_cache, CORE_UNTRACKED_CACHE_KEEP);
> +
> +	UPDATE_DEFAULT(r->settings->merge_directory_renames, MERGE_DIRECTORY_RENAMES_CONFLICT);
> +	UPDATE_DEFAULT(r->settings->fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
>  }
> diff --git a/repo-settings.h b/repo-settings.h
> index bac9b87d49..cecf7d0028 100644
> --- a/repo-settings.h
> +++ b/repo-settings.h
> @@ -4,12 +4,22 @@
>  #define CORE_UNTRACKED_CACHE_WRITE (1 << 0)
>  #define CORE_UNTRACKED_CACHE_KEEP (1 << 1)
>
> +#define MERGE_DIRECTORY_RENAMES_CONFLICT 1
> +#define MERGE_DIRECTORY_RENAMES_TRUE 2
> +
> +#define FETCH_NEGOTIATION_DEFAULT 1
> +#define FETCH_NEGOTIATION_SKIPPING 2
> +

Again, I'd prefer enums.

The rest looks pretty fine to me (I have to admit that I only glanced
over the tests, though).

Thanks!
Dscho

>  struct repo_settings {
>  	int core_commit_graph;
>  	int gc_write_commit_graph;
> -	int pack_use_sparse;
> +
>  	int index_version;
>  	int core_untracked_cache;
> +
> +	int pack_use_sparse;
> +	int merge_directory_renames;
> +	int fetch_negotiation_algorithm;
>  };
>
>  struct repository;
> diff --git a/t/t5552-skipping-fetch-negotiator.sh b/t/t5552-skipping-fetch-negotiator.sh
> index 8a14be51a1..f70cbcc9ca 100755
> --- a/t/t5552-skipping-fetch-negotiator.sh
> +++ b/t/t5552-skipping-fetch-negotiator.sh
> @@ -60,29 +60,6 @@ test_expect_success 'commits with no parents are sent regardless of skip distanc
>  	have_not_sent c6 c4 c3
>  '
>
> -test_expect_success 'unknown fetch.negotiationAlgorithm values error out' '
> -	rm -rf server client trace &&
> -	git init server &&
> -	test_commit -C server to_fetch &&
> -
> -	git init client &&
> -	test_commit -C client on_client &&
> -	git -C client checkout on_client &&
> -
> -	test_config -C client fetch.negotiationAlgorithm invalid &&
> -	test_must_fail git -C client fetch "$(pwd)/server" 2>err &&
> -	test_i18ngrep "unknown fetch negotiation algorithm" err &&
> -
> -	# Explicit "default" value
> -	test_config -C client fetch.negotiationAlgorithm default &&
> -	git -C client -c fetch.negotiationAlgorithm=default fetch "$(pwd)/server" &&
> -
> -	# Implementation detail: If there is nothing to fetch, we will not error out
> -	test_config -C client fetch.negotiationAlgorithm invalid &&
> -	git -C client fetch "$(pwd)/server" 2>err &&
> -	test_i18ngrep ! "unknown fetch negotiation algorithm" err
> -'
> -
>  test_expect_success 'when two skips collide, favor the larger one' '
>  	rm -rf server client trace &&
>  	git init server &&
> --
> gitgitgadget
>

Re: [PATCH 2/5] repo-settings: add feature.manyCommits setting

[Derrick Stolee]

On 7/23/2019 10:53 AM, Johannes Schindelin wrote:
> Hi Stolee,
> 
> On Mon, 22 Jul 2019, Derrick Stolee via GitGitGadget wrote:
> 
>> From: Derrick Stolee <dstolee@microsoft.com>
>>
>> When a repo has many commits, it is helpful to write and read the
>> commit-graph file. Future changes to Git may include new config
>> settings that are benefitial in this scenario.
> 
> s/benefitial/beneficial/
> 
>>
>> Create the 'feature.manyCommits' config setting that changes the
>> default values of 'core.commitGraph' and 'gc.writeCommitGraph' to
>> true.
> 
> Great!
> 
>> diff --git a/repo-settings.c b/repo-settings.c
>> index 13a9128f62..f328602fd7 100644
>> --- a/repo-settings.c
>> +++ b/repo-settings.c
>> @@ -3,10 +3,17 @@
>>  #include "config.h"
>>  #include "repo-settings.h"
>>
>> +#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
>> +
>>  static int git_repo_config(const char *key, const char *value, void *cb)
>>  {
>>  	struct repo_settings *rs = (struct repo_settings *)cb;
>>
>> +	if (!strcmp(key, "feature.manycommits")) {
>> +		UPDATE_DEFAULT(rs->core_commit_graph, 1);
>> +		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
>> +		return 0;
>> +	}
>>  	if (!strcmp(key, "core.commitgraph")) {
>>  		rs->core_commit_graph = git_config_bool(key, value);
>>  		return 0;
> 
> Okay, this one is tricky. The behavior I would want is for
> `feature.manycommits` to override the _default_. And if I set
> `feature.manycommits = false` (e.g. via `git -c
> feature.manycommits=false ...`), I would want the default to "go back".
> 
> So I'd really rather see this as
> 
> 	if (!repo_config_get_bool(r, "feature.manycommits", &b) && b) {
> 		rs->core_commit_graph = 1;
> 		rs->gc_write_commit_graph = 1;
> 	}
> 
> 	[...]
> 
> 	repo_config_get_bool(r, "core.commitgraph", &rs->core_commit_graph);
> 
> I.e. override the default only if `feature.manyCommits` was true *in the
> end*.
> 
> In any case, we really need to make sure to handle the `= false` case
> correctly here. We might want to override the setting from the
> command-line, or it might be set in `~/.gitconfig` and need to be
> overridden in a local repository. We need to handle it.

I had forgotten about this interaction. Thanks for finding it!

The way I would fix it is to add an "int feature_many_commits" to
the repo_settings struct, then check its value at the very end of
prepare_repo_settings() and then UPDATE_DEFAULT() for the settings
we want.

Thanks!
-Stolee

Re: [PATCH 3/5] repo-settings: parse core.untrackedCache

[Derrick Stolee]

On 7/23/2019 11:04 AM, Johannes Schindelin wrote:
> Hi Stolee,
> 
> On Mon, 22 Jul 2019, Derrick Stolee via GitGitGadget wrote:
> 
>> From: Derrick Stolee <dstolee@microsoft.com>
>>
>> The core.untrackedCache config setting is slightly complicated,
>> so clarify its use and centralize its parsing into the repo
>> settings.
>>
>> The default value is "keep" (returned as -1), which persists the
>> untracked cache if it exists.
>>
>> If the value is set as "false" (returned as 0), then remove the
>> untracked cache if it exists.
>>
>> If the value is set as "true" (returned as 1), then write the
>> untracked cache and persist it.
>>
>> Instead of relying on magic values of -1, 0, and 1, split these
>> options into bitflags CORE_UNTRACKED_CACHE_KEEP and
>> CORE_UNTRACKED_CACHE_WRITE. This allows the use of "-1" as a
>> default value. After parsing the config options, if the value is
>> unset we can initialize it to CORE_UNTRACKED_CACHE_KEEP.
> 
> Nice!
> 
>> [...]
>> diff --git a/read-cache.c b/read-cache.c
>> index ee1aaa8917..e67e6f6e3e 100644
>> --- a/read-cache.c
>> +++ b/read-cache.c
>> @@ -1846,18 +1846,17 @@ static void check_ce_order(struct index_state *istate)
>>
>>  static void tweak_untracked_cache(struct index_state *istate)
>>  {
>> -	switch (git_config_get_untracked_cache()) {
>> -	case -1: /* keep: do nothing */
>> -		break;
>> -	case 0: /* false */
>> +	struct repository *r = the_repository;
>> +
>> +	prepare_repo_settings(r);
>> +
>> +	if (!(r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_KEEP)) {
>>  		remove_untracked_cache(istate);
>> -		break;
>> -	case 1: /* true */
>> -		add_untracked_cache(istate);
>> -		break;
>> -	default: /* unknown value: do nothing */
>> -		break;
>> +		return;
>>  	}
>> +
>> +	if (r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_WRITE)
>> +		add_untracked_cache(istate);
> 
> This changes the flow in a subtle way: in the
> `CORE_UNTRACKED_CACHE_WRITE` case, we used to _not_ remove the untracked
> cache, but now we do.
> 
> I _think_ what you would want to do is replace the `!(..._KEEP)`
> condition by `..._REMOVE`.

I believe the code as written is correct, but confusing. The value is not an enum, but instead a bitflag. When the config setting is given as "true", then both _KEEP and _WRITE are set, so the flow is identical.

However, you already suggested switching to an enum, in which case using _REMOVE would be clearer.

> 
>>  }
>>
>>  static void tweak_split_index(struct index_state *istate)
>> diff --git a/repo-settings.c b/repo-settings.c
>> index f328602fd7..807c5a29d6 100644
>> --- a/repo-settings.c
>> +++ b/repo-settings.c
>> @@ -30,6 +30,20 @@ static int git_repo_config(const char *key, const char *value, void *cb)
>>  		rs->index_version = git_config_int(key, value);
>>  		return 0;
>>  	}
>> +	if (!strcmp(key, "core.untrackedcache")) {
>> +		int bool_value = git_parse_maybe_bool(value);
>> +		if (bool_value == 0)
>> +			rs->core_untracked_cache = 0;
>> +		else if (bool_value == 1)
>> +			rs->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP |
>> +						   CORE_UNTRACKED_CACHE_WRITE;
>> +		else if (!strcasecmp(value, "keep"))
>> +			rs->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
>> +		else
>> +			error(_("unknown core.untrackedCache value '%s'; "
>> +				"using 'keep' default value"), value);
>> +		return 0;
>> +	}
>>
>>  	return 1;
>>  }
>> @@ -46,6 +60,13 @@ void prepare_repo_settings(struct repository *r)
>>  	r->settings->gc_write_commit_graph = -1;
>>  	r->settings->pack_use_sparse = -1;
>>  	r->settings->index_version = -1;
>> +	r->settings->core_untracked_cache = -1;
> 
> At this point at the latest, I am starting to wonder whether it would
> not make more sense to use `memset(..., -1, sizeof(struct
> repo_settings)` instead.
> 
>>
>>  	repo_config(r, git_repo_config, r->settings);
>> +
>> +	/* Hack for test programs like test-dump-untracked-cache */
>> +	if (ignore_untracked_cache_config)
>> +		r->settings->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
>> +	else
>> +		UPDATE_DEFAULT(r->settings->core_untracked_cache, CORE_UNTRACKED_CACHE_KEEP);
>>  }
>> diff --git a/repo-settings.h b/repo-settings.h
>> index 1151c2193a..bac9b87d49 100644
>> --- a/repo-settings.h
>> +++ b/repo-settings.h
>> @@ -1,11 +1,15 @@
>>  #ifndef REPO_SETTINGS_H
>>  #define REPO_SETTINGS_H
>>
>> +#define CORE_UNTRACKED_CACHE_WRITE (1 << 0)
>> +#define CORE_UNTRACKED_CACHE_KEEP (1 << 1)
> 
> I think it would read even nicer as an enum. In any case, using `1<<1`
> suggests that this is a bit field, but I don't think that is what we
> actually want here. Instead, what `core_untracked_cache` seems to be (at
> least from my point of view) is a mode, where any two modes are mutually
> exclusive.
> 
> For example, what is the difference between `(_KEEP | _WRITE)` and
> `(_WRITE)` supposed to be? That the latter writes a fresh untracked
> cache without looking at the previous one? That's not how the existing
> code behaves, though...

Yes, there is no reason to have _WRITE without also _KEEP. An enum is better. Thanks!

> 
> Ciao,
> Dscho
> 
>> +
>>  struct repo_settings {
>>  	int core_commit_graph;
>>  	int gc_write_commit_graph;
>>  	int pack_use_sparse;
>>  	int index_version;
>> +	int core_untracked_cache;
>>  };
>>
>>  struct repository;
>> --
>> gitgitgadget
>>
>>

Re: [PATCH 5/5] repo-settings: create feature.experimental setting

[Derrick Stolee]

On 7/23/2019 11:20 AM, Johannes Schindelin wrote:
> Hi Stolee,
> 
> On Mon, 22 Jul 2019, Derrick Stolee via GitGitGadget wrote:
>>  struct path_hashmap_entry {
>>  	struct hashmap_entry e;
>> @@ -1375,10 +1376,14 @@ static int handle_rename_via_dir(struct merge_options *opt,
>>  	 * there is no content merge to do; just move the file into the
>>  	 * desired final location.
>>  	 */
>> +	struct repository *r = the_repository;
>>  	const struct rename *ren = ci->ren1;
>>  	const struct diff_filespec *dest = ren->pair->two;
>>  	char *file_path = dest->path;
>> -	int mark_conflicted = (opt->detect_directory_renames == 1);
> 
> I actually don't think that we want to do that; the `opt` parameter is
> passed to the merge recursive algorithm specifically so that it can be
> overridden by the caller.
> 
> Instead, what we should do, I think, is to change `init_merge_options()`
> (which already gets a parameter of type `struct repository *`) so that
> it does not hard-code the `detect_directory_renames` value to `1` but to
> query the repo settings instead.

[snip]
>> @@ -3687,7 +3690,6 @@ void init_merge_options(struct merge_options *opt,
>>  	opt->renormalize = 0;
>>  	opt->diff_detect_rename = -1;
>>  	opt->merge_detect_rename = -1;
>> -	opt->detect_directory_renames = 1;
> 
> In other words: here.

Yes, thanks! Your suggestion makes much more sense and makes the diff
much simpler. No reason to couple too closely to the config storage here.

I think you gave me enough feedback to merit a v2 before others chime
in, so I'll send one after I validate my current version.

Thanks,
-Stolee

[PATCH v2 0/5] Create 'feature.*' config area and some centralized config parsing

[Derrick Stolee via GitGitGadget]

This is a brand-new thread to replace ds/early-access. The discussion on
that thread was very helpful [1].

With this in mind, I propose instead a set of "feature.*" config settings
that form groups of "community recommended" settings (with some caveats). In
the space below, I'll list a set of possible feature names and the implied
config options.

First, the main two categories we've discussed so far: many commits and many
files. These two feature sets are for when your repo is large in one of
these dimensions. Perhaps there are other settings to include in these?

feature.manyFiles:
    index.version = 4
    core.untrackedCache = true

feature.manyCommits:
    core.commitGraph = true
    gc.writeCommitGraph = true
    (future: fetch.writeSplitCommitGraph = true)

Note: the fetch.writeSplitCommitGraph does not exist yet, but could be
introduced in a later release to write a new commit-graph (with --split) on
fetch.

The other category that has been discussed already is that of "experimental
features that we generally think are helpful but change behavior slightly in
some cases".

feature.experimental:
    pack.useSparse = true
    merge.directoryRenames = true
    fetch.negotiationAlgorithm = skipping

Specifically, this setting is for config values we are not sure will ever be
on by default, but additional testing is needed to be sure. This is
different than a possible 'feature.preview' setting that would include
config settings that we are committed to updating the defaults in a future
release. There are many ways we can take this idea in the future (including
more additions to these categories).

Thanks, -Stolee

Updates in V2: I'm responding to Dscho's comments early, as they were very
valuable.

 * Rearranged how we are checking for the feature.* settings, so if one
   layer turns the setting on and a later layer turns it off, we do not
   adjust the defaults.
   
   
 * Switched to using enums for the non-boolean config options.
   
   
 * Placed the repo_settings struct directly in the repository struct.
   
   
 * All struct members are initialized to -1 using memset.
   
   
 * The config changes around directory rename detection is greatly
   simplified.
   
   

[1] https://public-inbox.org/git/pull.254.git.gitgitgadget@gmail.com/

Derrick Stolee (5):
  repo-settings: consolidate some config settings
  repo-settings: add feature.manyCommits setting
  repo-settings: parse core.untrackedCache
  repo-settings: create feature.manyFiles setting
  repo-settings: create feature.experimental setting

 Documentation/config.txt             |  2 +
 Documentation/config/core.txt        |  7 ++-
 Documentation/config/feature.txt     | 42 +++++++++++++++
 Documentation/config/fetch.txt       |  3 +-
 Documentation/config/gc.txt          |  4 +-
 Documentation/config/index.txt       |  1 +
 Documentation/config/merge.txt       |  3 +-
 Documentation/config/pack.txt        |  3 +-
 Makefile                             |  1 +
 builtin/am.c                         |  1 +
 builtin/gc.c                         | 13 +++--
 builtin/pack-objects.c               |  9 ++--
 builtin/update-index.c               |  7 ++-
 commit-graph.c                       |  7 +--
 config.c                             | 24 ---------
 fetch-negotiator.c                   | 26 +++++-----
 fetch-negotiator.h                   |  5 +-
 fetch-pack.c                         | 11 ++--
 merge-recursive.c                    | 15 +++---
 read-cache.c                         | 31 +++++------
 repo-settings.c                      | 78 ++++++++++++++++++++++++++++
 repo-settings.h                      | 41 +++++++++++++++
 repository.h                         |  5 ++
 t/t1600-index.sh                     | 31 +++++++++--
 t/t5552-skipping-fetch-negotiator.sh | 23 --------
 25 files changed, 274 insertions(+), 119 deletions(-)
 create mode 100644 Documentation/config/feature.txt
 create mode 100644 repo-settings.c
 create mode 100644 repo-settings.h


base-commit: 9c9b961d7eb15fb583a2a812088713a68a85f1c0
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-292%2Fderrickstolee%2Frepo-settings%2Fhead-v2
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-292/derrickstolee/repo-settings/head-v2
Pull-Request: https://github.com/gitgitgadget/git/pull/292

Range-diff vs v1:

 1:  72f652b89c ! 1:  597ab7d621 repo-settings: consolidate some config settings
     @@ -64,7 +64,7 @@
      -					 NULL))
      -		return 1;
      +	prepare_repo_settings(the_repository);
     -+	if (the_repository->settings->gc_write_commit_graph == 1)
     ++	if (the_repository->settings.gc_write_commit_graph == 1)
      +		write_commit_graph_reachable(get_object_directory(),
      +					     !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
      +					     NULL);
     @@ -99,8 +99,8 @@
       
       	sparse = git_env_bool("GIT_TEST_PACK_SPARSE", 0);
      +	prepare_repo_settings(the_repository);
     -+	if (!sparse && the_repository->settings->pack_use_sparse != -1)
     -+		sparse = the_repository->settings->pack_use_sparse;
     ++	if (!sparse && the_repository->settings.pack_use_sparse != -1)
     ++		sparse = the_repository->settings.pack_use_sparse;
      +
       	reset_pack_idx_option(&pack_idx_opts);
       	git_config(git_pack_config, NULL);
     @@ -134,7 +134,7 @@
       	if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
      -	    (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
      -	    !config_value))
     -+	    r->settings->core_commit_graph != 1)
     ++	    r->settings.core_commit_graph != 1)
       		/*
       		 * This repository is not configured to use commit graphs, so
       		 * do not load one. (But report commit_graph_attempted anyway
     @@ -167,8 +167,8 @@
      -			version = value;
      +		prepare_repo_settings(r);
      +
     -+		if (r->settings->index_version >= 0)
     -+			version = r->settings->index_version;
     ++		if (r->settings.index_version >= 0)
     ++			version = r->settings.index_version;
       		if (version < INDEX_FORMAT_LB || INDEX_FORMAT_UB < version) {
       			warning(_("index.version set, but the value is invalid.\n"
       				  "Using version %i"), INDEX_FORMAT_DEFAULT);
     @@ -192,44 +192,28 @@
      +#include "config.h"
      +#include "repo-settings.h"
      +
     -+static int git_repo_config(const char *key, const char *value, void *cb)
     ++void prepare_repo_settings(struct repository *r)
      +{
     -+	struct repo_settings *rs = (struct repo_settings *)cb;
     ++	int value;
      +
     -+	if (!strcmp(key, "core.commitgraph")) {
     -+		rs->core_commit_graph = git_config_bool(key, value);
     -+		return 0;
     -+	}
     -+	if (!strcmp(key, "gc.writecommitgraph")) {
     -+		rs->gc_write_commit_graph = git_config_bool(key, value);
     -+		return 0;
     -+	}
     -+	if (!strcmp(key, "pack.usesparse")) {
     -+		rs->pack_use_sparse = git_config_bool(key, value);
     -+		return 0;
     -+	}
     -+	if (!strcmp(key, "index.version")) {
     -+		rs->index_version = git_config_int(key, value);
     -+		return 0;
     -+	}
     ++	if (r->settings_initialized)
     ++		return;
      +
     -+	return 1;
     -+}
     ++	/* Defaults */
     ++	memset(&r->settings, -1, sizeof(r->settings));
      +
     -+void prepare_repo_settings(struct repository *r)
     -+{
     -+	if (r->settings)
     -+		return;
     ++	if (!repo_config_get_bool(r, "core.commitgraph", &value))
     ++		r->settings.core_commit_graph = value;
     ++	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
     ++		r->settings.gc_write_commit_graph = value;
      +
     -+	r->settings = xmalloc(sizeof(*r->settings));
     ++	if (!repo_config_get_bool(r, "index.version", &value))
     ++		r->settings.index_version = value;
      +
     -+	/* Defaults */
     -+	r->settings->core_commit_graph = -1;
     -+	r->settings->gc_write_commit_graph = -1;
     -+	r->settings->pack_use_sparse = -1;
     -+	r->settings->index_version = -1;
     ++	if (!repo_config_get_bool(r, "pack.usesparse", &value))
     ++		r->settings.pack_use_sparse = value;
      +
     -+	repo_config(r, git_repo_config, r->settings);
     ++	r->settings_initialized = 1;
      +}
      
       diff --git a/repo-settings.h b/repo-settings.h
     @@ -243,8 +227,10 @@
      +struct repo_settings {
      +	int core_commit_graph;
      +	int gc_write_commit_graph;
     -+	int pack_use_sparse;
     ++
      +	int index_version;
     ++
     ++	int pack_use_sparse;
      +};
      +
      +struct repository;
     @@ -257,7 +243,10 @@
       --- a/repository.h
       +++ b/repository.h
      @@
     + #define REPOSITORY_H
     + 
       #include "path.h"
     ++#include "repo-settings.h"
       
       struct config_set;
      +struct repo_settings;
     @@ -268,7 +257,8 @@
       	 */
       	char *submodule_prefix;
       
     -+	struct repo_settings *settings;
     ++	int settings_initialized;
     ++	struct repo_settings settings;
      +
       	/* Subsystems */
       	/*
 2:  e00a1be75b ! 2:  86380c7832 repo-settings: add feature.manyCommits setting
     @@ -4,7 +4,7 @@
      
          When a repo has many commits, it is helpful to write and read the
          commit-graph file. Future changes to Git may include new config
     -    settings that are benefitial in this scenario.
     +    settings that are beneficial in this scenario.
      
          Create the 'feature.manyCommits' config setting that changes the
          default values of 'core.commitGraph' and 'gc.writeCommitGraph' to
     @@ -85,15 +85,17 @@
       
      +#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
      +
     - static int git_repo_config(const char *key, const char *value, void *cb)
     + void prepare_repo_settings(struct repository *r)
       {
     - 	struct repo_settings *rs = (struct repo_settings *)cb;
     + 	int value;
     +@@
     + 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
     + 		r->settings.pack_use_sparse = value;
       
     -+	if (!strcmp(key, "feature.manycommits")) {
     -+		UPDATE_DEFAULT(rs->core_commit_graph, 1);
     -+		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
     -+		return 0;
     ++	if (!repo_config_get_bool(r, "feature.manycommits", &value) && value) {
     ++		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
     ++		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
      +	}
     - 	if (!strcmp(key, "core.commitgraph")) {
     - 		rs->core_commit_graph = git_config_bool(key, value);
     - 		return 0;
     ++
     + 	r->settings_initialized = 1;
     + }
 3:  47ae3e7d4d ! 3:  49be7a7345 repo-settings: parse core.untrackedCache
     @@ -16,10 +16,9 @@
          untracked cache and persist it.
      
          Instead of relying on magic values of -1, 0, and 1, split these
     -    options into bitflags CORE_UNTRACKED_CACHE_KEEP and
     -    CORE_UNTRACKED_CACHE_WRITE. This allows the use of "-1" as a
     +    options into an enum. This allows the use of "-1" as a
          default value. After parsing the config options, if the value is
     -    unset we can initialize it to CORE_UNTRACKED_CACHE_KEEP.
     +    unset we can initialize it to UNTRACKED_CACHE_KEEP.
      
          Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
      
     @@ -52,7 +51,7 @@
       		break;
       	case UC_DISABLE:
      -		if (git_config_get_untracked_cache() == 1)
     -+		if (r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_WRITE)
     ++		if (r->settings.core_untracked_cache == UNTRACKED_CACHE_WRITE)
       			warning(_("core.untrackedCache is set to true; "
       				  "remove or change it, if you really want to "
       				  "disable the untracked cache"));
     @@ -61,7 +60,7 @@
       	case UC_ENABLE:
       	case UC_FORCE:
      -		if (git_config_get_untracked_cache() == 0)
     -+		if (r->settings->core_untracked_cache == 0)
     ++		if (r->settings.core_untracked_cache == UNTRACKED_CACHE_REMOVE)
       			warning(_("core.untrackedCache is set to false; "
       				  "remove or change it, if you really want to "
       				  "enable the untracked cache"));
     @@ -116,7 +115,7 @@
      +
      +	prepare_repo_settings(r);
      +
     -+	if (!(r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_KEEP)) {
     ++	if (r->settings.core_untracked_cache  == UNTRACKED_CACHE_REMOVE) {
       		remove_untracked_cache(istate);
      -		break;
      -	case 1: /* true */
     @@ -127,7 +126,7 @@
      +		return;
       	}
      +
     -+	if (r->settings->core_untracked_cache & CORE_UNTRACKED_CACHE_WRITE)
     ++	if (r->settings.core_untracked_cache == UNTRACKED_CACHE_WRITE)
      +		add_untracked_cache(istate);
       }
       
     @@ -137,39 +136,46 @@
       --- a/repo-settings.c
       +++ b/repo-settings.c
      @@
     - 		rs->index_version = git_config_int(key, value);
     - 		return 0;
     - 	}
     -+	if (!strcmp(key, "core.untrackedcache")) {
     -+		int bool_value = git_parse_maybe_bool(value);
     -+		if (bool_value == 0)
     -+			rs->core_untracked_cache = 0;
     -+		else if (bool_value == 1)
     -+			rs->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP |
     -+						   CORE_UNTRACKED_CACHE_WRITE;
     -+		else if (!strcasecmp(value, "keep"))
     -+			rs->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
     -+		else
     -+			error(_("unknown core.untrackedCache value '%s'; "
     -+				"using 'keep' default value"), value);
     -+		return 0;
     -+	}
     + void prepare_repo_settings(struct repository *r)
     + {
     + 	int value;
     ++	char *strval;
       
     - 	return 1;
     - }
     + 	if (r->settings_initialized)
     + 		return;
      @@
     - 	r->settings->gc_write_commit_graph = -1;
     - 	r->settings->pack_use_sparse = -1;
     - 	r->settings->index_version = -1;
     -+	r->settings->core_untracked_cache = -1;
       
     - 	repo_config(r, git_repo_config, r->settings);
     + 	if (!repo_config_get_bool(r, "index.version", &value))
     + 		r->settings.index_version = value;
     ++	if (!repo_config_get_maybe_bool(r, "core.untrackedcache", &value)) {
     ++		if (value == 0)
     ++			r->settings.core_untracked_cache = UNTRACKED_CACHE_REMOVE;
     ++		else
     ++			r->settings.core_untracked_cache = UNTRACKED_CACHE_WRITE;
     ++	} else if (!repo_config_get_string(r, "core.untrackedcache", &strval)) {
     ++		if (!strcasecmp(strval, "keep"))
     ++			r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
     ++
     ++		free(strval);
     ++	}
      +
     + 
     + 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
     + 		r->settings.pack_use_sparse = value;
     +-
     ++	
     + 	if (!repo_config_get_bool(r, "feature.manycommits", &value) && value) {
     + 		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
     + 		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
     + 	}
     + 
      +	/* Hack for test programs like test-dump-untracked-cache */
      +	if (ignore_untracked_cache_config)
     -+		r->settings->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
     ++		r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
      +	else
     -+		UPDATE_DEFAULT(r->settings->core_untracked_cache, CORE_UNTRACKED_CACHE_KEEP);
     ++		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
     ++
     + 	r->settings_initialized = 1;
       }
      
       diff --git a/repo-settings.h b/repo-settings.h
     @@ -179,15 +185,19 @@
       #ifndef REPO_SETTINGS_H
       #define REPO_SETTINGS_H
       
     -+#define CORE_UNTRACKED_CACHE_WRITE (1 << 0)
     -+#define CORE_UNTRACKED_CACHE_KEEP (1 << 1)
     ++enum untracked_cache_setting {
     ++	UNTRACKED_CACHE_UNSET = -1,
     ++	UNTRACKED_CACHE_REMOVE = 0,
     ++	UNTRACKED_CACHE_KEEP = 1,
     ++	UNTRACKED_CACHE_WRITE = 2
     ++};
      +
       struct repo_settings {
       	int core_commit_graph;
       	int gc_write_commit_graph;
     - 	int pack_use_sparse;
     + 
       	int index_version;
     -+	int core_untracked_cache;
     - };
     ++	enum untracked_cache_setting core_untracked_cache;
       
     - struct repository;
     + 	int pack_use_sparse;
     + };
 4:  63b522a858 ! 4:  86a5a0c589 repo-settings: create feature.manyFiles setting
     @@ -63,18 +63,16 @@
       --- a/repo-settings.c
       +++ b/repo-settings.c
      @@
     - 		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
     - 		return 0;
     + 		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
     + 		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
       	}
     -+	if (!strcmp(key, "feature.manyfiles")) {
     -+		UPDATE_DEFAULT(rs->index_version, 4);
     -+		UPDATE_DEFAULT(rs->core_untracked_cache,
     -+			       CORE_UNTRACKED_CACHE_KEEP | CORE_UNTRACKED_CACHE_WRITE);
     -+		return 0;
     ++	if (!repo_config_get_bool(r, "feature.manyfiles", &value) && value) {
     ++		UPDATE_DEFAULT(r->settings.index_version, 4);
     ++		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
      +	}
     - 	if (!strcmp(key, "core.commitgraph")) {
     - 		rs->core_commit_graph = git_config_bool(key, value);
     - 		return 0;
     + 
     + 	/* Hack for test programs like test-dump-untracked-cache */
     + 	if (ignore_untracked_cache_config)
      
       diff --git a/t/t1600-index.sh b/t/t1600-index.sh
       --- a/t/t1600-index.sh
 5:  da7685936f ! 5:  b8a631683a repo-settings: create feature.experimental setting
     @@ -104,16 +104,6 @@
       
       /**
        * Returns the length of the first line of msg.
     -@@
     - 	o.branch1 = "HEAD";
     - 	their_tree_name = xstrfmt("%.*s", linelen(state->msg), state->msg);
     - 	o.branch2 = their_tree_name;
     --	o.detect_directory_renames = 0;
     -+	prepare_repo_settings(the_repository);
     -+	the_repository->settings->merge_directory_renames = 0;
     - 
     - 	if (state->quiet)
     - 		o.verbosity = 0;
      
       diff --git a/fetch-negotiator.c b/fetch-negotiator.c
       --- a/fetch-negotiator.c
     @@ -140,7 +130,7 @@
      -			die("unknown fetch negotiation algorithm '%s'", algorithm);
      -		}
      +	prepare_repo_settings(r);
     -+	switch(r->settings->fetch_negotiation_algorithm) {
     ++	switch(r->settings.fetch_negotiation_algorithm) {
      +	case FETCH_NEGOTIATION_SKIPPING:
      +		skipping_negotiator_init(negotiator);
      +		return;
     @@ -248,83 +238,6 @@
       
       struct path_hashmap_entry {
       	struct hashmap_entry e;
     -@@
     - 	 * there is no content merge to do; just move the file into the
     - 	 * desired final location.
     - 	 */
     -+	struct repository *r = the_repository;
     - 	const struct rename *ren = ci->ren1;
     - 	const struct diff_filespec *dest = ren->pair->two;
     - 	char *file_path = dest->path;
     --	int mark_conflicted = (opt->detect_directory_renames == 1);
     -+	int mark_conflicted;
     -+
     -+	prepare_repo_settings(r);
     -+	mark_conflicted = (r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_CONFLICT);
     - 	assert(ren->dir_rename_original_dest);
     - 
     - 	if (!opt->call_depth && would_lose_untracked(opt, dest->path)) {
     -@@
     - 				      struct string_list *entries,
     - 				      struct rename_info *ri)
     - {
     -+	struct repository *r = the_repository;
     - 	struct diff_queue_struct *head_pairs, *merge_pairs;
     - 	struct hashmap *dir_re_head, *dir_re_merge;
     - 	int clean = 1;
     -@@
     - 	head_pairs = get_diffpairs(opt, common, head);
     - 	merge_pairs = get_diffpairs(opt, common, merge);
     - 
     --	if (opt->detect_directory_renames) {
     -+	prepare_repo_settings(r);
     -+	if (r->settings->merge_directory_renames) {
     - 		dir_re_head = get_directory_renames(head_pairs);
     - 		dir_re_merge = get_directory_renames(merge_pairs);
     - 
     -@@
     - 				const struct diff_filespec *b,
     - 				struct rename_conflict_info *ci)
     - {
     -+	struct repository *r = the_repository;
     - 	struct rename *ren = ci->ren1;
     - 	struct merge_file_info mfi;
     - 	int clean;
     -@@
     - 	clean = handle_content_merge(&mfi, opt, path, was_dirty(opt, path),
     - 				     o, a, b, ci);
     - 
     --	if (clean && opt->detect_directory_renames == 1 &&
     -+	prepare_repo_settings(r);
     -+	if (clean &&
     -+	    r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_CONFLICT &&
     - 	    ren->dir_rename_original_dest) {
     - 		if (update_stages(opt, path,
     - 				  NULL,
     -@@
     - static int warn_about_dir_renamed_entries(struct merge_options *opt,
     - 					  struct rename *ren)
     - {
     -+	struct repository *r = the_repository;
     - 	const char *msg;
     - 	int clean = 1, is_add;
     - 
     -@@
     - 		return clean;
     - 
     - 	/* Sanity checks */
     --	assert(opt->detect_directory_renames > 0);
     -+	prepare_repo_settings(r);
     -+	assert(r->settings->merge_directory_renames > 0);
     - 	assert(ren->dir_rename_original_type == 'A' ||
     - 	       ren->dir_rename_original_type == 'R');
     - 
     - 	/* Check whether to treat directory renames as a conflict */
     --	clean = (opt->detect_directory_renames == 2);
     -+	clean = (r->settings->merge_directory_renames == MERGE_DIRECTORY_RENAMES_TRUE);
     - 
     - 	is_add = (ren->dir_rename_original_type == 'A');
     - 	if (ren->dir_rename_original_type == 'A' && clean) {
      @@
       		opt->merge_detect_rename = git_config_rename("merge.renames", value);
       		free(value);
     @@ -342,115 +255,92 @@
       }
       
      @@
     - 	opt->renormalize = 0;
       	opt->diff_detect_rename = -1;
       	opt->merge_detect_rename = -1;
     --	opt->detect_directory_renames = 1;
     + 	opt->detect_directory_renames = 1;
     ++
     ++	prepare_repo_settings(repo);
     ++	if (repo->settings.merge_directory_renames >= 0)
     ++		opt->detect_directory_renames = repo->settings.merge_directory_renames;
     ++
       	merge_recursive_config(opt);
       	merge_verbosity = getenv("GIT_MERGE_VERBOSITY");
       	if (merge_verbosity)
      
     - diff --git a/merge-recursive.h b/merge-recursive.h
     - --- a/merge-recursive.h
     - +++ b/merge-recursive.h
     -@@
     - 	unsigned renormalize : 1;
     - 	long xdl_opts;
     - 	int verbosity;
     --	int detect_directory_renames;
     - 	int diff_detect_rename;
     - 	int merge_detect_rename;
     - 	int diff_rename_limit;
     -
       diff --git a/repo-settings.c b/repo-settings.c
       --- a/repo-settings.c
       +++ b/repo-settings.c
      @@
     - {
     - 	struct repo_settings *rs = (struct repo_settings *)cb;
     - 
     -+	if (!strcmp(key, "feature.experimental")) {
     -+		UPDATE_DEFAULT(rs->pack_use_sparse, 1);
     -+		UPDATE_DEFAULT(rs->merge_directory_renames, MERGE_DIRECTORY_RENAMES_TRUE);
     -+		UPDATE_DEFAULT(rs->fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
     -+		return 0;
     -+	}
     - 	if (!strcmp(key, "feature.manycommits")) {
     - 		UPDATE_DEFAULT(rs->core_commit_graph, 1);
     - 		UPDATE_DEFAULT(rs->gc_write_commit_graph, 1);
     -@@
     - 				"using 'keep' default value"), value);
     - 		return 0;
     + 		free(strval);
       	}
     -+	if (!strcmp(key, "merge.directoryrenames")) {
     -+		int bool_value = git_parse_maybe_bool(value);
     -+		if (0 <= bool_value) {
     -+			rs->merge_directory_renames = bool_value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
     -+		} else if (!strcasecmp(value, "conflict")) {
     -+			rs->merge_directory_renames = MERGE_DIRECTORY_RENAMES_CONFLICT;
     -+		}
     -+		return 0;
     + 
     ++	if (!repo_config_get_maybe_bool(r, "merge.directoryrenames", &value))
     ++		r->settings.merge_directory_renames = value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
     ++	else if (!repo_config_get_string(r, "merge.directoryrenames", &strval)) {
     ++		if (!strcasecmp(strval, "conflict"))
     ++			r->settings.merge_directory_renames = MERGE_DIRECTORY_RENAMES_CONFLICT;
      +	}
     -+	if (!strcmp(key, "fetch.negotiationalgorithm"))	{
     -+		if (!strcasecmp(value, "skipping")) {
     -+			rs->fetch_negotiation_algorithm = FETCH_NEGOTIATION_SKIPPING;
     -+		} else {
     -+			rs->fetch_negotiation_algorithm = FETCH_NEGOTIATION_DEFAULT;
     -+		}
     -+		return 0;
     ++	if (!repo_config_get_string(r, "fetch.negotiationalgorithm", &strval)) {
     ++		if (!strcasecmp(strval, "skipping"))
     ++			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_SKIPPING;
     ++		else
     ++			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_DEFAULT;
      +	}
       
     - 	return 1;
     - }
     + 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
     + 		r->settings.pack_use_sparse = value;
      @@
     - 	/* Defaults */
     - 	r->settings->core_commit_graph = -1;
     - 	r->settings->gc_write_commit_graph = -1;
     --	r->settings->pack_use_sparse = -1;
     -+
     - 	r->settings->index_version = -1;
     - 	r->settings->core_untracked_cache = -1;
     - 
     -+	r->settings->pack_use_sparse = -1;
     -+	r->settings->merge_directory_renames = -1;
     -+	r->settings->fetch_negotiation_algorithm = -1;
     -+
     - 	repo_config(r, git_repo_config, r->settings);
     + 		UPDATE_DEFAULT(r->settings.index_version, 4);
     + 		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
     + 	}
     ++	if (!repo_config_get_bool(r, "feature.experimental", &value) && value) {
     ++		UPDATE_DEFAULT(r->settings.pack_use_sparse, 1);
     ++		UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_TRUE);
     ++		UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
     ++	}
       
       	/* Hack for test programs like test-dump-untracked-cache */
     + 	if (ignore_untracked_cache_config)
      @@
     - 		r->settings->core_untracked_cache = CORE_UNTRACKED_CACHE_KEEP;
       	else
     - 		UPDATE_DEFAULT(r->settings->core_untracked_cache, CORE_UNTRACKED_CACHE_KEEP);
     + 		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
     + 
     ++	UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_CONFLICT);
     ++	UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
      +
     -+	UPDATE_DEFAULT(r->settings->merge_directory_renames, MERGE_DIRECTORY_RENAMES_CONFLICT);
     -+	UPDATE_DEFAULT(r->settings->fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
     + 	r->settings_initialized = 1;
       }
      
       diff --git a/repo-settings.h b/repo-settings.h
       --- a/repo-settings.h
       +++ b/repo-settings.h
      @@
     - #define CORE_UNTRACKED_CACHE_WRITE (1 << 0)
     - #define CORE_UNTRACKED_CACHE_KEEP (1 << 1)
     + 	UNTRACKED_CACHE_WRITE = 2
     + };
       
     -+#define MERGE_DIRECTORY_RENAMES_CONFLICT 1
     -+#define MERGE_DIRECTORY_RENAMES_TRUE 2
     ++enum merge_directory_renames_setting {
     ++	MERGE_DIRECTORY_RENAMES_UNSET = -1,
     ++	MERGE_DIRECTORY_RENAMES_NONE = 0,
     ++	MERGE_DIRECTORY_RENAMES_CONFLICT = 1,
     ++	MERGE_DIRECTORY_RENAMES_TRUE = 2,
     ++};
      +
     -+#define FETCH_NEGOTIATION_DEFAULT 1
     -+#define FETCH_NEGOTIATION_SKIPPING 2
     ++enum fetch_negotiation_setting {
     ++	FETCH_NEGOTIATION_UNSET = -1,
     ++	FETCH_NEGOTIATION_NONE = 0,
     ++	FETCH_NEGOTIATION_DEFAULT = 1,
     ++	FETCH_NEGOTIATION_SKIPPING = 2,
     ++};
      +
       struct repo_settings {
       	int core_commit_graph;
       	int gc_write_commit_graph;
     --	int pack_use_sparse;
     -+
     - 	int index_version;
     - 	int core_untracked_cache;
     -+
     -+	int pack_use_sparse;
     -+	int merge_directory_renames;
     -+	int fetch_negotiation_algorithm;
     +@@
     + 	enum untracked_cache_setting core_untracked_cache;
     + 
     + 	int pack_use_sparse;
     ++	enum merge_directory_renames_setting merge_directory_renames;
     ++	enum fetch_negotiation_setting fetch_negotiation_algorithm;
       };
       
       struct repository;

-- 
gitgitgadget

[PATCH v2 1/5] repo-settings: consolidate some config settings

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

There are a few important config settings that are not loaded
during git_default_config. These are instead loaded on-demand.

Centralize these config options to a single scan, and store
all of the values in a repo_settings struct. The values for
each setting are initialized as negative to indicate "unset".

This centralization will be particularly important in a later
change to introduce "meta" config settings that change the
defaults for these config settings.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Makefile               |  1 +
 builtin/gc.c           | 13 ++++++-------
 builtin/pack-objects.c |  9 +++++----
 commit-graph.c         |  7 ++++---
 read-cache.c           | 12 +++++++-----
 repo-settings.c        | 28 ++++++++++++++++++++++++++++
 repo-settings.h        | 17 +++++++++++++++++
 repository.h           |  5 +++++
 8 files changed, 73 insertions(+), 19 deletions(-)
 create mode 100644 repo-settings.c
 create mode 100644 repo-settings.h

diff --git a/Makefile b/Makefile
index 11ccea4071..032fe9b5f9 100644
--- a/Makefile
+++ b/Makefile
@@ -964,6 +964,7 @@ LIB_OBJS += refspec.o
 LIB_OBJS += ref-filter.o
 LIB_OBJS += remote.o
 LIB_OBJS += replace-object.o
+LIB_OBJS += repo-settings.o
 LIB_OBJS += repository.o
 LIB_OBJS += rerere.o
 LIB_OBJS += resolve-undo.o
diff --git a/builtin/gc.c b/builtin/gc.c
index c18efadda5..5e365049aa 100644
--- a/builtin/gc.c
+++ b/builtin/gc.c
@@ -27,6 +27,7 @@
 #include "pack-objects.h"
 #include "blob.h"
 #include "tree.h"
+#include "repo-settings.h"
 
 #define FAILED_RUN "failed to run %s"
 
@@ -41,7 +42,6 @@ static int aggressive_depth = 50;
 static int aggressive_window = 250;
 static int gc_auto_threshold = 6700;
 static int gc_auto_pack_limit = 50;
-static int gc_write_commit_graph;
 static int detach_auto = 1;
 static timestamp_t gc_log_expire_time;
 static const char *gc_log_expire = "1.day.ago";
@@ -148,7 +148,6 @@ static void gc_config(void)
 	git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 	git_config_get_int("gc.auto", &gc_auto_threshold);
 	git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
-	git_config_get_bool("gc.writecommitgraph", &gc_write_commit_graph);
 	git_config_get_bool("gc.autodetach", &detach_auto);
 	git_config_get_expiry("gc.pruneexpire", &prune_expire);
 	git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
@@ -685,11 +684,11 @@ int cmd_gc(int argc, const char **argv, const char *prefix)
 		clean_pack_garbage();
 	}
 
-	if (gc_write_commit_graph &&
-	    write_commit_graph_reachable(get_object_directory(),
-					 !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
-					 NULL))
-		return 1;
+	prepare_repo_settings(the_repository);
+	if (the_repository->settings.gc_write_commit_graph == 1)
+		write_commit_graph_reachable(get_object_directory(),
+					     !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
+					     NULL);
 
 	if (auto_gc && too_many_loose_objects())
 		warning(_("There are too many unreachable loose objects; "
diff --git a/builtin/pack-objects.c b/builtin/pack-objects.c
index 267c562b1f..20bcd57033 100644
--- a/builtin/pack-objects.c
+++ b/builtin/pack-objects.c
@@ -34,6 +34,7 @@
 #include "dir.h"
 #include "midx.h"
 #include "trace2.h"
+#include "repo-settings.h"
 
 #define IN_PACK(obj) oe_in_pack(&to_pack, obj)
 #define SIZE(obj) oe_size(&to_pack, obj)
@@ -2709,10 +2710,6 @@ static int git_pack_config(const char *k, const char *v, void *cb)
 		use_bitmap_index_default = git_config_bool(k, v);
 		return 0;
 	}
-	if (!strcmp(k, "pack.usesparse")) {
-		sparse = git_config_bool(k, v);
-		return 0;
-	}
 	if (!strcmp(k, "pack.threads")) {
 		delta_search_threads = git_config_int(k, v);
 		if (delta_search_threads < 0)
@@ -3332,6 +3329,10 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 	read_replace_refs = 0;
 
 	sparse = git_env_bool("GIT_TEST_PACK_SPARSE", 0);
+	prepare_repo_settings(the_repository);
+	if (!sparse && the_repository->settings.pack_use_sparse != -1)
+		sparse = the_repository->settings.pack_use_sparse;
+
 	reset_pack_idx_option(&pack_idx_opts);
 	git_config(git_pack_config, NULL);
 
diff --git a/commit-graph.c b/commit-graph.c
index b3c4de79b6..0c2dadab9b 100644
--- a/commit-graph.c
+++ b/commit-graph.c
@@ -16,6 +16,7 @@
 #include "hashmap.h"
 #include "replace-object.h"
 #include "progress.h"
+#include "repo-settings.h"
 
 #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
 #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
@@ -466,7 +467,6 @@ static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 static int prepare_commit_graph(struct repository *r)
 {
 	struct object_directory *odb;
-	int config_value;
 
 	if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
 		die("dying as requested by the '%s' variable on commit-graph load!",
@@ -476,9 +476,10 @@ static int prepare_commit_graph(struct repository *r)
 		return !!r->objects->commit_graph;
 	r->objects->commit_graph_attempted = 1;
 
+	prepare_repo_settings(r);
+
 	if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
-	    (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
-	    !config_value))
+	    r->settings.core_commit_graph != 1)
 		/*
 		 * This repository is not configured to use commit graphs, so
 		 * do not load one. (But report commit_graph_attempted anyway
diff --git a/read-cache.c b/read-cache.c
index c701f7f8b8..bda665a25a 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -25,6 +25,7 @@
 #include "fsmonitor.h"
 #include "thread-utils.h"
 #include "progress.h"
+#include "repo-settings.h"
 
 /* Mask for the name length in ce_flags in the on-disk index */
 
@@ -1599,16 +1600,17 @@ struct cache_entry *refresh_cache_entry(struct index_state *istate,
 
 #define INDEX_FORMAT_DEFAULT 3
 
-static unsigned int get_index_format_default(void)
+static unsigned int get_index_format_default(struct repository *r)
 {
 	char *envversion = getenv("GIT_INDEX_VERSION");
 	char *endp;
-	int value;
 	unsigned int version = INDEX_FORMAT_DEFAULT;
 
 	if (!envversion) {
-		if (!git_config_get_int("index.version", &value))
-			version = value;
+		prepare_repo_settings(r);
+
+		if (r->settings.index_version >= 0)
+			version = r->settings.index_version;
 		if (version < INDEX_FORMAT_LB || INDEX_FORMAT_UB < version) {
 			warning(_("index.version set, but the value is invalid.\n"
 				  "Using version %i"), INDEX_FORMAT_DEFAULT);
@@ -2765,7 +2767,7 @@ static int do_write_index(struct index_state *istate, struct tempfile *tempfile,
 	}
 
 	if (!istate->version) {
-		istate->version = get_index_format_default();
+		istate->version = get_index_format_default(the_repository);
 		if (git_env_bool("GIT_TEST_SPLIT_INDEX", 0))
 			init_split_index(istate);
 	}
diff --git a/repo-settings.c b/repo-settings.c
new file mode 100644
index 0000000000..2c33810cc6
--- /dev/null
+++ b/repo-settings.c
@@ -0,0 +1,28 @@
+#include "cache.h"
+#include "repository.h"
+#include "config.h"
+#include "repo-settings.h"
+
+void prepare_repo_settings(struct repository *r)
+{
+	int value;
+
+	if (r->settings_initialized)
+		return;
+
+	/* Defaults */
+	memset(&r->settings, -1, sizeof(r->settings));
+
+	if (!repo_config_get_bool(r, "core.commitgraph", &value))
+		r->settings.core_commit_graph = value;
+	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
+		r->settings.gc_write_commit_graph = value;
+
+	if (!repo_config_get_bool(r, "index.version", &value))
+		r->settings.index_version = value;
+
+	if (!repo_config_get_bool(r, "pack.usesparse", &value))
+		r->settings.pack_use_sparse = value;
+
+	r->settings_initialized = 1;
+}
diff --git a/repo-settings.h b/repo-settings.h
new file mode 100644
index 0000000000..89fb0159bf
--- /dev/null
+++ b/repo-settings.h
@@ -0,0 +1,17 @@
+#ifndef REPO_SETTINGS_H
+#define REPO_SETTINGS_H
+
+struct repo_settings {
+	int core_commit_graph;
+	int gc_write_commit_graph;
+
+	int index_version;
+
+	int pack_use_sparse;
+};
+
+struct repository;
+
+void prepare_repo_settings(struct repository *r);
+
+#endif /* REPO_SETTINGS_H */
diff --git a/repository.h b/repository.h
index 4fb6a5885f..a817486825 100644
--- a/repository.h
+++ b/repository.h
@@ -2,8 +2,10 @@
 #define REPOSITORY_H
 
 #include "path.h"
+#include "repo-settings.h"
 
 struct config_set;
+struct repo_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -72,6 +74,9 @@ struct repository {
 	 */
 	char *submodule_prefix;
 
+	int settings_initialized;
+	struct repo_settings settings;
+
 	/* Subsystems */
 	/*
 	 * Repository's config which contains key-value pairs from the usual
-- 
gitgitgadget

[PATCH v2 2/5] repo-settings: add feature.manyCommits setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

When a repo has many commits, it is helpful to write and read the
commit-graph file. Future changes to Git may include new config
settings that are beneficial in this scenario.

Create the 'feature.manyCommits' config setting that changes the
default values of 'core.commitGraph' and 'gc.writeCommitGraph' to
true.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config.txt         |  2 ++
 Documentation/config/core.txt    |  3 ++-
 Documentation/config/feature.txt | 15 +++++++++++++++
 Documentation/config/gc.txt      |  4 ++--
 repo-settings.c                  |  7 +++++++
 5 files changed, 28 insertions(+), 3 deletions(-)
 create mode 100644 Documentation/config/feature.txt

diff --git a/Documentation/config.txt b/Documentation/config.txt
index e3f5bc3396..77f3b1486b 100644
--- a/Documentation/config.txt
+++ b/Documentation/config.txt
@@ -345,6 +345,8 @@ include::config/difftool.txt[]
 
 include::config/fastimport.txt[]
 
+include::config/feature.txt[]
+
 include::config/fetch.txt[]
 
 include::config/format.txt[]
diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index 75538d27e7..d80162681a 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -577,7 +577,8 @@ the `GIT_NOTES_REF` environment variable.  See linkgit:git-notes[1].
 
 core.commitGraph::
 	If true, then git will read the commit-graph file (if it exists)
-	to parse the graph structure of commits. Defaults to false. See
+	to parse the graph structure of commits. Defaults to false, unless
+	`feature.manyCommits` is enabled. See
 	linkgit:git-commit-graph[1] for more information.
 
 core.useReplaceRefs::
diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
new file mode 100644
index 0000000000..f74314ae90
--- /dev/null
+++ b/Documentation/config/feature.txt
@@ -0,0 +1,15 @@
+feature.*::
+	The config settings that start with `feature.` modify the defaults of
+	a group of other config settings. These groups are created by the Git
+	developer community as recommended defaults and are subject to change.
+	In particular, new config options may be added with different defaults.
+
+feature.manyCommits::
+	Enable config options that optimize for repos with many commits. This
+	setting is recommended for repos with at least 100,000 commits. The
+	new default values are:
++
+* `core.commitGraph=true` enables reading the commit-graph file.
++
+* `gc.writeCommitGraph=true` enables writing the commit-graph file during
+garbage collection.
\ No newline at end of file
diff --git a/Documentation/config/gc.txt b/Documentation/config/gc.txt
index 02b92b18b5..31a5fc4f75 100644
--- a/Documentation/config/gc.txt
+++ b/Documentation/config/gc.txt
@@ -63,8 +63,8 @@ gc.writeCommitGraph::
 	If true, then gc will rewrite the commit-graph file when
 	linkgit:git-gc[1] is run. When using `git gc --auto`
 	the commit-graph will be updated if housekeeping is
-	required. Default is false. See linkgit:git-commit-graph[1]
-	for details.
+	required. Default is false, unless `feature.manyCommits`
+	is enabled. See linkgit:git-commit-graph[1] for details.
 
 gc.logExpiry::
 	If the file gc.log exists, then `git gc --auto` will print
diff --git a/repo-settings.c b/repo-settings.c
index 2c33810cc6..3face7d8b9 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -3,6 +3,8 @@
 #include "config.h"
 #include "repo-settings.h"
 
+#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
+
 void prepare_repo_settings(struct repository *r)
 {
 	int value;
@@ -24,5 +26,10 @@ void prepare_repo_settings(struct repository *r)
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
 
+	if (!repo_config_get_bool(r, "feature.manycommits", &value) && value) {
+		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
+		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
+	}
+
 	r->settings_initialized = 1;
 }
-- 
gitgitgadget

[PATCH v2 3/5] repo-settings: parse core.untrackedCache

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The core.untrackedCache config setting is slightly complicated,
so clarify its use and centralize its parsing into the repo
settings.

The default value is "keep" (returned as -1), which persists the
untracked cache if it exists.

If the value is set as "false" (returned as 0), then remove the
untracked cache if it exists.

If the value is set as "true" (returned as 1), then write the
untracked cache and persist it.

Instead of relying on magic values of -1, 0, and 1, split these
options into an enum. This allows the use of "-1" as a
default value. After parsing the config options, if the value is
unset we can initialize it to UNTRACKED_CACHE_KEEP.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 builtin/update-index.c |  7 +++++--
 config.c               | 24 ------------------------
 read-cache.c           | 19 +++++++++----------
 repo-settings.c        | 21 ++++++++++++++++++++-
 repo-settings.h        |  8 ++++++++
 5 files changed, 42 insertions(+), 37 deletions(-)

diff --git a/builtin/update-index.c b/builtin/update-index.c
index dff2f4b837..5ced51c1ee 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -18,6 +18,7 @@
 #include "dir.h"
 #include "split-index.h"
 #include "fsmonitor.h"
+#include "repo-settings.h"
 
 /*
  * Default to not allowing changes to the list of files. The
@@ -966,6 +967,7 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	struct parse_opt_ctx_t ctx;
 	strbuf_getline_fn getline_fn;
 	int parseopt_state = PARSE_OPT_UNKNOWN;
+	struct repository *r = the_repository;
 	struct option options[] = {
 		OPT_BIT('q', NULL, &refresh_args.flags,
 			N_("continue refresh even when index needs update"),
@@ -1180,11 +1182,12 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 		remove_split_index(&the_index);
 	}
 
+	prepare_repo_settings(r);
 	switch (untracked_cache) {
 	case UC_UNSPECIFIED:
 		break;
 	case UC_DISABLE:
-		if (git_config_get_untracked_cache() == 1)
+		if (r->settings.core_untracked_cache == UNTRACKED_CACHE_WRITE)
 			warning(_("core.untrackedCache is set to true; "
 				  "remove or change it, if you really want to "
 				  "disable the untracked cache"));
@@ -1196,7 +1199,7 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 		return !test_if_untracked_cache_is_supported();
 	case UC_ENABLE:
 	case UC_FORCE:
-		if (git_config_get_untracked_cache() == 0)
+		if (r->settings.core_untracked_cache == UNTRACKED_CACHE_REMOVE)
 			warning(_("core.untrackedCache is set to false; "
 				  "remove or change it, if you really want to "
 				  "enable the untracked cache"));
diff --git a/config.c b/config.c
index faa57e436c..3241dbc54d 100644
--- a/config.c
+++ b/config.c
@@ -2277,30 +2277,6 @@ int git_config_get_expiry_in_days(const char *key, timestamp_t *expiry, timestam
 	return -1; /* thing exists but cannot be parsed */
 }
 
-int git_config_get_untracked_cache(void)
-{
-	int val = -1;
-	const char *v;
-
-	/* Hack for test programs like test-dump-untracked-cache */
-	if (ignore_untracked_cache_config)
-		return -1;
-
-	if (!git_config_get_maybe_bool("core.untrackedcache", &val))
-		return val;
-
-	if (!git_config_get_value("core.untrackedcache", &v)) {
-		if (!strcasecmp(v, "keep"))
-			return -1;
-
-		error(_("unknown core.untrackedCache value '%s'; "
-			"using 'keep' default value"), v);
-		return -1;
-	}
-
-	return -1; /* default value */
-}
-
 int git_config_get_split_index(void)
 {
 	int val;
diff --git a/read-cache.c b/read-cache.c
index bda665a25a..3761b9a171 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -1846,18 +1846,17 @@ static void check_ce_order(struct index_state *istate)
 
 static void tweak_untracked_cache(struct index_state *istate)
 {
-	switch (git_config_get_untracked_cache()) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
+	struct repository *r = the_repository;
+
+	prepare_repo_settings(r);
+
+	if (r->settings.core_untracked_cache  == UNTRACKED_CACHE_REMOVE) {
 		remove_untracked_cache(istate);
-		break;
-	case 1: /* true */
-		add_untracked_cache(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
+		return;
 	}
+
+	if (r->settings.core_untracked_cache == UNTRACKED_CACHE_WRITE)
+		add_untracked_cache(istate);
 }
 
 static void tweak_split_index(struct index_state *istate)
diff --git a/repo-settings.c b/repo-settings.c
index 3face7d8b9..65b980403c 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -8,6 +8,7 @@
 void prepare_repo_settings(struct repository *r)
 {
 	int value;
+	char *strval;
 
 	if (r->settings_initialized)
 		return;
@@ -22,14 +23,32 @@ void prepare_repo_settings(struct repository *r)
 
 	if (!repo_config_get_bool(r, "index.version", &value))
 		r->settings.index_version = value;
+	if (!repo_config_get_maybe_bool(r, "core.untrackedcache", &value)) {
+		if (value == 0)
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_REMOVE;
+		else
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_WRITE;
+	} else if (!repo_config_get_string(r, "core.untrackedcache", &strval)) {
+		if (!strcasecmp(strval, "keep"))
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
+
+		free(strval);
+	}
+
 
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
-
+	
 	if (!repo_config_get_bool(r, "feature.manycommits", &value) && value) {
 		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
 		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
 	}
 
+	/* Hack for test programs like test-dump-untracked-cache */
+	if (ignore_untracked_cache_config)
+		r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
+	else
+		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
+
 	r->settings_initialized = 1;
 }
diff --git a/repo-settings.h b/repo-settings.h
index 89fb0159bf..e9b028e1a9 100644
--- a/repo-settings.h
+++ b/repo-settings.h
@@ -1,11 +1,19 @@
 #ifndef REPO_SETTINGS_H
 #define REPO_SETTINGS_H
 
+enum untracked_cache_setting {
+	UNTRACKED_CACHE_UNSET = -1,
+	UNTRACKED_CACHE_REMOVE = 0,
+	UNTRACKED_CACHE_KEEP = 1,
+	UNTRACKED_CACHE_WRITE = 2
+};
+
 struct repo_settings {
 	int core_commit_graph;
 	int gc_write_commit_graph;
 
 	int index_version;
+	enum untracked_cache_setting core_untracked_cache;
 
 	int pack_use_sparse;
 };
-- 
gitgitgadget

[PATCH v2 4/5] repo-settings: create feature.manyFiles setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The feature.manyFiles setting is suitable for repos with many
files in the working directory. By setting index.version=4 and
core.untrackedCache=true, commands such as 'git status' should
improve.

While adding this setting, modify the index version precedence
tests to check how this setting overrides the default for
index.version is unset.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config/core.txt    |  4 +++-
 Documentation/config/feature.txt | 12 +++++++++++-
 Documentation/config/index.txt   |  1 +
 repo-settings.c                  |  4 ++++
 t/t1600-index.sh                 | 31 ++++++++++++++++++++++++++-----
 5 files changed, 45 insertions(+), 7 deletions(-)

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index d80162681a..7a2a33bc8c 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -86,7 +86,9 @@ core.untrackedCache::
 	it will automatically be removed, if set to `false`. Before
 	setting it to `true`, you should check that mtime is working
 	properly on your system.
-	See linkgit:git-update-index[1]. `keep` by default.
+	See linkgit:git-update-index[1]. `keep` by default, unless
+	`feature.manyFiles` is enabled which sets this setting to
+	`true` by default.
 
 core.checkStat::
 	When missing or is set to `default`, many fields in the stat
diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
index f74314ae90..c2d9ef7473 100644
--- a/Documentation/config/feature.txt
+++ b/Documentation/config/feature.txt
@@ -12,4 +12,14 @@ feature.manyCommits::
 * `core.commitGraph=true` enables reading the commit-graph file.
 +
 * `gc.writeCommitGraph=true` enables writing the commit-graph file during
-garbage collection.
\ No newline at end of file
+garbage collection.
+
+feature.manyFiles::
+	Enable config options that optimize for repos with many files in the
+	working directory. With many files, commands such as `git status` and
+	`git checkout` may be slow and these new defaults improve performance:
++
+* `index.version=4` enables path-prefix compression in the index.
++
+* `core.untrackedCache=true` enables the untracked cache. This setting assumes
+that mtime is working on your machine.
\ No newline at end of file
diff --git a/Documentation/config/index.txt b/Documentation/config/index.txt
index f181503041..7cb50b37e9 100644
--- a/Documentation/config/index.txt
+++ b/Documentation/config/index.txt
@@ -24,3 +24,4 @@ index.threads::
 index.version::
 	Specify the version with which new index files should be
 	initialized.  This does not affect existing repositories.
+	If `feature.manyFiles` is enabled, then the default is 4.
diff --git a/repo-settings.c b/repo-settings.c
index 65b980403c..3a5cf33814 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -43,6 +43,10 @@ void prepare_repo_settings(struct repository *r)
 		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
 		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
 	}
+	if (!repo_config_get_bool(r, "feature.manyfiles", &value) && value) {
+		UPDATE_DEFAULT(r->settings.index_version, 4);
+		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
+	}
 
 	/* Hack for test programs like test-dump-untracked-cache */
 	if (ignore_untracked_cache_config)
diff --git a/t/t1600-index.sh b/t/t1600-index.sh
index 42962ed7d4..c77721b580 100755
--- a/t/t1600-index.sh
+++ b/t/t1600-index.sh
@@ -59,17 +59,38 @@ test_expect_success 'out of bounds index.version issues warning' '
 	)
 '
 
-test_expect_success 'GIT_INDEX_VERSION takes precedence over config' '
+test_index_version () {
+	INDEX_VERSION_CONFIG=$1 &&
+	FEATURE_MANY_FILES=$2 &&
+	ENV_VAR_VERSION=$3
+	EXPECTED_OUTPUT_VERSION=$4 &&
 	(
 		rm -f .git/index &&
-		GIT_INDEX_VERSION=4 &&
-		export GIT_INDEX_VERSION &&
-		git config --add index.version 2 &&
+		rm -f .git/config &&
+		if test "$INDEX_VERSION_CONFIG" -ne 0
+		then
+			git config --add index.version $INDEX_VERSION_CONFIG
+		fi &&
+		git config --add feature.manyFiles $FEATURE_MANY_FILES
+		if test "$ENV_VAR_VERSION" -ne 0
+		then
+			GIT_INDEX_VERSION=$ENV_VAR_VERSION &&
+			export GIT_INDEX_VERSION
+		else
+			unset GIT_INDEX_VERSION
+		fi &&
 		git add a 2>&1 &&
-		echo 4 >expect &&
+		echo $EXPECTED_OUTPUT_VERSION >expect &&
 		test-tool index-version <.git/index >actual &&
 		test_cmp expect actual
 	)
+}
+
+test_expect_success 'index version config precedence' '
+	test_index_version 2 false 4 4 &&
+	test_index_version 2 true 0 2 &&
+	test_index_version 0 true 0 4 &&
+	test_index_version 0 true 2 2
 '
 
 test_done
-- 
gitgitgadget

[PATCH v2 5/5] repo-settings: create feature.experimental setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The 'feature.experimental' setting includes config options that are
not committed to become defaults, but could use additional testing.

Update the following config settings to take new defaults, and to
use the repo_settings struct if not already using them:

* 'pack.useSparse=true'

* 'merge.directoryRenames=true'

* 'fetch.negotiationAlgorithm=skipping'

In the case of fetch.negotiationAlgorithm, the existing logic
would load the config option only when about to use the setting,
so had a die() statement on an unknown string value. This is
removed as now the config is parsed under prepare_repo_settings().
In general, this die() is probably misplaced and not valuable.
A test was removed that checked this die() statement executed.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config/feature.txt     | 17 +++++++++++++++++
 Documentation/config/fetch.txt       |  3 ++-
 Documentation/config/merge.txt       |  3 ++-
 Documentation/config/pack.txt        |  3 ++-
 builtin/am.c                         |  1 +
 fetch-negotiator.c                   | 26 ++++++++++++++------------
 fetch-negotiator.h                   |  5 +++--
 fetch-pack.c                         | 11 +++++------
 merge-recursive.c                    | 15 ++++++---------
 repo-settings.c                      | 20 ++++++++++++++++++++
 repo-settings.h                      | 16 ++++++++++++++++
 t/t5552-skipping-fetch-negotiator.sh | 23 -----------------------
 12 files changed, 88 insertions(+), 55 deletions(-)

diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
index c2d9ef7473..287621b1e8 100644
--- a/Documentation/config/feature.txt
+++ b/Documentation/config/feature.txt
@@ -4,6 +4,23 @@ feature.*::
 	developer community as recommended defaults and are subject to change.
 	In particular, new config options may be added with different defaults.
 
+feature.experimental::
+	Enable config options that are new to Git, and are being considered for
+	future defaults. Config settings included here may be added or removed
+	with each release, including minor version updates. These settings may
+	have unintended interactions since they are so new. Please enable this
+	setting if you are interested in providing feedback on experimental
+	features. The new default values are:
++
+* `pack.useSparse=true` uses a new algorithm when constructing a pack-file
+which can improve `git push` performance in repos with many files.
++
+* `merge.directoryRenames=true` uses a new algorithm for detecting renames by
+using entire directories at a time instead of single files at a time.
++
+* `fetch.negotiationAlgorithm=skipping` may improve fetch negotiation times by
+skipping more commits at a time, reducing the number of round trips.
+
 feature.manyCommits::
 	Enable config options that optimize for repos with many commits. This
 	setting is recommended for repos with at least 100,000 commits. The
diff --git a/Documentation/config/fetch.txt b/Documentation/config/fetch.txt
index ba890b5884..d402110638 100644
--- a/Documentation/config/fetch.txt
+++ b/Documentation/config/fetch.txt
@@ -59,7 +59,8 @@ fetch.negotiationAlgorithm::
 	effort to converge faster, but may result in a larger-than-necessary
 	packfile; The default is "default" which instructs Git to use the default algorithm
 	that never skips commits (unless the server has acknowledged it or one
-	of its descendants).
+	of its descendants). If `feature.experimental` is enabled, then this
+	setting defaults to "skipping".
 	Unknown values will cause 'git fetch' to error out.
 +
 See also the `--negotiation-tip` option for linkgit:git-fetch[1].
diff --git a/Documentation/config/merge.txt b/Documentation/config/merge.txt
index 6a313937f8..e8def2d63c 100644
--- a/Documentation/config/merge.txt
+++ b/Documentation/config/merge.txt
@@ -54,7 +54,8 @@ merge.directoryRenames::
 	moved into the new directory.  If set to "conflict", a conflict
 	will be reported for such paths.  If merge.renames is false,
 	merge.directoryRenames is ignored and treated as false.  Defaults
-	to "conflict".
+	to "conflict" unless `feature.experimental` is enabled and the
+	default is "true".
 
 merge.renormalize::
 	Tell Git that canonical representation of files in the
diff --git a/Documentation/config/pack.txt b/Documentation/config/pack.txt
index 9cdcfa7324..1d66f0c992 100644
--- a/Documentation/config/pack.txt
+++ b/Documentation/config/pack.txt
@@ -112,7 +112,8 @@ pack.useSparse::
 	objects. This can have significant performance benefits when
 	computing a pack to send a small change. However, it is possible
 	that extra objects are added to the pack-file if the included
-	commits contain certain types of direct renames.
+	commits contain certain types of direct renames. Default is `false`
+	unless `feature.experimental` is enabled.
 
 pack.writeBitmaps (deprecated)::
 	This is a deprecated synonym for `repack.writeBitmaps`.
diff --git a/builtin/am.c b/builtin/am.c
index 1aea657a7f..912c6155b8 100644
--- a/builtin/am.c
+++ b/builtin/am.c
@@ -34,6 +34,7 @@
 #include "string-list.h"
 #include "packfile.h"
 #include "repository.h"
+#include "repo-settings.h"
 
 /**
  * Returns the length of the first line of msg.
diff --git a/fetch-negotiator.c b/fetch-negotiator.c
index d6d685cba0..1a68be1a2e 100644
--- a/fetch-negotiator.c
+++ b/fetch-negotiator.c
@@ -2,19 +2,21 @@
 #include "fetch-negotiator.h"
 #include "negotiator/default.h"
 #include "negotiator/skipping.h"
+#include "repository.h"
+#include "repo-settings.h"
 
-void fetch_negotiator_init(struct fetch_negotiator *negotiator,
-			   const char *algorithm)
+void fetch_negotiator_init(struct repository *r,
+			   struct fetch_negotiator *negotiator)
 {
-	if (algorithm) {
-		if (!strcmp(algorithm, "skipping")) {
-			skipping_negotiator_init(negotiator);
-			return;
-		} else if (!strcmp(algorithm, "default")) {
-			/* Fall through to default initialization */
-		} else {
-			die("unknown fetch negotiation algorithm '%s'", algorithm);
-		}
+	prepare_repo_settings(r);
+	switch(r->settings.fetch_negotiation_algorithm) {
+	case FETCH_NEGOTIATION_SKIPPING:
+		skipping_negotiator_init(negotiator);
+		return;
+
+	case FETCH_NEGOTIATION_DEFAULT:
+	default:
+		default_negotiator_init(negotiator);
+		return;
 	}
-	default_negotiator_init(negotiator);
 }
diff --git a/fetch-negotiator.h b/fetch-negotiator.h
index 9e3967ce66..ea78868504 100644
--- a/fetch-negotiator.h
+++ b/fetch-negotiator.h
@@ -2,6 +2,7 @@
 #define FETCH_NEGOTIATOR_H
 
 struct commit;
+struct repository;
 
 /*
  * An object that supplies the information needed to negotiate the contents of
@@ -52,7 +53,7 @@ struct fetch_negotiator {
 	void *data;
 };
 
-void fetch_negotiator_init(struct fetch_negotiator *negotiator,
-			   const char *algorithm);
+void fetch_negotiator_init(struct repository *r,
+			   struct fetch_negotiator *negotiator);
 
 #endif
diff --git a/fetch-pack.c b/fetch-pack.c
index 65be043f2a..d81f47c07b 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -36,7 +36,6 @@ static int agent_supported;
 static int server_supports_filtering;
 static struct lock_file shallow_lock;
 static const char *alternate_shallow_file;
-static char *negotiation_algorithm;
 static struct strbuf fsck_msg_types = STRBUF_INIT;
 
 /* Remember to update object flag allocation in object.h */
@@ -892,12 +891,13 @@ static struct ref *do_fetch_pack(struct fetch_pack_args *args,
 				 struct shallow_info *si,
 				 char **pack_lockfile)
 {
+	struct repository *r = the_repository;
 	struct ref *ref = copy_ref_list(orig_ref);
 	struct object_id oid;
 	const char *agent_feature;
 	int agent_len;
 	struct fetch_negotiator negotiator;
-	fetch_negotiator_init(&negotiator, negotiation_algorithm);
+	fetch_negotiator_init(r, &negotiator);
 
 	sort_ref_list(&ref, ref_compare_name);
 	QSORT(sought, nr_sought, cmp_ref_by_name);
@@ -911,7 +911,7 @@ static struct ref *do_fetch_pack(struct fetch_pack_args *args,
 
 	if (server_supports("shallow"))
 		print_verbose(args, _("Server supports %s"), "shallow");
-	else if (args->depth > 0 || is_repository_shallow(the_repository))
+	else if (args->depth > 0 || is_repository_shallow(r))
 		die(_("Server does not support shallow clients"));
 	if (args->depth > 0 || args->deepen_since || args->deepen_not)
 		args->deepen = 1;
@@ -1379,6 +1379,7 @@ static struct ref *do_fetch_pack_v2(struct fetch_pack_args *args,
 				    struct shallow_info *si,
 				    char **pack_lockfile)
 {
+	struct repository *r = the_repository;
 	struct ref *ref = copy_ref_list(orig_ref);
 	enum fetch_state state = FETCH_CHECK_LOCAL;
 	struct oidset common = OIDSET_INIT;
@@ -1386,7 +1387,7 @@ static struct ref *do_fetch_pack_v2(struct fetch_pack_args *args,
 	int in_vain = 0;
 	int haves_to_send = INITIAL_FLUSH;
 	struct fetch_negotiator negotiator;
-	fetch_negotiator_init(&negotiator, negotiation_algorithm);
+	fetch_negotiator_init(r, &negotiator);
 	packet_reader_init(&reader, fd[0], NULL, 0,
 			   PACKET_READ_CHOMP_NEWLINE |
 			   PACKET_READ_DIE_ON_ERR_PACKET);
@@ -1505,8 +1506,6 @@ static void fetch_pack_config(void)
 	git_config_get_bool("repack.usedeltabaseoffset", &prefer_ofs_delta);
 	git_config_get_bool("fetch.fsckobjects", &fetch_fsck_objects);
 	git_config_get_bool("transfer.fsckobjects", &transfer_fsck_objects);
-	git_config_get_string("fetch.negotiationalgorithm",
-			      &negotiation_algorithm);
 
 	git_config(fetch_pack_config_cb, NULL);
 }
diff --git a/merge-recursive.c b/merge-recursive.c
index 12300131fc..f39c31c9c4 100644
--- a/merge-recursive.c
+++ b/merge-recursive.c
@@ -28,6 +28,7 @@
 #include "submodule.h"
 #include "revision.h"
 #include "commit-reach.h"
+#include "repo-settings.h"
 
 struct path_hashmap_entry {
 	struct hashmap_entry e;
@@ -3662,15 +3663,6 @@ static void merge_recursive_config(struct merge_options *opt)
 		opt->merge_detect_rename = git_config_rename("merge.renames", value);
 		free(value);
 	}
-	if (!git_config_get_string("merge.directoryrenames", &value)) {
-		int boolval = git_parse_maybe_bool(value);
-		if (0 <= boolval) {
-			opt->detect_directory_renames = boolval ? 2 : 0;
-		} else if (!strcasecmp(value, "conflict")) {
-			opt->detect_directory_renames = 1;
-		} /* avoid erroring on values from future versions of git */
-		free(value);
-	}
 	git_config(git_xmerge_config, NULL);
 }
 
@@ -3688,6 +3680,11 @@ void init_merge_options(struct merge_options *opt,
 	opt->diff_detect_rename = -1;
 	opt->merge_detect_rename = -1;
 	opt->detect_directory_renames = 1;
+
+	prepare_repo_settings(repo);
+	if (repo->settings.merge_directory_renames >= 0)
+		opt->detect_directory_renames = repo->settings.merge_directory_renames;
+
 	merge_recursive_config(opt);
 	merge_verbosity = getenv("GIT_MERGE_VERBOSITY");
 	if (merge_verbosity)
diff --git a/repo-settings.c b/repo-settings.c
index 3a5cf33814..bfb3288fda 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -35,6 +35,18 @@ void prepare_repo_settings(struct repository *r)
 		free(strval);
 	}
 
+	if (!repo_config_get_maybe_bool(r, "merge.directoryrenames", &value))
+		r->settings.merge_directory_renames = value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
+	else if (!repo_config_get_string(r, "merge.directoryrenames", &strval)) {
+		if (!strcasecmp(strval, "conflict"))
+			r->settings.merge_directory_renames = MERGE_DIRECTORY_RENAMES_CONFLICT;
+	}
+	if (!repo_config_get_string(r, "fetch.negotiationalgorithm", &strval)) {
+		if (!strcasecmp(strval, "skipping"))
+			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_SKIPPING;
+		else
+			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_DEFAULT;
+	}
 
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
@@ -47,6 +59,11 @@ void prepare_repo_settings(struct repository *r)
 		UPDATE_DEFAULT(r->settings.index_version, 4);
 		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
 	}
+	if (!repo_config_get_bool(r, "feature.experimental", &value) && value) {
+		UPDATE_DEFAULT(r->settings.pack_use_sparse, 1);
+		UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_TRUE);
+		UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
+	}
 
 	/* Hack for test programs like test-dump-untracked-cache */
 	if (ignore_untracked_cache_config)
@@ -54,5 +71,8 @@ void prepare_repo_settings(struct repository *r)
 	else
 		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
 
+	UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_CONFLICT);
+	UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
+
 	r->settings_initialized = 1;
 }
diff --git a/repo-settings.h b/repo-settings.h
index e9b028e1a9..95d3392df2 100644
--- a/repo-settings.h
+++ b/repo-settings.h
@@ -8,6 +8,20 @@ enum untracked_cache_setting {
 	UNTRACKED_CACHE_WRITE = 2
 };
 
+enum merge_directory_renames_setting {
+	MERGE_DIRECTORY_RENAMES_UNSET = -1,
+	MERGE_DIRECTORY_RENAMES_NONE = 0,
+	MERGE_DIRECTORY_RENAMES_CONFLICT = 1,
+	MERGE_DIRECTORY_RENAMES_TRUE = 2,
+};
+
+enum fetch_negotiation_setting {
+	FETCH_NEGOTIATION_UNSET = -1,
+	FETCH_NEGOTIATION_NONE = 0,
+	FETCH_NEGOTIATION_DEFAULT = 1,
+	FETCH_NEGOTIATION_SKIPPING = 2,
+};
+
 struct repo_settings {
 	int core_commit_graph;
 	int gc_write_commit_graph;
@@ -16,6 +30,8 @@ struct repo_settings {
 	enum untracked_cache_setting core_untracked_cache;
 
 	int pack_use_sparse;
+	enum merge_directory_renames_setting merge_directory_renames;
+	enum fetch_negotiation_setting fetch_negotiation_algorithm;
 };
 
 struct repository;
diff --git a/t/t5552-skipping-fetch-negotiator.sh b/t/t5552-skipping-fetch-negotiator.sh
index 8a14be51a1..f70cbcc9ca 100755
--- a/t/t5552-skipping-fetch-negotiator.sh
+++ b/t/t5552-skipping-fetch-negotiator.sh
@@ -60,29 +60,6 @@ test_expect_success 'commits with no parents are sent regardless of skip distanc
 	have_not_sent c6 c4 c3
 '
 
-test_expect_success 'unknown fetch.negotiationAlgorithm values error out' '
-	rm -rf server client trace &&
-	git init server &&
-	test_commit -C server to_fetch &&
-
-	git init client &&
-	test_commit -C client on_client &&
-	git -C client checkout on_client &&
-
-	test_config -C client fetch.negotiationAlgorithm invalid &&
-	test_must_fail git -C client fetch "$(pwd)/server" 2>err &&
-	test_i18ngrep "unknown fetch negotiation algorithm" err &&
-
-	# Explicit "default" value
-	test_config -C client fetch.negotiationAlgorithm default &&
-	git -C client -c fetch.negotiationAlgorithm=default fetch "$(pwd)/server" &&
-
-	# Implementation detail: If there is nothing to fetch, we will not error out
-	test_config -C client fetch.negotiationAlgorithm invalid &&
-	git -C client fetch "$(pwd)/server" 2>err &&
-	test_i18ngrep ! "unknown fetch negotiation algorithm" err
-'
-
 test_expect_success 'when two skips collide, favor the larger one' '
 	rm -rf server client trace &&
 	git init server &&
-- 
gitgitgadget

Re: [PATCH v2 1/5] repo-settings: consolidate some config settings

[Johannes Schindelin]

Hi Stolee,

On Wed, 24 Jul 2019, Derrick Stolee via GitGitGadget wrote:

> diff --git a/repo-settings.h b/repo-settings.h
> new file mode 100644
> index 0000000000..89fb0159bf
> --- /dev/null
> +++ b/repo-settings.h
> @@ -0,0 +1,17 @@
> +#ifndef REPO_SETTINGS_H
> +#define REPO_SETTINGS_H
> +
> +struct repo_settings {
> +	int core_commit_graph;
> +	int gc_write_commit_graph;
> +
> +	int index_version;
> +
> +	int pack_use_sparse;
> +};
> +
> +struct repository;
> +
> +void prepare_repo_settings(struct repository *r);
> +
> +#endif /* REPO_SETTINGS_H */
> diff --git a/repository.h b/repository.h
> index 4fb6a5885f..a817486825 100644
> --- a/repository.h
> +++ b/repository.h
> @@ -2,8 +2,10 @@
>  #define REPOSITORY_H
>
>  #include "path.h"
> +#include "repo-settings.h"

I still think that the `repo_settings` struct could just as easily be
declared in `repository.h`. No need to invent a new header file.

>
>  struct config_set;
> +struct repo_settings;

In any case, this is no longer necessary.

>  struct git_hash_algo;
>  struct index_state;
>  struct lock_file;
> @@ -72,6 +74,9 @@ struct repository {
>  	 */
>  	char *submodule_prefix;
>
> +	int settings_initialized;

Or maybe

	unsigned settings_initialized:1;

?

> +	struct repo_settings settings;
> +

Or maybe even fold the `initialized` flag into that struct?

Thanks,
Dscho

>  	/* Subsystems */
>  	/*
>  	 * Repository's config which contains key-value pairs from the usual
> --
> gitgitgadget
>
>

Re: [PATCH v2 3/5] repo-settings: parse core.untrackedCache

[Johannes Schindelin]

Hi Stolee,

On Wed, 24 Jul 2019, Derrick Stolee via GitGitGadget wrote:

> diff --git a/builtin/update-index.c b/builtin/update-index.c
> index dff2f4b837..5ced51c1ee 100644
> --- a/builtin/update-index.c
> +++ b/builtin/update-index.c
> @@ -18,6 +18,7 @@
>  #include "dir.h"
>  #include "split-index.h"
>  #include "fsmonitor.h"
> +#include "repo-settings.h"

Sorry, I only realized 10 seconds ago that these `#include`s are no
longer necessary because `repo-settings` is already included from
`repository.h` (and needs to be, because the `repository` struct now
contains a full `struct repo_settings` field, no longer just a pointer).

But really, I would just fold the `repo-settings.h` declarations
directly into `repository.h` (but keep `repo-settings.c`, of course!) as
it make everything a bit simpler.

Thanks,
Dscho

Re: [PATCH v2 0/5] Create 'feature.*' config area and some centralized config parsing

[Johannes Schindelin]

Hi Stolee,

On Wed, 24 Jul 2019, Derrick Stolee via GitGitGadget wrote:

> Updates in V2: I'm responding to Dscho's comments early, as they were very
> valuable.
>
>  * Rearranged how we are checking for the feature.* settings, so if one
>    layer turns the setting on and a later layer turns it off, we do not
>    adjust the defaults.
>
>
>  * Switched to using enums for the non-boolean config options.
>
>
>  * Placed the repo_settings struct directly in the repository struct.
>
>
>  * All struct members are initialized to -1 using memset.
>
>
>  * The config changes around directory rename detection is greatly
>    simplified.

I left only a couple of comments, and quite honestly, I am really happy
with the direction this is going.

It even leaves the door open to let users define their custom feature
groups, at least I have a couple of ideas how that could be accomplished
building on top of `repo-settings.c`.

Thanks!
Dscho

[PATCH v3 0/5] Create 'feature.*' config area and some centralized config parsing

[Derrick Stolee via GitGitGadget]

This is a brand-new thread to replace ds/early-access. The discussion on
that thread was very helpful [1].

With this in mind, I propose instead a set of "feature.*" config settings
that form groups of "community recommended" settings (with some caveats). In
the space below, I'll list a set of possible feature names and the implied
config options.

First, the main two categories we've discussed so far: many commits and many
files. These two feature sets are for when your repo is large in one of
these dimensions. Perhaps there are other settings to include in these?

feature.manyFiles:
    index.version = 4
    core.untrackedCache = true

feature.manyCommits:
    core.commitGraph = true
    gc.writeCommitGraph = true
    (future: fetch.writeSplitCommitGraph = true)

Note: the fetch.writeSplitCommitGraph does not exist yet, but could be
introduced in a later release to write a new commit-graph (with --split) on
fetch.

The other category that has been discussed already is that of "experimental
features that we generally think are helpful but change behavior slightly in
some cases".

feature.experimental:
    pack.useSparse = true
    merge.directoryRenames = true
    fetch.negotiationAlgorithm = skipping

Specifically, this setting is for config values we are not sure will ever be
on by default, but additional testing is needed to be sure. This is
different than a possible 'feature.preview' setting that would include
config settings that we are committed to updating the defaults in a future
release. There are many ways we can take this idea in the future (including
more additions to these categories).

Thanks, -Stolee

Updates in V2: I'm responding to Dscho's comments early, as they were very
valuable.

 * Rearranged how we are checking for the feature.* settings, so if one
   layer turns the setting on and a later layer turns it off, we do not
   adjust the defaults.
   
   
 * Switched to using enums for the non-boolean config options.
   
   
 * Placed the repo_settings struct directly in the repository struct.
   
   
 * All struct members are initialized to -1 using memset.
   
   
 * The config changes around directory rename detection is greatly
   simplified.
   
   

Updates in V3: These are more Dscho comments, and are quite small.

 * Remove repo-settings.h by placing the struct and function definitions in
   repository.h.
   
   
 * Move the initialization int into "struct repo_settings".
   
   

[1] https://public-inbox.org/git/pull.254.git.gitgitgadget@gmail.com/

Derrick Stolee (5):
  repo-settings: consolidate some config settings
  repo-settings: add feature.manyCommits setting
  repo-settings: parse core.untrackedCache
  repo-settings: create feature.manyFiles setting
  repo-settings: create feature.experimental setting

 Documentation/config.txt             |  2 +
 Documentation/config/core.txt        |  7 ++-
 Documentation/config/feature.txt     | 42 ++++++++++++++++
 Documentation/config/fetch.txt       |  3 +-
 Documentation/config/gc.txt          |  4 +-
 Documentation/config/index.txt       |  1 +
 Documentation/config/merge.txt       |  3 +-
 Documentation/config/pack.txt        |  3 +-
 Makefile                             |  1 +
 builtin/gc.c                         | 12 ++---
 builtin/pack-objects.c               |  8 +--
 builtin/update-index.c               |  6 ++-
 commit-graph.c                       |  6 +--
 config.c                             | 24 ---------
 fetch-negotiator.c                   | 25 +++++-----
 fetch-negotiator.h                   |  5 +-
 fetch-pack.c                         | 11 ++--
 merge-recursive.c                    | 14 ++----
 read-cache.c                         | 30 +++++------
 repo-settings.c                      | 75 ++++++++++++++++++++++++++++
 repository.h                         | 39 +++++++++++++++
 t/t1600-index.sh                     | 31 ++++++++++--
 t/t5552-skipping-fetch-negotiator.sh | 23 ---------
 23 files changed, 256 insertions(+), 119 deletions(-)
 create mode 100644 Documentation/config/feature.txt
 create mode 100644 repo-settings.c


base-commit: 9c9b961d7eb15fb583a2a812088713a68a85f1c0
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-292%2Fderrickstolee%2Frepo-settings%2Fhead-v3
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-292/derrickstolee/repo-settings/head-v3
Pull-Request: https://github.com/gitgitgadget/git/pull/292

Range-diff vs v2:

 1:  597ab7d621 ! 1:  4d0db57ecb repo-settings: consolidate some config settings
     @@ -30,14 +30,6 @@
       diff --git a/builtin/gc.c b/builtin/gc.c
       --- a/builtin/gc.c
       +++ b/builtin/gc.c
     -@@
     - #include "pack-objects.h"
     - #include "blob.h"
     - #include "tree.h"
     -+#include "repo-settings.h"
     - 
     - #define FAILED_RUN "failed to run %s"
     - 
      @@
       static int aggressive_window = 250;
       static int gc_auto_threshold = 6700;
     @@ -75,14 +67,6 @@
       diff --git a/builtin/pack-objects.c b/builtin/pack-objects.c
       --- a/builtin/pack-objects.c
       +++ b/builtin/pack-objects.c
     -@@
     - #include "dir.h"
     - #include "midx.h"
     - #include "trace2.h"
     -+#include "repo-settings.h"
     - 
     - #define IN_PACK(obj) oe_in_pack(&to_pack, obj)
     - #define SIZE(obj) oe_size(&to_pack, obj)
      @@
       		use_bitmap_index_default = git_config_bool(k, v);
       		return 0;
     @@ -109,14 +93,6 @@
       diff --git a/commit-graph.c b/commit-graph.c
       --- a/commit-graph.c
       +++ b/commit-graph.c
     -@@
     - #include "hashmap.h"
     - #include "replace-object.h"
     - #include "progress.h"
     -+#include "repo-settings.h"
     - 
     - #define GRAPH_SIGNATURE 0x43475048 /* "CGPH" */
     - #define GRAPH_CHUNKID_OIDFANOUT 0x4f494446 /* "OIDF" */
      @@
       static int prepare_commit_graph(struct repository *r)
       {
     @@ -142,14 +118,6 @@
       diff --git a/read-cache.c b/read-cache.c
       --- a/read-cache.c
       +++ b/read-cache.c
     -@@
     - #include "fsmonitor.h"
     - #include "thread-utils.h"
     - #include "progress.h"
     -+#include "repo-settings.h"
     - 
     - /* Mask for the name length in ce_flags in the on-disk index */
     - 
      @@
       
       #define INDEX_FORMAT_DEFAULT 3
     @@ -188,15 +156,14 @@
       +++ b/repo-settings.c
      @@
      +#include "cache.h"
     -+#include "repository.h"
      +#include "config.h"
     -+#include "repo-settings.h"
     ++#include "repository.h"
      +
      +void prepare_repo_settings(struct repository *r)
      +{
      +	int value;
      +
     -+	if (r->settings_initialized)
     ++	if (r->settings.initialized)
      +		return;
      +
      +	/* Defaults */
     @@ -212,54 +179,50 @@
      +
      +	if (!repo_config_get_bool(r, "pack.usesparse", &value))
      +		r->settings.pack_use_sparse = value;
     -+
     -+	r->settings_initialized = 1;
      +}
      
     - diff --git a/repo-settings.h b/repo-settings.h
     - new file mode 100644
     - --- /dev/null
     - +++ b/repo-settings.h
     -@@
     -+#ifndef REPO_SETTINGS_H
     -+#define REPO_SETTINGS_H
     -+
     -+struct repo_settings {
     -+	int core_commit_graph;
     -+	int gc_write_commit_graph;
     -+
     -+	int index_version;
     -+
     -+	int pack_use_sparse;
     -+};
     -+
     -+struct repository;
     -+
     -+void prepare_repo_settings(struct repository *r);
     -+
     -+#endif /* REPO_SETTINGS_H */
     -
       diff --git a/repository.h b/repository.h
       --- a/repository.h
       +++ b/repository.h
      @@
     - #define REPOSITORY_H
     - 
       #include "path.h"
     -+#include "repo-settings.h"
       
       struct config_set;
      +struct repo_settings;
       struct git_hash_algo;
       struct index_state;
       struct lock_file;
     +@@
     + struct raw_object_store;
     + struct submodule_cache;
     + 
     ++struct repo_settings {
     ++	int initialized;
     ++
     ++	int core_commit_graph;
     ++	int gc_write_commit_graph;
     ++
     ++	int index_version;
     ++
     ++	int pack_use_sparse;
     ++};
     ++
     + struct repository {
     + 	/* Environment */
     + 	/*
      @@
       	 */
       	char *submodule_prefix;
       
     -+	int settings_initialized;
      +	struct repo_settings settings;
      +
       	/* Subsystems */
       	/*
       	 * Repository's config which contains key-value pairs from the usual
     +@@
     +  */
     + void repo_update_index_if_able(struct repository *, struct lock_file *);
     + 
     ++void prepare_repo_settings(struct repository *r);
     + 
     + #endif /* REPOSITORY_H */
 2:  86380c7832 ! 2:  c0129066a0 repo-settings: add feature.manyCommits setting
     @@ -81,7 +81,7 @@
       +++ b/repo-settings.c
      @@
       #include "config.h"
     - #include "repo-settings.h"
     + #include "repository.h"
       
      +#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
      +
     @@ -89,13 +89,12 @@
       {
       	int value;
      @@
     + 
       	if (!repo_config_get_bool(r, "pack.usesparse", &value))
       		r->settings.pack_use_sparse = value;
     - 
     ++
      +	if (!repo_config_get_bool(r, "feature.manycommits", &value) && value) {
      +		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
      +		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
      +	}
     -+
     - 	r->settings_initialized = 1;
       }
 3:  49be7a7345 ! 3:  b0a2d0e188 repo-settings: parse core.untrackedCache
     @@ -25,14 +25,6 @@
       diff --git a/builtin/update-index.c b/builtin/update-index.c
       --- a/builtin/update-index.c
       +++ b/builtin/update-index.c
     -@@
     - #include "dir.h"
     - #include "split-index.h"
     - #include "fsmonitor.h"
     -+#include "repo-settings.h"
     - 
     - /*
     -  * Default to not allowing changes to the list of files. The
      @@
       	struct parse_opt_ctx_t ctx;
       	strbuf_getline_fn getline_fn;
     @@ -141,7 +133,7 @@
       	int value;
      +	char *strval;
       
     - 	if (r->settings_initialized)
     + 	if (r->settings.initialized)
       		return;
      @@
       
     @@ -168,22 +160,21 @@
       		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
       		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
       	}
     - 
     ++
      +	/* Hack for test programs like test-dump-untracked-cache */
      +	if (ignore_untracked_cache_config)
      +		r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
      +	else
      +		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
      +
     - 	r->settings_initialized = 1;
       }
      
     - diff --git a/repo-settings.h b/repo-settings.h
     - --- a/repo-settings.h
     - +++ b/repo-settings.h
     + diff --git a/repository.h b/repository.h
     + --- a/repository.h
     + +++ b/repository.h
      @@
     - #ifndef REPO_SETTINGS_H
     - #define REPO_SETTINGS_H
     + struct raw_object_store;
     + struct submodule_cache;
       
      +enum untracked_cache_setting {
      +	UNTRACKED_CACHE_UNSET = -1,
     @@ -193,7 +184,9 @@
      +};
      +
       struct repo_settings {
     - 	int core_commit_graph;
     + 	int initialized;
     + 
     +@@
       	int gc_write_commit_graph;
       
       	int index_version;
 4:  86a5a0c589 = 4:  0d4774d45f repo-settings: create feature.manyFiles setting
 5:  b8a631683a ! 5:  2e153fac22 repo-settings: create feature.experimental setting
     @@ -93,18 +93,6 @@
       pack.writeBitmaps (deprecated)::
       	This is a deprecated synonym for `repack.writeBitmaps`.
      
     - diff --git a/builtin/am.c b/builtin/am.c
     - --- a/builtin/am.c
     - +++ b/builtin/am.c
     -@@
     - #include "string-list.h"
     - #include "packfile.h"
     - #include "repository.h"
     -+#include "repo-settings.h"
     - 
     - /**
     -  * Returns the length of the first line of msg.
     -
       diff --git a/fetch-negotiator.c b/fetch-negotiator.c
       --- a/fetch-negotiator.c
       +++ b/fetch-negotiator.c
     @@ -113,7 +101,6 @@
       #include "negotiator/default.h"
       #include "negotiator/skipping.h"
      +#include "repository.h"
     -+#include "repo-settings.h"
       
      -void fetch_negotiator_init(struct fetch_negotiator *negotiator,
      -			   const char *algorithm)
     @@ -230,14 +217,6 @@
       diff --git a/merge-recursive.c b/merge-recursive.c
       --- a/merge-recursive.c
       +++ b/merge-recursive.c
     -@@
     - #include "submodule.h"
     - #include "revision.h"
     - #include "commit-reach.h"
     -+#include "repo-settings.h"
     - 
     - struct path_hashmap_entry {
     - 	struct hashmap_entry e;
      @@
       		opt->merge_detect_rename = git_config_rename("merge.renames", value);
       		free(value);
     @@ -307,13 +286,11 @@
       
      +	UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_CONFLICT);
      +	UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
     -+
     - 	r->settings_initialized = 1;
       }
      
     - diff --git a/repo-settings.h b/repo-settings.h
     - --- a/repo-settings.h
     - +++ b/repo-settings.h
     + diff --git a/repository.h b/repository.h
     + --- a/repository.h
     + +++ b/repository.h
      @@
       	UNTRACKED_CACHE_WRITE = 2
       };
     @@ -333,8 +310,8 @@
      +};
      +
       struct repo_settings {
     - 	int core_commit_graph;
     - 	int gc_write_commit_graph;
     + 	int initialized;
     + 
      @@
       	enum untracked_cache_setting core_untracked_cache;
       
     @@ -343,7 +320,7 @@
      +	enum fetch_negotiation_setting fetch_negotiation_algorithm;
       };
       
     - struct repository;
     + struct repository {
      
       diff --git a/t/t5552-skipping-fetch-negotiator.sh b/t/t5552-skipping-fetch-negotiator.sh
       --- a/t/t5552-skipping-fetch-negotiator.sh

-- 
gitgitgadget

[PATCH v3 1/5] repo-settings: consolidate some config settings

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

There are a few important config settings that are not loaded
during git_default_config. These are instead loaded on-demand.

Centralize these config options to a single scan, and store
all of the values in a repo_settings struct. The values for
each setting are initialized as negative to indicate "unset".

This centralization will be particularly important in a later
change to introduce "meta" config settings that change the
defaults for these config settings.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Makefile               |  1 +
 builtin/gc.c           | 12 +++++-------
 builtin/pack-objects.c |  8 ++++----
 commit-graph.c         |  6 +++---
 read-cache.c           | 11 ++++++-----
 repo-settings.c        | 25 +++++++++++++++++++++++++
 repository.h           | 15 +++++++++++++++
 7 files changed, 59 insertions(+), 19 deletions(-)
 create mode 100644 repo-settings.c

diff --git a/Makefile b/Makefile
index 11ccea4071..032fe9b5f9 100644
--- a/Makefile
+++ b/Makefile
@@ -964,6 +964,7 @@ LIB_OBJS += refspec.o
 LIB_OBJS += ref-filter.o
 LIB_OBJS += remote.o
 LIB_OBJS += replace-object.o
+LIB_OBJS += repo-settings.o
 LIB_OBJS += repository.o
 LIB_OBJS += rerere.o
 LIB_OBJS += resolve-undo.o
diff --git a/builtin/gc.c b/builtin/gc.c
index c18efadda5..4b8fbb9965 100644
--- a/builtin/gc.c
+++ b/builtin/gc.c
@@ -41,7 +41,6 @@ static int aggressive_depth = 50;
 static int aggressive_window = 250;
 static int gc_auto_threshold = 6700;
 static int gc_auto_pack_limit = 50;
-static int gc_write_commit_graph;
 static int detach_auto = 1;
 static timestamp_t gc_log_expire_time;
 static const char *gc_log_expire = "1.day.ago";
@@ -148,7 +147,6 @@ static void gc_config(void)
 	git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 	git_config_get_int("gc.auto", &gc_auto_threshold);
 	git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
-	git_config_get_bool("gc.writecommitgraph", &gc_write_commit_graph);
 	git_config_get_bool("gc.autodetach", &detach_auto);
 	git_config_get_expiry("gc.pruneexpire", &prune_expire);
 	git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
@@ -685,11 +683,11 @@ int cmd_gc(int argc, const char **argv, const char *prefix)
 		clean_pack_garbage();
 	}
 
-	if (gc_write_commit_graph &&
-	    write_commit_graph_reachable(get_object_directory(),
-					 !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
-					 NULL))
-		return 1;
+	prepare_repo_settings(the_repository);
+	if (the_repository->settings.gc_write_commit_graph == 1)
+		write_commit_graph_reachable(get_object_directory(),
+					     !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
+					     NULL);
 
 	if (auto_gc && too_many_loose_objects())
 		warning(_("There are too many unreachable loose objects; "
diff --git a/builtin/pack-objects.c b/builtin/pack-objects.c
index 267c562b1f..f9f7181f3f 100644
--- a/builtin/pack-objects.c
+++ b/builtin/pack-objects.c
@@ -2709,10 +2709,6 @@ static int git_pack_config(const char *k, const char *v, void *cb)
 		use_bitmap_index_default = git_config_bool(k, v);
 		return 0;
 	}
-	if (!strcmp(k, "pack.usesparse")) {
-		sparse = git_config_bool(k, v);
-		return 0;
-	}
 	if (!strcmp(k, "pack.threads")) {
 		delta_search_threads = git_config_int(k, v);
 		if (delta_search_threads < 0)
@@ -3332,6 +3328,10 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 	read_replace_refs = 0;
 
 	sparse = git_env_bool("GIT_TEST_PACK_SPARSE", 0);
+	prepare_repo_settings(the_repository);
+	if (!sparse && the_repository->settings.pack_use_sparse != -1)
+		sparse = the_repository->settings.pack_use_sparse;
+
 	reset_pack_idx_option(&pack_idx_opts);
 	git_config(git_pack_config, NULL);
 
diff --git a/commit-graph.c b/commit-graph.c
index b3c4de79b6..7854e491b2 100644
--- a/commit-graph.c
+++ b/commit-graph.c
@@ -466,7 +466,6 @@ static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 static int prepare_commit_graph(struct repository *r)
 {
 	struct object_directory *odb;
-	int config_value;
 
 	if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
 		die("dying as requested by the '%s' variable on commit-graph load!",
@@ -476,9 +475,10 @@ static int prepare_commit_graph(struct repository *r)
 		return !!r->objects->commit_graph;
 	r->objects->commit_graph_attempted = 1;
 
+	prepare_repo_settings(r);
+
 	if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
-	    (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
-	    !config_value))
+	    r->settings.core_commit_graph != 1)
 		/*
 		 * This repository is not configured to use commit graphs, so
 		 * do not load one. (But report commit_graph_attempted anyway
diff --git a/read-cache.c b/read-cache.c
index c701f7f8b8..59dbebc15d 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -1599,16 +1599,17 @@ struct cache_entry *refresh_cache_entry(struct index_state *istate,
 
 #define INDEX_FORMAT_DEFAULT 3
 
-static unsigned int get_index_format_default(void)
+static unsigned int get_index_format_default(struct repository *r)
 {
 	char *envversion = getenv("GIT_INDEX_VERSION");
 	char *endp;
-	int value;
 	unsigned int version = INDEX_FORMAT_DEFAULT;
 
 	if (!envversion) {
-		if (!git_config_get_int("index.version", &value))
-			version = value;
+		prepare_repo_settings(r);
+
+		if (r->settings.index_version >= 0)
+			version = r->settings.index_version;
 		if (version < INDEX_FORMAT_LB || INDEX_FORMAT_UB < version) {
 			warning(_("index.version set, but the value is invalid.\n"
 				  "Using version %i"), INDEX_FORMAT_DEFAULT);
@@ -2765,7 +2766,7 @@ static int do_write_index(struct index_state *istate, struct tempfile *tempfile,
 	}
 
 	if (!istate->version) {
-		istate->version = get_index_format_default();
+		istate->version = get_index_format_default(the_repository);
 		if (git_env_bool("GIT_TEST_SPLIT_INDEX", 0))
 			init_split_index(istate);
 	}
diff --git a/repo-settings.c b/repo-settings.c
new file mode 100644
index 0000000000..309577f6bc
--- /dev/null
+++ b/repo-settings.c
@@ -0,0 +1,25 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+
+void prepare_repo_settings(struct repository *r)
+{
+	int value;
+
+	if (r->settings.initialized)
+		return;
+
+	/* Defaults */
+	memset(&r->settings, -1, sizeof(r->settings));
+
+	if (!repo_config_get_bool(r, "core.commitgraph", &value))
+		r->settings.core_commit_graph = value;
+	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
+		r->settings.gc_write_commit_graph = value;
+
+	if (!repo_config_get_bool(r, "index.version", &value))
+		r->settings.index_version = value;
+
+	if (!repo_config_get_bool(r, "pack.usesparse", &value))
+		r->settings.pack_use_sparse = value;
+}
diff --git a/repository.h b/repository.h
index 4fb6a5885f..2bb2bc3eea 100644
--- a/repository.h
+++ b/repository.h
@@ -4,6 +4,7 @@
 #include "path.h"
 
 struct config_set;
+struct repo_settings;
 struct git_hash_algo;
 struct index_state;
 struct lock_file;
@@ -11,6 +12,17 @@ struct pathspec;
 struct raw_object_store;
 struct submodule_cache;
 
+struct repo_settings {
+	int initialized;
+
+	int core_commit_graph;
+	int gc_write_commit_graph;
+
+	int index_version;
+
+	int pack_use_sparse;
+};
+
 struct repository {
 	/* Environment */
 	/*
@@ -72,6 +84,8 @@ struct repository {
 	 */
 	char *submodule_prefix;
 
+	struct repo_settings settings;
+
 	/* Subsystems */
 	/*
 	 * Repository's config which contains key-value pairs from the usual
@@ -157,5 +171,6 @@ int repo_read_index_unmerged(struct repository *);
  */
 void repo_update_index_if_able(struct repository *, struct lock_file *);
 
+void prepare_repo_settings(struct repository *r);
 
 #endif /* REPOSITORY_H */
-- 
gitgitgadget

[PATCH v3 2/5] repo-settings: add feature.manyCommits setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

When a repo has many commits, it is helpful to write and read the
commit-graph file. Future changes to Git may include new config
settings that are beneficial in this scenario.

Create the 'feature.manyCommits' config setting that changes the
default values of 'core.commitGraph' and 'gc.writeCommitGraph' to
true.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config.txt         |  2 ++
 Documentation/config/core.txt    |  3 ++-
 Documentation/config/feature.txt | 15 +++++++++++++++
 Documentation/config/gc.txt      |  4 ++--
 repo-settings.c                  |  7 +++++++
 5 files changed, 28 insertions(+), 3 deletions(-)
 create mode 100644 Documentation/config/feature.txt

diff --git a/Documentation/config.txt b/Documentation/config.txt
index e3f5bc3396..77f3b1486b 100644
--- a/Documentation/config.txt
+++ b/Documentation/config.txt
@@ -345,6 +345,8 @@ include::config/difftool.txt[]
 
 include::config/fastimport.txt[]
 
+include::config/feature.txt[]
+
 include::config/fetch.txt[]
 
 include::config/format.txt[]
diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index 75538d27e7..d80162681a 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -577,7 +577,8 @@ the `GIT_NOTES_REF` environment variable.  See linkgit:git-notes[1].
 
 core.commitGraph::
 	If true, then git will read the commit-graph file (if it exists)
-	to parse the graph structure of commits. Defaults to false. See
+	to parse the graph structure of commits. Defaults to false, unless
+	`feature.manyCommits` is enabled. See
 	linkgit:git-commit-graph[1] for more information.
 
 core.useReplaceRefs::
diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
new file mode 100644
index 0000000000..f74314ae90
--- /dev/null
+++ b/Documentation/config/feature.txt
@@ -0,0 +1,15 @@
+feature.*::
+	The config settings that start with `feature.` modify the defaults of
+	a group of other config settings. These groups are created by the Git
+	developer community as recommended defaults and are subject to change.
+	In particular, new config options may be added with different defaults.
+
+feature.manyCommits::
+	Enable config options that optimize for repos with many commits. This
+	setting is recommended for repos with at least 100,000 commits. The
+	new default values are:
++
+* `core.commitGraph=true` enables reading the commit-graph file.
++
+* `gc.writeCommitGraph=true` enables writing the commit-graph file during
+garbage collection.
\ No newline at end of file
diff --git a/Documentation/config/gc.txt b/Documentation/config/gc.txt
index 02b92b18b5..31a5fc4f75 100644
--- a/Documentation/config/gc.txt
+++ b/Documentation/config/gc.txt
@@ -63,8 +63,8 @@ gc.writeCommitGraph::
 	If true, then gc will rewrite the commit-graph file when
 	linkgit:git-gc[1] is run. When using `git gc --auto`
 	the commit-graph will be updated if housekeeping is
-	required. Default is false. See linkgit:git-commit-graph[1]
-	for details.
+	required. Default is false, unless `feature.manyCommits`
+	is enabled. See linkgit:git-commit-graph[1] for details.
 
 gc.logExpiry::
 	If the file gc.log exists, then `git gc --auto` will print
diff --git a/repo-settings.c b/repo-settings.c
index 309577f6bc..fc05f4fbc4 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -2,6 +2,8 @@
 #include "config.h"
 #include "repository.h"
 
+#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
+
 void prepare_repo_settings(struct repository *r)
 {
 	int value;
@@ -22,4 +24,9 @@ void prepare_repo_settings(struct repository *r)
 
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
+
+	if (!repo_config_get_bool(r, "feature.manycommits", &value) && value) {
+		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
+		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
+	}
 }
-- 
gitgitgadget

[PATCH v3 4/5] repo-settings: create feature.manyFiles setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The feature.manyFiles setting is suitable for repos with many
files in the working directory. By setting index.version=4 and
core.untrackedCache=true, commands such as 'git status' should
improve.

While adding this setting, modify the index version precedence
tests to check how this setting overrides the default for
index.version is unset.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config/core.txt    |  4 +++-
 Documentation/config/feature.txt | 12 +++++++++++-
 Documentation/config/index.txt   |  1 +
 repo-settings.c                  |  4 ++++
 t/t1600-index.sh                 | 31 ++++++++++++++++++++++++++-----
 5 files changed, 45 insertions(+), 7 deletions(-)

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index d80162681a..7a2a33bc8c 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -86,7 +86,9 @@ core.untrackedCache::
 	it will automatically be removed, if set to `false`. Before
 	setting it to `true`, you should check that mtime is working
 	properly on your system.
-	See linkgit:git-update-index[1]. `keep` by default.
+	See linkgit:git-update-index[1]. `keep` by default, unless
+	`feature.manyFiles` is enabled which sets this setting to
+	`true` by default.
 
 core.checkStat::
 	When missing or is set to `default`, many fields in the stat
diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
index f74314ae90..c2d9ef7473 100644
--- a/Documentation/config/feature.txt
+++ b/Documentation/config/feature.txt
@@ -12,4 +12,14 @@ feature.manyCommits::
 * `core.commitGraph=true` enables reading the commit-graph file.
 +
 * `gc.writeCommitGraph=true` enables writing the commit-graph file during
-garbage collection.
\ No newline at end of file
+garbage collection.
+
+feature.manyFiles::
+	Enable config options that optimize for repos with many files in the
+	working directory. With many files, commands such as `git status` and
+	`git checkout` may be slow and these new defaults improve performance:
++
+* `index.version=4` enables path-prefix compression in the index.
++
+* `core.untrackedCache=true` enables the untracked cache. This setting assumes
+that mtime is working on your machine.
\ No newline at end of file
diff --git a/Documentation/config/index.txt b/Documentation/config/index.txt
index f181503041..7cb50b37e9 100644
--- a/Documentation/config/index.txt
+++ b/Documentation/config/index.txt
@@ -24,3 +24,4 @@ index.threads::
 index.version::
 	Specify the version with which new index files should be
 	initialized.  This does not affect existing repositories.
+	If `feature.manyFiles` is enabled, then the default is 4.
diff --git a/repo-settings.c b/repo-settings.c
index 9ed83e6e54..af93696343 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -42,6 +42,10 @@ void prepare_repo_settings(struct repository *r)
 		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
 		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
 	}
+	if (!repo_config_get_bool(r, "feature.manyfiles", &value) && value) {
+		UPDATE_DEFAULT(r->settings.index_version, 4);
+		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
+	}
 
 	/* Hack for test programs like test-dump-untracked-cache */
 	if (ignore_untracked_cache_config)
diff --git a/t/t1600-index.sh b/t/t1600-index.sh
index 42962ed7d4..c77721b580 100755
--- a/t/t1600-index.sh
+++ b/t/t1600-index.sh
@@ -59,17 +59,38 @@ test_expect_success 'out of bounds index.version issues warning' '
 	)
 '
 
-test_expect_success 'GIT_INDEX_VERSION takes precedence over config' '
+test_index_version () {
+	INDEX_VERSION_CONFIG=$1 &&
+	FEATURE_MANY_FILES=$2 &&
+	ENV_VAR_VERSION=$3
+	EXPECTED_OUTPUT_VERSION=$4 &&
 	(
 		rm -f .git/index &&
-		GIT_INDEX_VERSION=4 &&
-		export GIT_INDEX_VERSION &&
-		git config --add index.version 2 &&
+		rm -f .git/config &&
+		if test "$INDEX_VERSION_CONFIG" -ne 0
+		then
+			git config --add index.version $INDEX_VERSION_CONFIG
+		fi &&
+		git config --add feature.manyFiles $FEATURE_MANY_FILES
+		if test "$ENV_VAR_VERSION" -ne 0
+		then
+			GIT_INDEX_VERSION=$ENV_VAR_VERSION &&
+			export GIT_INDEX_VERSION
+		else
+			unset GIT_INDEX_VERSION
+		fi &&
 		git add a 2>&1 &&
-		echo 4 >expect &&
+		echo $EXPECTED_OUTPUT_VERSION >expect &&
 		test-tool index-version <.git/index >actual &&
 		test_cmp expect actual
 	)
+}
+
+test_expect_success 'index version config precedence' '
+	test_index_version 2 false 4 4 &&
+	test_index_version 2 true 0 2 &&
+	test_index_version 0 true 0 4 &&
+	test_index_version 0 true 2 2
 '
 
 test_done
-- 
gitgitgadget

[PATCH v3 3/5] repo-settings: parse core.untrackedCache

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The core.untrackedCache config setting is slightly complicated,
so clarify its use and centralize its parsing into the repo
settings.

The default value is "keep" (returned as -1), which persists the
untracked cache if it exists.

If the value is set as "false" (returned as 0), then remove the
untracked cache if it exists.

If the value is set as "true" (returned as 1), then write the
untracked cache and persist it.

Instead of relying on magic values of -1, 0, and 1, split these
options into an enum. This allows the use of "-1" as a
default value. After parsing the config options, if the value is
unset we can initialize it to UNTRACKED_CACHE_KEEP.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 builtin/update-index.c |  6 ++++--
 config.c               | 24 ------------------------
 read-cache.c           | 19 +++++++++----------
 repo-settings.c        | 22 +++++++++++++++++++++-
 repository.h           |  8 ++++++++
 5 files changed, 42 insertions(+), 37 deletions(-)

diff --git a/builtin/update-index.c b/builtin/update-index.c
index dff2f4b837..49302d98c5 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -966,6 +966,7 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	struct parse_opt_ctx_t ctx;
 	strbuf_getline_fn getline_fn;
 	int parseopt_state = PARSE_OPT_UNKNOWN;
+	struct repository *r = the_repository;
 	struct option options[] = {
 		OPT_BIT('q', NULL, &refresh_args.flags,
 			N_("continue refresh even when index needs update"),
@@ -1180,11 +1181,12 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 		remove_split_index(&the_index);
 	}
 
+	prepare_repo_settings(r);
 	switch (untracked_cache) {
 	case UC_UNSPECIFIED:
 		break;
 	case UC_DISABLE:
-		if (git_config_get_untracked_cache() == 1)
+		if (r->settings.core_untracked_cache == UNTRACKED_CACHE_WRITE)
 			warning(_("core.untrackedCache is set to true; "
 				  "remove or change it, if you really want to "
 				  "disable the untracked cache"));
@@ -1196,7 +1198,7 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 		return !test_if_untracked_cache_is_supported();
 	case UC_ENABLE:
 	case UC_FORCE:
-		if (git_config_get_untracked_cache() == 0)
+		if (r->settings.core_untracked_cache == UNTRACKED_CACHE_REMOVE)
 			warning(_("core.untrackedCache is set to false; "
 				  "remove or change it, if you really want to "
 				  "enable the untracked cache"));
diff --git a/config.c b/config.c
index faa57e436c..3241dbc54d 100644
--- a/config.c
+++ b/config.c
@@ -2277,30 +2277,6 @@ int git_config_get_expiry_in_days(const char *key, timestamp_t *expiry, timestam
 	return -1; /* thing exists but cannot be parsed */
 }
 
-int git_config_get_untracked_cache(void)
-{
-	int val = -1;
-	const char *v;
-
-	/* Hack for test programs like test-dump-untracked-cache */
-	if (ignore_untracked_cache_config)
-		return -1;
-
-	if (!git_config_get_maybe_bool("core.untrackedcache", &val))
-		return val;
-
-	if (!git_config_get_value("core.untrackedcache", &v)) {
-		if (!strcasecmp(v, "keep"))
-			return -1;
-
-		error(_("unknown core.untrackedCache value '%s'; "
-			"using 'keep' default value"), v);
-		return -1;
-	}
-
-	return -1; /* default value */
-}
-
 int git_config_get_split_index(void)
 {
 	int val;
diff --git a/read-cache.c b/read-cache.c
index 59dbebc15d..7a07286164 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -1845,18 +1845,17 @@ static void check_ce_order(struct index_state *istate)
 
 static void tweak_untracked_cache(struct index_state *istate)
 {
-	switch (git_config_get_untracked_cache()) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
+	struct repository *r = the_repository;
+
+	prepare_repo_settings(r);
+
+	if (r->settings.core_untracked_cache  == UNTRACKED_CACHE_REMOVE) {
 		remove_untracked_cache(istate);
-		break;
-	case 1: /* true */
-		add_untracked_cache(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
+		return;
 	}
+
+	if (r->settings.core_untracked_cache == UNTRACKED_CACHE_WRITE)
+		add_untracked_cache(istate);
 }
 
 static void tweak_split_index(struct index_state *istate)
diff --git a/repo-settings.c b/repo-settings.c
index fc05f4fbc4..9ed83e6e54 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -7,6 +7,7 @@
 void prepare_repo_settings(struct repository *r)
 {
 	int value;
+	char *strval;
 
 	if (r->settings.initialized)
 		return;
@@ -21,12 +22,31 @@ void prepare_repo_settings(struct repository *r)
 
 	if (!repo_config_get_bool(r, "index.version", &value))
 		r->settings.index_version = value;
+	if (!repo_config_get_maybe_bool(r, "core.untrackedcache", &value)) {
+		if (value == 0)
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_REMOVE;
+		else
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_WRITE;
+	} else if (!repo_config_get_string(r, "core.untrackedcache", &strval)) {
+		if (!strcasecmp(strval, "keep"))
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
+
+		free(strval);
+	}
+
 
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
-
+	
 	if (!repo_config_get_bool(r, "feature.manycommits", &value) && value) {
 		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
 		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
 	}
+
+	/* Hack for test programs like test-dump-untracked-cache */
+	if (ignore_untracked_cache_config)
+		r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
+	else
+		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
+
 }
diff --git a/repository.h b/repository.h
index 2bb2bc3eea..e7a72e2341 100644
--- a/repository.h
+++ b/repository.h
@@ -12,6 +12,13 @@ struct pathspec;
 struct raw_object_store;
 struct submodule_cache;
 
+enum untracked_cache_setting {
+	UNTRACKED_CACHE_UNSET = -1,
+	UNTRACKED_CACHE_REMOVE = 0,
+	UNTRACKED_CACHE_KEEP = 1,
+	UNTRACKED_CACHE_WRITE = 2
+};
+
 struct repo_settings {
 	int initialized;
 
@@ -19,6 +26,7 @@ struct repo_settings {
 	int gc_write_commit_graph;
 
 	int index_version;
+	enum untracked_cache_setting core_untracked_cache;
 
 	int pack_use_sparse;
 };
-- 
gitgitgadget

[PATCH v3 5/5] repo-settings: create feature.experimental setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The 'feature.experimental' setting includes config options that are
not committed to become defaults, but could use additional testing.

Update the following config settings to take new defaults, and to
use the repo_settings struct if not already using them:

* 'pack.useSparse=true'

* 'merge.directoryRenames=true'

* 'fetch.negotiationAlgorithm=skipping'

In the case of fetch.negotiationAlgorithm, the existing logic
would load the config option only when about to use the setting,
so had a die() statement on an unknown string value. This is
removed as now the config is parsed under prepare_repo_settings().
In general, this die() is probably misplaced and not valuable.
A test was removed that checked this die() statement executed.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config/feature.txt     | 17 +++++++++++++++++
 Documentation/config/fetch.txt       |  3 ++-
 Documentation/config/merge.txt       |  3 ++-
 Documentation/config/pack.txt        |  3 ++-
 fetch-negotiator.c                   | 25 +++++++++++++------------
 fetch-negotiator.h                   |  5 +++--
 fetch-pack.c                         | 11 +++++------
 merge-recursive.c                    | 14 +++++---------
 repo-settings.c                      | 19 +++++++++++++++++++
 repository.h                         | 16 ++++++++++++++++
 t/t5552-skipping-fetch-negotiator.sh | 23 -----------------------
 11 files changed, 84 insertions(+), 55 deletions(-)

diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
index c2d9ef7473..287621b1e8 100644
--- a/Documentation/config/feature.txt
+++ b/Documentation/config/feature.txt
@@ -4,6 +4,23 @@ feature.*::
 	developer community as recommended defaults and are subject to change.
 	In particular, new config options may be added with different defaults.
 
+feature.experimental::
+	Enable config options that are new to Git, and are being considered for
+	future defaults. Config settings included here may be added or removed
+	with each release, including minor version updates. These settings may
+	have unintended interactions since they are so new. Please enable this
+	setting if you are interested in providing feedback on experimental
+	features. The new default values are:
++
+* `pack.useSparse=true` uses a new algorithm when constructing a pack-file
+which can improve `git push` performance in repos with many files.
++
+* `merge.directoryRenames=true` uses a new algorithm for detecting renames by
+using entire directories at a time instead of single files at a time.
++
+* `fetch.negotiationAlgorithm=skipping` may improve fetch negotiation times by
+skipping more commits at a time, reducing the number of round trips.
+
 feature.manyCommits::
 	Enable config options that optimize for repos with many commits. This
 	setting is recommended for repos with at least 100,000 commits. The
diff --git a/Documentation/config/fetch.txt b/Documentation/config/fetch.txt
index ba890b5884..d402110638 100644
--- a/Documentation/config/fetch.txt
+++ b/Documentation/config/fetch.txt
@@ -59,7 +59,8 @@ fetch.negotiationAlgorithm::
 	effort to converge faster, but may result in a larger-than-necessary
 	packfile; The default is "default" which instructs Git to use the default algorithm
 	that never skips commits (unless the server has acknowledged it or one
-	of its descendants).
+	of its descendants). If `feature.experimental` is enabled, then this
+	setting defaults to "skipping".
 	Unknown values will cause 'git fetch' to error out.
 +
 See also the `--negotiation-tip` option for linkgit:git-fetch[1].
diff --git a/Documentation/config/merge.txt b/Documentation/config/merge.txt
index 6a313937f8..e8def2d63c 100644
--- a/Documentation/config/merge.txt
+++ b/Documentation/config/merge.txt
@@ -54,7 +54,8 @@ merge.directoryRenames::
 	moved into the new directory.  If set to "conflict", a conflict
 	will be reported for such paths.  If merge.renames is false,
 	merge.directoryRenames is ignored and treated as false.  Defaults
-	to "conflict".
+	to "conflict" unless `feature.experimental` is enabled and the
+	default is "true".
 
 merge.renormalize::
 	Tell Git that canonical representation of files in the
diff --git a/Documentation/config/pack.txt b/Documentation/config/pack.txt
index 9cdcfa7324..1d66f0c992 100644
--- a/Documentation/config/pack.txt
+++ b/Documentation/config/pack.txt
@@ -112,7 +112,8 @@ pack.useSparse::
 	objects. This can have significant performance benefits when
 	computing a pack to send a small change. However, it is possible
 	that extra objects are added to the pack-file if the included
-	commits contain certain types of direct renames.
+	commits contain certain types of direct renames. Default is `false`
+	unless `feature.experimental` is enabled.
 
 pack.writeBitmaps (deprecated)::
 	This is a deprecated synonym for `repack.writeBitmaps`.
diff --git a/fetch-negotiator.c b/fetch-negotiator.c
index d6d685cba0..0a1357dc9d 100644
--- a/fetch-negotiator.c
+++ b/fetch-negotiator.c
@@ -2,19 +2,20 @@
 #include "fetch-negotiator.h"
 #include "negotiator/default.h"
 #include "negotiator/skipping.h"
+#include "repository.h"
 
-void fetch_negotiator_init(struct fetch_negotiator *negotiator,
-			   const char *algorithm)
+void fetch_negotiator_init(struct repository *r,
+			   struct fetch_negotiator *negotiator)
 {
-	if (algorithm) {
-		if (!strcmp(algorithm, "skipping")) {
-			skipping_negotiator_init(negotiator);
-			return;
-		} else if (!strcmp(algorithm, "default")) {
-			/* Fall through to default initialization */
-		} else {
-			die("unknown fetch negotiation algorithm '%s'", algorithm);
-		}
+	prepare_repo_settings(r);
+	switch(r->settings.fetch_negotiation_algorithm) {
+	case FETCH_NEGOTIATION_SKIPPING:
+		skipping_negotiator_init(negotiator);
+		return;
+
+	case FETCH_NEGOTIATION_DEFAULT:
+	default:
+		default_negotiator_init(negotiator);
+		return;
 	}
-	default_negotiator_init(negotiator);
 }
diff --git a/fetch-negotiator.h b/fetch-negotiator.h
index 9e3967ce66..ea78868504 100644
--- a/fetch-negotiator.h
+++ b/fetch-negotiator.h
@@ -2,6 +2,7 @@
 #define FETCH_NEGOTIATOR_H
 
 struct commit;
+struct repository;
 
 /*
  * An object that supplies the information needed to negotiate the contents of
@@ -52,7 +53,7 @@ struct fetch_negotiator {
 	void *data;
 };
 
-void fetch_negotiator_init(struct fetch_negotiator *negotiator,
-			   const char *algorithm);
+void fetch_negotiator_init(struct repository *r,
+			   struct fetch_negotiator *negotiator);
 
 #endif
diff --git a/fetch-pack.c b/fetch-pack.c
index 65be043f2a..d81f47c07b 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -36,7 +36,6 @@ static int agent_supported;
 static int server_supports_filtering;
 static struct lock_file shallow_lock;
 static const char *alternate_shallow_file;
-static char *negotiation_algorithm;
 static struct strbuf fsck_msg_types = STRBUF_INIT;
 
 /* Remember to update object flag allocation in object.h */
@@ -892,12 +891,13 @@ static struct ref *do_fetch_pack(struct fetch_pack_args *args,
 				 struct shallow_info *si,
 				 char **pack_lockfile)
 {
+	struct repository *r = the_repository;
 	struct ref *ref = copy_ref_list(orig_ref);
 	struct object_id oid;
 	const char *agent_feature;
 	int agent_len;
 	struct fetch_negotiator negotiator;
-	fetch_negotiator_init(&negotiator, negotiation_algorithm);
+	fetch_negotiator_init(r, &negotiator);
 
 	sort_ref_list(&ref, ref_compare_name);
 	QSORT(sought, nr_sought, cmp_ref_by_name);
@@ -911,7 +911,7 @@ static struct ref *do_fetch_pack(struct fetch_pack_args *args,
 
 	if (server_supports("shallow"))
 		print_verbose(args, _("Server supports %s"), "shallow");
-	else if (args->depth > 0 || is_repository_shallow(the_repository))
+	else if (args->depth > 0 || is_repository_shallow(r))
 		die(_("Server does not support shallow clients"));
 	if (args->depth > 0 || args->deepen_since || args->deepen_not)
 		args->deepen = 1;
@@ -1379,6 +1379,7 @@ static struct ref *do_fetch_pack_v2(struct fetch_pack_args *args,
 				    struct shallow_info *si,
 				    char **pack_lockfile)
 {
+	struct repository *r = the_repository;
 	struct ref *ref = copy_ref_list(orig_ref);
 	enum fetch_state state = FETCH_CHECK_LOCAL;
 	struct oidset common = OIDSET_INIT;
@@ -1386,7 +1387,7 @@ static struct ref *do_fetch_pack_v2(struct fetch_pack_args *args,
 	int in_vain = 0;
 	int haves_to_send = INITIAL_FLUSH;
 	struct fetch_negotiator negotiator;
-	fetch_negotiator_init(&negotiator, negotiation_algorithm);
+	fetch_negotiator_init(r, &negotiator);
 	packet_reader_init(&reader, fd[0], NULL, 0,
 			   PACKET_READ_CHOMP_NEWLINE |
 			   PACKET_READ_DIE_ON_ERR_PACKET);
@@ -1505,8 +1506,6 @@ static void fetch_pack_config(void)
 	git_config_get_bool("repack.usedeltabaseoffset", &prefer_ofs_delta);
 	git_config_get_bool("fetch.fsckobjects", &fetch_fsck_objects);
 	git_config_get_bool("transfer.fsckobjects", &transfer_fsck_objects);
-	git_config_get_string("fetch.negotiationalgorithm",
-			      &negotiation_algorithm);
 
 	git_config(fetch_pack_config_cb, NULL);
 }
diff --git a/merge-recursive.c b/merge-recursive.c
index 12300131fc..4f77b4b3bb 100644
--- a/merge-recursive.c
+++ b/merge-recursive.c
@@ -3662,15 +3662,6 @@ static void merge_recursive_config(struct merge_options *opt)
 		opt->merge_detect_rename = git_config_rename("merge.renames", value);
 		free(value);
 	}
-	if (!git_config_get_string("merge.directoryrenames", &value)) {
-		int boolval = git_parse_maybe_bool(value);
-		if (0 <= boolval) {
-			opt->detect_directory_renames = boolval ? 2 : 0;
-		} else if (!strcasecmp(value, "conflict")) {
-			opt->detect_directory_renames = 1;
-		} /* avoid erroring on values from future versions of git */
-		free(value);
-	}
 	git_config(git_xmerge_config, NULL);
 }
 
@@ -3688,6 +3679,11 @@ void init_merge_options(struct merge_options *opt,
 	opt->diff_detect_rename = -1;
 	opt->merge_detect_rename = -1;
 	opt->detect_directory_renames = 1;
+
+	prepare_repo_settings(repo);
+	if (repo->settings.merge_directory_renames >= 0)
+		opt->detect_directory_renames = repo->settings.merge_directory_renames;
+
 	merge_recursive_config(opt);
 	merge_verbosity = getenv("GIT_MERGE_VERBOSITY");
 	if (merge_verbosity)
diff --git a/repo-settings.c b/repo-settings.c
index af93696343..e0673938c0 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -34,6 +34,18 @@ void prepare_repo_settings(struct repository *r)
 		free(strval);
 	}
 
+	if (!repo_config_get_maybe_bool(r, "merge.directoryrenames", &value))
+		r->settings.merge_directory_renames = value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
+	else if (!repo_config_get_string(r, "merge.directoryrenames", &strval)) {
+		if (!strcasecmp(strval, "conflict"))
+			r->settings.merge_directory_renames = MERGE_DIRECTORY_RENAMES_CONFLICT;
+	}
+	if (!repo_config_get_string(r, "fetch.negotiationalgorithm", &strval)) {
+		if (!strcasecmp(strval, "skipping"))
+			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_SKIPPING;
+		else
+			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_DEFAULT;
+	}
 
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
@@ -46,6 +58,11 @@ void prepare_repo_settings(struct repository *r)
 		UPDATE_DEFAULT(r->settings.index_version, 4);
 		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
 	}
+	if (!repo_config_get_bool(r, "feature.experimental", &value) && value) {
+		UPDATE_DEFAULT(r->settings.pack_use_sparse, 1);
+		UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_TRUE);
+		UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
+	}
 
 	/* Hack for test programs like test-dump-untracked-cache */
 	if (ignore_untracked_cache_config)
@@ -53,4 +70,6 @@ void prepare_repo_settings(struct repository *r)
 	else
 		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
 
+	UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_CONFLICT);
+	UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
 }
diff --git a/repository.h b/repository.h
index e7a72e2341..b8e52dd48f 100644
--- a/repository.h
+++ b/repository.h
@@ -19,6 +19,20 @@ enum untracked_cache_setting {
 	UNTRACKED_CACHE_WRITE = 2
 };
 
+enum merge_directory_renames_setting {
+	MERGE_DIRECTORY_RENAMES_UNSET = -1,
+	MERGE_DIRECTORY_RENAMES_NONE = 0,
+	MERGE_DIRECTORY_RENAMES_CONFLICT = 1,
+	MERGE_DIRECTORY_RENAMES_TRUE = 2,
+};
+
+enum fetch_negotiation_setting {
+	FETCH_NEGOTIATION_UNSET = -1,
+	FETCH_NEGOTIATION_NONE = 0,
+	FETCH_NEGOTIATION_DEFAULT = 1,
+	FETCH_NEGOTIATION_SKIPPING = 2,
+};
+
 struct repo_settings {
 	int initialized;
 
@@ -29,6 +43,8 @@ struct repo_settings {
 	enum untracked_cache_setting core_untracked_cache;
 
 	int pack_use_sparse;
+	enum merge_directory_renames_setting merge_directory_renames;
+	enum fetch_negotiation_setting fetch_negotiation_algorithm;
 };
 
 struct repository {
diff --git a/t/t5552-skipping-fetch-negotiator.sh b/t/t5552-skipping-fetch-negotiator.sh
index 8a14be51a1..f70cbcc9ca 100755
--- a/t/t5552-skipping-fetch-negotiator.sh
+++ b/t/t5552-skipping-fetch-negotiator.sh
@@ -60,29 +60,6 @@ test_expect_success 'commits with no parents are sent regardless of skip distanc
 	have_not_sent c6 c4 c3
 '
 
-test_expect_success 'unknown fetch.negotiationAlgorithm values error out' '
-	rm -rf server client trace &&
-	git init server &&
-	test_commit -C server to_fetch &&
-
-	git init client &&
-	test_commit -C client on_client &&
-	git -C client checkout on_client &&
-
-	test_config -C client fetch.negotiationAlgorithm invalid &&
-	test_must_fail git -C client fetch "$(pwd)/server" 2>err &&
-	test_i18ngrep "unknown fetch negotiation algorithm" err &&
-
-	# Explicit "default" value
-	test_config -C client fetch.negotiationAlgorithm default &&
-	git -C client -c fetch.negotiationAlgorithm=default fetch "$(pwd)/server" &&
-
-	# Implementation detail: If there is nothing to fetch, we will not error out
-	test_config -C client fetch.negotiationAlgorithm invalid &&
-	git -C client fetch "$(pwd)/server" 2>err &&
-	test_i18ngrep ! "unknown fetch negotiation algorithm" err
-'
-
 test_expect_success 'when two skips collide, favor the larger one' '
 	rm -rf server client trace &&
 	git init server &&
-- 
gitgitgadget

Re: [PATCH v3 1/5] repo-settings: consolidate some config settings

[Junio C Hamano]

"Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:

> diff --git a/repository.h b/repository.h
> index 4fb6a5885f..2bb2bc3eea 100644
> --- a/repository.h
> +++ b/repository.h
> @@ -4,6 +4,7 @@
>  #include "path.h"
>  
>  struct config_set;
> +struct repo_settings;

Given that the next hunk you introduce the real thing, and nobody
refers to it until then, I do not see why we want to have a forward
declaration.

>  struct git_hash_algo;
>  struct index_state;
>  struct lock_file;
> @@ -11,6 +12,17 @@ struct pathspec;
>  struct raw_object_store;
>  struct submodule_cache;
>  
> +struct repo_settings {
> +	int initialized;
> +
> +	int core_commit_graph;
> +	int gc_write_commit_graph;
> +
> +	int index_version;
> +
> +	int pack_use_sparse;
> +};
> +
>  struct repository {
>  	/* Environment */
>  	/*
> @@ -72,6 +84,8 @@ struct repository {
>  	 */
>  	char *submodule_prefix;
>  
> +	struct repo_settings settings;
> +
>  	/* Subsystems */
>  	/*
>  	 * Repository's config which contains key-value pairs from the usual
> @@ -157,5 +171,6 @@ int repo_read_index_unmerged(struct repository *);
>   */
>  void repo_update_index_if_able(struct repository *, struct lock_file *);
>  
> +void prepare_repo_settings(struct repository *r);
>  
>  #endif /* REPOSITORY_H */

Re: [PATCH v3 2/5] repo-settings: add feature.manyCommits setting

[Junio C Hamano]

"Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:

> diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
> new file mode 100644
> index 0000000000..f74314ae90
> --- /dev/null
> +++ b/Documentation/config/feature.txt
> @@ -0,0 +1,15 @@
> +feature.*::
> +...
> +* `gc.writeCommitGraph=true` enables writing the commit-graph file during
> +garbage collection.
> \ No newline at end of file

No newline at end of file

> diff --git a/Documentation/config/gc.txt b/Documentation/config/gc.txt
> index 02b92b18b5..31a5fc4f75 100644
> --- a/Documentation/config/gc.txt
> +++ b/Documentation/config/gc.txt
> @@ -63,8 +63,8 @@ gc.writeCommitGraph::
>  	If true, then gc will rewrite the commit-graph file when
>  	linkgit:git-gc[1] is run. When using `git gc --auto`
>  	the commit-graph will be updated if housekeeping is
> -	required. Default is false. See linkgit:git-commit-graph[1]
> -	for details.
> +	required. Default is false, unless `feature.manyCommits`
> +	is enabled. See linkgit:git-commit-graph[1] for details.
>  
>  gc.logExpiry::
>  	If the file gc.log exists, then `git gc --auto` will print
> diff --git a/repo-settings.c b/repo-settings.c
> index 309577f6bc..fc05f4fbc4 100644
> --- a/repo-settings.c
> +++ b/repo-settings.c
> @@ -2,6 +2,8 @@
>  #include "config.h"
>  #include "repository.h"
>  
> +#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)

A few points:

 1. give 's' and 'v' a bit better name, perhaps 'slot' and 'value'?

 2. "do { if ((s) == -1) { (s) = (v); } } while(0)"

 3. When we learn to set default values for variables that are not
    boolean in the future, we will regret that we did not name it
    UPDATE_DEFAULT_BOOL(slot, value).

Re: [PATCH v3 2/5] repo-settings: add feature.manyCommits setting

[Johannes Schindelin]

Hi Junio,

On Tue, 30 Jul 2019, Junio C Hamano wrote:

> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>
> > +#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
>
> [...]
>  3. When we learn to set default values for variables that are not
>     boolean in the future, we will regret that we did not name it
>     UPDATE_DEFAULT_BOOL(slot, value).

On the other hand, as we never promised any kind of API (and this is not
even an internal API to begin with), it will be _easy_ to rename it in
the unlikely event that we would ever introduce non-boolean defaults to
override, wouldn't you agree?

We have plenty of precedent where patch series start by refactoring,
whether it is to rename functions or variables or files or extracting
functions. Preparing for a future that might never come strikes me as
premature optimization.

Ciao,
Dscho

Re: [PATCH v3 2/5] repo-settings: add feature.manyCommits setting

[Ævar Arnfjörð Bjarmason]

On Tue, Jul 30 2019, Derrick Stolee via GitGitGadget wrote:

> +feature.*::
> +	The config settings that start with `feature.` modify the defaults of
> +	a group of other config settings. These groups are created by the Git
> +	developer community as recommended defaults and are subject to change.
> +	In particular, new config options may be added with different defaults.
> +
> +feature.manyCommits::
> +	Enable config options that optimize for repos with many commits. This
> +	setting is recommended for repos with at least 100,000 commits. The
> +	new default values are:
> ++
> +* `core.commitGraph=true` enables reading the commit-graph file.
> ++
> +* `gc.writeCommitGraph=true` enables writing the commit-graph file during
> +garbage collection.

During the whole new commit graph format discussion (which has now
landed) we discussed just auto toggling this:
https://public-inbox.org/git/87zhobr4fl.fsf@evledraar.gmail.com/

This looks fine, but have we backed out of simply enabling this at this
point? I don't see why not, regardless of commit count...

Re: [PATCH v3 2/5] repo-settings: add feature.manyCommits setting

[Junio C Hamano]

Johannes Schindelin <Johannes.Schindelin@gmx.de> writes:

> On Tue, 30 Jul 2019, Junio C Hamano wrote:
>
>> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>
>> > +#define UPDATE_DEFAULT(s,v) do { if (s == -1) { s = v; } } while(0)
>>
>> [...]
>>  3. When we learn to set default values for variables that are not
>>     boolean in the future, we will regret that we did not name it
>>     UPDATE_DEFAULT_BOOL(slot, value).
>
> On the other hand, as we never promised any kind of API (and this is not
> even an internal API to begin with), it will be _easy_ to rename it in
> the unlikely event that we would ever introduce non-boolean defaults to
> override, wouldn't you agree?

I agree that it is easy to say that it is easy to rename it later
and burden somebody else with the task.

I know that the renaming itself is easy, when you limit yourself
within the scope of a single topic, whether done now or later.  I
also know that having to worry about other topics in flight has
non-zero cost.  I also know that you are not the one who will bear
it---I will be.

So from my point of view, if we can make a prediction, even with
limited knowledge that a name may need to be renamed in the future,
it is better not pick such a name and instead use one that we think
it has a better chance of surviving without needing a rename.

Re: [PATCH v3 2/5] repo-settings: add feature.manyCommits setting

[Derrick Stolee]

On 7/31/2019 11:01 AM, Ævar Arnfjörð Bjarmason wrote:
> 
> On Tue, Jul 30 2019, Derrick Stolee via GitGitGadget wrote:
> 
>> +feature.*::
>> +	The config settings that start with `feature.` modify the defaults of
>> +	a group of other config settings. These groups are created by the Git
>> +	developer community as recommended defaults and are subject to change.
>> +	In particular, new config options may be added with different defaults.
>> +
>> +feature.manyCommits::
>> +	Enable config options that optimize for repos with many commits. This
>> +	setting is recommended for repos with at least 100,000 commits. The
>> +	new default values are:
>> ++
>> +* `core.commitGraph=true` enables reading the commit-graph file.
>> ++
>> +* `gc.writeCommitGraph=true` enables writing the commit-graph file during
>> +garbage collection.
> 
> During the whole new commit graph format discussion (which has now
> landed) we discussed just auto toggling this:
> https://public-inbox.org/git/87zhobr4fl.fsf@evledraar.gmail.com/
> 
> This looks fine, but have we backed out of simply enabling this at this
> point? I don't see why not, regardless of commit count...

I would be happy to drop feature.manyCommits and instead swap the
defaults of core.commitGraph and gc.writeCommitGraph to true if we think
that is what we want to do for 2.24.0. We can use the repo settings and
UPDATE_DEFAULT[_BOOL] for this purpose.

Thanks,
-Stolee

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Elijah Newren]

Sorry for the late reply...

On Tue, Jul 30, 2019 at 3:49 PM Derrick Stolee via GitGitGadget
<gitgitgadget@gmail.com> wrote:
>
> --- a/Documentation/config/merge.txt
> +++ b/Documentation/config/merge.txt
> @@ -54,7 +54,8 @@ merge.directoryRenames::
>         moved into the new directory.  If set to "conflict", a conflict
>         will be reported for such paths.  If merge.renames is false,
>         merge.directoryRenames is ignored and treated as false.  Defaults
> -       to "conflict".
> +       to "conflict" unless `feature.experimental` is enabled and the
> +       default is "true".

I have a hard time parsing that changed sentence.  Perhaps something like:
   ...unless `feature.experimental` is enabled in which case the
default is "true".
?

...
> diff --git a/repo-settings.c b/repo-settings.c
> index af93696343..e0673938c0 100644
> --- a/repo-settings.c
> +++ b/repo-settings.c
> @@ -34,6 +34,18 @@ void prepare_repo_settings(struct repository *r)
>                 free(strval);
>         }
>
> +       if (!repo_config_get_maybe_bool(r, "merge.directoryrenames", &value))
> +               r->settings.merge_directory_renames = value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;

Shouldn't that be "MERGE_DIRECTORY_RENAMES_NONE" instead of "0"?

> diff --git a/repository.h b/repository.h
> index e7a72e2341..b8e52dd48f 100644
> --- a/repository.h
> +++ b/repository.h
> @@ -19,6 +19,20 @@ enum untracked_cache_setting {
>         UNTRACKED_CACHE_WRITE = 2
>  };
>
> +enum merge_directory_renames_setting {
> +       MERGE_DIRECTORY_RENAMES_UNSET = -1,
> +       MERGE_DIRECTORY_RENAMES_NONE = 0,
> +       MERGE_DIRECTORY_RENAMES_CONFLICT = 1,
> +       MERGE_DIRECTORY_RENAMES_TRUE = 2,
> +};

Thanks for adding these; makes things much nicer.  :-)


Cheers,
Elijah

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Derrick Stolee]

On 8/8/2019 2:34 PM, Elijah Newren wrote:
> Sorry for the late reply...
> 
> On Tue, Jul 30, 2019 at 3:49 PM Derrick Stolee via GitGitGadget
> <gitgitgadget@gmail.com> wrote:
>>
>> --- a/Documentation/config/merge.txt
>> +++ b/Documentation/config/merge.txt
>> @@ -54,7 +54,8 @@ merge.directoryRenames::
>>         moved into the new directory.  If set to "conflict", a conflict
>>         will be reported for such paths.  If merge.renames is false,
>>         merge.directoryRenames is ignored and treated as false.  Defaults
>> -       to "conflict".
>> +       to "conflict" unless `feature.experimental` is enabled and the
>> +       default is "true".
> 
> I have a hard time parsing that changed sentence.  Perhaps something like:
>    ...unless `feature.experimental` is enabled in which case the
> default is "true".
> ?
> 
> ...
>> diff --git a/repo-settings.c b/repo-settings.c
>> index af93696343..e0673938c0 100644
>> --- a/repo-settings.c
>> +++ b/repo-settings.c
>> @@ -34,6 +34,18 @@ void prepare_repo_settings(struct repository *r)
>>                 free(strval);
>>         }
>>
>> +       if (!repo_config_get_maybe_bool(r, "merge.directoryrenames", &value))
>> +               r->settings.merge_directory_renames = value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
> 
> Shouldn't that be "MERGE_DIRECTORY_RENAMES_NONE" instead of "0"?
> 
>> diff --git a/repository.h b/repository.h
>> index e7a72e2341..b8e52dd48f 100644
>> --- a/repository.h
>> +++ b/repository.h
>> @@ -19,6 +19,20 @@ enum untracked_cache_setting {
>>         UNTRACKED_CACHE_WRITE = 2
>>  };
>>
>> +enum merge_directory_renames_setting {
>> +       MERGE_DIRECTORY_RENAMES_UNSET = -1,
>> +       MERGE_DIRECTORY_RENAMES_NONE = 0,
>> +       MERGE_DIRECTORY_RENAMES_CONFLICT = 1,
>> +       MERGE_DIRECTORY_RENAMES_TRUE = 2,
>> +};
> 
> Thanks for adding these; makes things much nicer.  :-)

Thanks for the feedback! It's just in time for me to send a v4.

-Stolee

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Junio C Hamano]

Elijah Newren <newren@gmail.com> writes:

>> --- a/Documentation/config/merge.txt
>> +++ b/Documentation/config/merge.txt
>> @@ -54,7 +54,8 @@ merge.directoryRenames::
>>         moved into the new directory.  If set to "conflict", a conflict
>>         will be reported for such paths.  If merge.renames is false,
>>         merge.directoryRenames is ignored and treated as false.  Defaults
>> -       to "conflict".
>> +       to "conflict" unless `feature.experimental` is enabled and the
>> +       default is "true".
>
> I have a hard time parsing that changed sentence.  Perhaps something like:
>    ...unless `feature.experimental` is enabled in which case the
> default is "true".
> ?

That reads better.

But I am not sure about the wisdom of controlling between conflict
and true with this feature macro in the first place.

Between "conflict" and "true", the former forces the end user to
verify (or allows the end user to veto) the auto resolution by the
heuristics and is always a safer if more cumbersome option.  It's
not like blindly trusting the directory rename heuristics is the
bright future for all users, is it?

I would not set rerere.autoUpdate to true when feature.experimental
is set; for exactly the same reason, I do not find it reasonable to
set this to true with feature.experimental macro.

> ...
>> diff --git a/repo-settings.c b/repo-settings.c
>> index af93696343..e0673938c0 100644
>> --- a/repo-settings.c
>> +++ b/repo-settings.c
>> @@ -34,6 +34,18 @@ void prepare_repo_settings(struct repository *r)
>>                 free(strval);
>>         }
>>
>> +       if (!repo_config_get_maybe_bool(r, "merge.directoryrenames", &value))
>> +               r->settings.merge_directory_renames = value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
>
> Shouldn't that be "MERGE_DIRECTORY_RENAMES_NONE" instead of "0"?
>
>> diff --git a/repository.h b/repository.h
>> index e7a72e2341..b8e52dd48f 100644
>> --- a/repository.h
>> +++ b/repository.h
>> @@ -19,6 +19,20 @@ enum untracked_cache_setting {
>>         UNTRACKED_CACHE_WRITE = 2
>>  };
>>
>> +enum merge_directory_renames_setting {
>> +       MERGE_DIRECTORY_RENAMES_UNSET = -1,
>> +       MERGE_DIRECTORY_RENAMES_NONE = 0,
>> +       MERGE_DIRECTORY_RENAMES_CONFLICT = 1,
>> +       MERGE_DIRECTORY_RENAMES_TRUE = 2,
>> +};
>
> Thanks for adding these; makes things much nicer.  :-)
>
>
> Cheers,
> Elijah

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Derrick Stolee]

On 8/8/2019 2:59 PM, Junio C Hamano wrote:
> Elijah Newren <newren@gmail.com> writes:
> 
>>> --- a/Documentation/config/merge.txt
>>> +++ b/Documentation/config/merge.txt
>>> @@ -54,7 +54,8 @@ merge.directoryRenames::
>>>         moved into the new directory.  If set to "conflict", a conflict
>>>         will be reported for such paths.  If merge.renames is false,
>>>         merge.directoryRenames is ignored and treated as false.  Defaults
>>> -       to "conflict".
>>> +       to "conflict" unless `feature.experimental` is enabled and the
>>> +       default is "true".
>>
>> I have a hard time parsing that changed sentence.  Perhaps something like:
>>    ...unless `feature.experimental` is enabled in which case the
>> default is "true".
>> ?
> 
> That reads better.
> 
> But I am not sure about the wisdom of controlling between conflict
> and true with this feature macro in the first place.
> 
> Between "conflict" and "true", the former forces the end user to
> verify (or allows the end user to veto) the auto resolution by the
> heuristics and is always a safer if more cumbersome option.  It's
> not like blindly trusting the directory rename heuristics is the
> bright future for all users, is it?
> 
> I would not set rerere.autoUpdate to true when feature.experimental
> is set; for exactly the same reason, I do not find it reasonable to
> set this to true with feature.experimental macro.

OK. I can remove it from the feature.experimental variable.

Shall I keep the enum logic and the use of repo-settings.c? I can split
them out into a separate patch.

-Stolee

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Elijah Newren]

On Thu, Aug 8, 2019 at 12:00 PM Junio C Hamano <gitster@pobox.com> wrote:
>
> Elijah Newren <newren@gmail.com> writes:
>
> >> --- a/Documentation/config/merge.txt
> >> +++ b/Documentation/config/merge.txt
> >> @@ -54,7 +54,8 @@ merge.directoryRenames::
> >>         moved into the new directory.  If set to "conflict", a conflict
> >>         will be reported for such paths.  If merge.renames is false,
> >>         merge.directoryRenames is ignored and treated as false.  Defaults
> >> -       to "conflict".
> >> +       to "conflict" unless `feature.experimental` is enabled and the
> >> +       default is "true".
> >
> > I have a hard time parsing that changed sentence.  Perhaps something like:
> >    ...unless `feature.experimental` is enabled in which case the
> > default is "true".
> > ?
>
> That reads better.

:-)

> But I am not sure about the wisdom of controlling between conflict
> and true with this feature macro in the first place.
>
> Between "conflict" and "true", the former forces the end user to
> verify (or allows the end user to veto) the auto resolution by the
> heuristics and is always a safer if more cumbersome option.  It's
> not like blindly trusting the directory rename heuristics is the
> bright future for all users, is it?

I'm afraid I don't quite know what the intended usecase for
feature.experimental is, and I'm unfamiliar with both pack.useSparse
and fetch.negotiationAlgorithm for comparison.  If the intended
usecase is stuff that is expected to become the future for all users,
then I agree with you; "conflict" is the right default for both now
and in the future so feature.experimental should not change it.
However, Stollee's commit message started with "The
'feature.experimental' setting includes config options that are not
committed to become defaults".

If these are settings that are "intended" but not "committed" to
become defaults, then yes, you're absolutely right.  (And in such a
case, it might be nice to add that distinction to the commit message.)
 If there is some other purpose, it depends on what that purpose is.

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Junio C Hamano]

Elijah Newren <newren@gmail.com> writes:

> However, Stollee's commit message started with "The
> 'feature.experimental' setting includes config options that are not
> committed to become defaults".
>
> If these are settings that are "intended" but not "committed" to
> become defaults, then yes, you're absolutely right.  (And in such a
> case, it might be nice to add that distinction to the commit message.)

My understanding is that the only purpose of feature.experimental
is: "we want to change the default of X this way, but we do not know
if we missed an important use case that would be harmed by such a
change, so let's see if those who volunteered to be guinea pigs by
setting feature.experimental to true find glitches in our thinking
that led us to think the change of the default is a good idea".

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Elijah Newren]

On Thu, Aug 8, 2019 at 12:12 PM Derrick Stolee <stolee@gmail.com> wrote:
>
> On 8/8/2019 2:59 PM, Junio C Hamano wrote:
> > Elijah Newren <newren@gmail.com> writes:
> >
> >>> --- a/Documentation/config/merge.txt
> >>> +++ b/Documentation/config/merge.txt
> >>> @@ -54,7 +54,8 @@ merge.directoryRenames::
> >>>         moved into the new directory.  If set to "conflict", a conflict
> >>>         will be reported for such paths.  If merge.renames is false,
> >>>         merge.directoryRenames is ignored and treated as false.  Defaults
> >>> -       to "conflict".
> >>> +       to "conflict" unless `feature.experimental` is enabled and the
> >>> +       default is "true".
> >>
> >> I have a hard time parsing that changed sentence.  Perhaps something like:
> >>    ...unless `feature.experimental` is enabled in which case the
> >> default is "true".
> >> ?
> >
> > That reads better.
> >
> > But I am not sure about the wisdom of controlling between conflict
> > and true with this feature macro in the first place.
> >
> > Between "conflict" and "true", the former forces the end user to
> > verify (or allows the end user to veto) the auto resolution by the
> > heuristics and is always a safer if more cumbersome option.  It's
> > not like blindly trusting the directory rename heuristics is the
> > bright future for all users, is it?
> >
> > I would not set rerere.autoUpdate to true when feature.experimental
> > is set; for exactly the same reason, I do not find it reasonable to
> > set this to true with feature.experimental macro.
>
> OK. I can remove it from the feature.experimental variable.
>
> Shall I keep the enum logic and the use of repo-settings.c? I can split
> them out into a separate patch.

Good question.  In part, I was looking at this ds/feature-macros
series because my cleanup-merge-api series[1] has some minor conflicts
with it.  I'm a little unsure what route I should take with my series
now.  Some choices:
  * keep this logic in your series in a separate patch, and have me
rebase my series on yours.
  * drop this logic from your series since it may not be needed
anymore, making our two series independent again.
  * move this logic into a separate patch of yours but making that
patch part of my series instead; that'd be easy with the enum logic,
but the repo-settings.c stuff appears to depend on your other
patches...

Thoughts or preferences?

[1] https://public-inbox.org/git/20190726155258.28561-1-newren@gmail.com/

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Derrick Stolee]

On 8/8/2019 4:07 PM, Junio C Hamano wrote:
> Elijah Newren <newren@gmail.com> writes:
> 
>> However, Stollee's commit message started with "The
>> 'feature.experimental' setting includes config options that are not
>> committed to become defaults".
>>
>> If these are settings that are "intended" but not "committed" to
>> become defaults, then yes, you're absolutely right.  (And in such a
>> case, it might be nice to add that distinction to the commit message.)
> 
> My understanding is that the only purpose of feature.experimental
> is: "we want to change the default of X this way, but we do not know
> if we missed an important use case that would be harmed by such a
> change, so let's see if those who volunteered to be guinea pigs by
> setting feature.experimental to true find glitches in our thinking
> that led us to think the change of the default is a good idea".

This is the basic idea. "Here are some new features that we think
are good, please opt-in to test them out and find corner cases."

Thanks,
-Stolee

Re: [PATCH v3 5/5] repo-settings: create feature.experimental setting

[Derrick Stolee]

On 8/8/2019 4:31 PM, Elijah Newren wrote:
> On Thu, Aug 8, 2019 at 12:12 PM Derrick Stolee <stolee@gmail.com> wrote:
>>
>> On 8/8/2019 2:59 PM, Junio C Hamano wrote:
>>> Elijah Newren <newren@gmail.com> writes:
>>>
>>>>> --- a/Documentation/config/merge.txt
>>>>> +++ b/Documentation/config/merge.txt
>>>>> @@ -54,7 +54,8 @@ merge.directoryRenames::
>>>>>         moved into the new directory.  If set to "conflict", a conflict
>>>>>         will be reported for such paths.  If merge.renames is false,
>>>>>         merge.directoryRenames is ignored and treated as false.  Defaults
>>>>> -       to "conflict".
>>>>> +       to "conflict" unless `feature.experimental` is enabled and the
>>>>> +       default is "true".
>>>>
>>>> I have a hard time parsing that changed sentence.  Perhaps something like:
>>>>    ...unless `feature.experimental` is enabled in which case the
>>>> default is "true".
>>>> ?
>>>
>>> That reads better.
>>>
>>> But I am not sure about the wisdom of controlling between conflict
>>> and true with this feature macro in the first place.
>>>
>>> Between "conflict" and "true", the former forces the end user to
>>> verify (or allows the end user to veto) the auto resolution by the
>>> heuristics and is always a safer if more cumbersome option.  It's
>>> not like blindly trusting the directory rename heuristics is the
>>> bright future for all users, is it?
>>>
>>> I would not set rerere.autoUpdate to true when feature.experimental
>>> is set; for exactly the same reason, I do not find it reasonable to
>>> set this to true with feature.experimental macro.
>>
>> OK. I can remove it from the feature.experimental variable.
>>
>> Shall I keep the enum logic and the use of repo-settings.c? I can split
>> them out into a separate patch.
> 
> Good question.  In part, I was looking at this ds/feature-macros
> series because my cleanup-merge-api series[1] has some minor conflicts
> with it.  I'm a little unsure what route I should take with my series
> now.  Some choices:
>   * keep this logic in your series in a separate patch, and have me
> rebase my series on yours.
>   * drop this logic from your series since it may not be needed
> anymore, making our two series independent again.
>   * move this logic into a separate patch of yours but making that
> patch part of my series instead; that'd be easy with the enum logic,
> but the repo-settings.c stuff appears to depend on your other
> patches...
> 
> Thoughts or preferences?

I've done the work in my GGG PR [1] to split the commit into two,
with the merge.directoryRenames on top. I created a tag [2] at that
commit and I'll send my v4 without changes to merge.directoryRenames.

I'll let you decide if you like placing the config in repo-settings
or not (or want to adapt the enum logic in a different place).

Thanks,
-Stolee

[1] https://github.com/gitgitgadget/git/pull/292
[2] https://github.com/derrickstolee/git/releases/tag/merge.directoryRenames

[PATCH v4 0/6] Create 'feature.*' config area and some centralized config parsing

[Derrick Stolee via GitGitGadget]

This is a brand-new thread to replace ds/early-access. The discussion on
that thread was very helpful [1].

With this in mind, I propose instead a set of "feature.*" config settings
that form groups of "community recommended" settings (with some caveats). In
the space below, I'll list a set of possible feature names and the implied
config options.

First, the main two categories we've discussed so far: many commits and many
files. These two feature sets are for when your repo is large in one of
these dimensions. Perhaps there are other settings to include in these?

feature.manyFiles:
    index.version = 4
    core.untrackedCache = true

UPDATE: Instead of "feature.manyCommits", this series includes making
core.commitGraph and gc.writeCommitGraph true by default!

The other category that has been discussed already is that of "experimental
features that we generally think are helpful but change behavior slightly in
some cases".

feature.experimental:
    pack.useSparse = true
    merge.directoryRenames = true
    fetch.negotiationAlgorithm = skipping

Specifically, this setting is for config values we are not sure will ever be
on by default, but additional testing is needed to be sure. This is
different than a possible 'feature.preview' setting that would include
config settings that we are committed to updating the defaults in a future
release. There are many ways we can take this idea in the future (including
more additions to these categories).

Thanks, -Stolee

Updates in V2: I'm responding to Dscho's comments early, as they were very
valuable.

 * Rearranged how we are checking for the feature.* settings, so if one
   layer turns the setting on and a later layer turns it off, we do not
   adjust the defaults.
   
   
 * Switched to using enums for the non-boolean config options.
   
   
 * Placed the repo_settings struct directly in the repository struct.
   
   
 * All struct members are initialized to -1 using memset.
   
   
 * The config changes around directory rename detection is greatly
   simplified.
   
   

Updates in V3: These are more Dscho comments, and are quite small.

 * Remove repo-settings.h by placing the struct and function definitions in
   repository.h.
   
   
 * Move the initialization int into "struct repo_settings".
   
   

Updates in V4:

 * Remove feature.manyCommits in favor of turning commit-graph on by
   default.
   
   
 * Update docs and 0 constant in feature.experimental around
   merge.directoryRenames.
   
   

[1] https://public-inbox.org/git/pull.254.git.gitgitgadget@gmail.com/

Derrick Stolee (6):
  repo-settings: consolidate some config settings
  t6501: use 'git gc' in quiet mode
  commit-graph: turn on commit-graph by default
  repo-settings: parse core.untrackedCache
  repo-settings: create feature.manyFiles setting
  repo-settings: create feature.experimental setting

 Documentation/config.txt             |  2 +
 Documentation/config/core.txt        |  6 ++-
 Documentation/config/feature.txt     | 29 +++++++++++++
 Documentation/config/fetch.txt       |  3 +-
 Documentation/config/gc.txt          |  2 +-
 Documentation/config/index.txt       |  1 +
 Documentation/config/pack.txt        |  3 +-
 Makefile                             |  1 +
 builtin/gc.c                         | 12 +++---
 builtin/pack-objects.c               |  8 ++--
 builtin/update-index.c               |  6 ++-
 commit-graph.c                       |  6 +--
 config.c                             | 24 -----------
 fetch-negotiator.c                   | 25 +++++------
 fetch-negotiator.h                   |  5 ++-
 fetch-pack.c                         | 11 +++--
 read-cache.c                         | 30 ++++++-------
 repo-settings.c                      | 64 ++++++++++++++++++++++++++++
 repository.h                         | 30 +++++++++++++
 t/t0410-partial-clone.sh             |  2 +-
 t/t1600-index.sh                     | 31 +++++++++++---
 t/t5307-pack-missing-commit.sh       |  4 +-
 t/t5324-split-commit-graph.sh        |  2 +
 t/t5552-skipping-fetch-negotiator.sh | 23 ----------
 t/t6011-rev-list-with-bad-commit.sh  |  2 +-
 t/t6501-freshen-objects.sh           |  6 +--
 26 files changed, 223 insertions(+), 115 deletions(-)
 create mode 100644 Documentation/config/feature.txt
 create mode 100644 repo-settings.c


base-commit: 9c9b961d7eb15fb583a2a812088713a68a85f1c0
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-292%2Fderrickstolee%2Frepo-settings%2Fhead-v4
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-292/derrickstolee/repo-settings/head-v4
Pull-Request: https://github.com/gitgitgadget/git/pull/292

Range-diff vs v3:

 1:  4d0db57ecb ! 1:  1a0b30b88a repo-settings: consolidate some config settings
     @@ -184,14 +184,6 @@
       diff --git a/repository.h b/repository.h
       --- a/repository.h
       +++ b/repository.h
     -@@
     - #include "path.h"
     - 
     - struct config_set;
     -+struct repo_settings;
     - struct git_hash_algo;
     - struct index_state;
     - struct lock_file;
      @@
       struct raw_object_store;
       struct submodule_cache;
 2:  c0129066a0 < -:  ---------- repo-settings: add feature.manyCommits setting
 -:  ---------- > 2:  6bf3584a60 t6501: use 'git gc' in quiet mode
 -:  ---------- > 3:  651e2d526b commit-graph: turn on commit-graph by default
 3:  b0a2d0e188 ! 4:  ec0abffd8b repo-settings: parse core.untrackedCache
     @@ -154,19 +154,12 @@
       
       	if (!repo_config_get_bool(r, "pack.usesparse", &value))
       		r->settings.pack_use_sparse = value;
     --
     -+	
     - 	if (!repo_config_get_bool(r, "feature.manycommits", &value) && value) {
     - 		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
     - 		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
     - 	}
      +
      +	/* Hack for test programs like test-dump-untracked-cache */
      +	if (ignore_untracked_cache_config)
      +		r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
      +	else
     -+		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
     -+
     ++		UPDATE_DEFAULT_BOOL(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
       }
      
       diff --git a/repository.h b/repository.h
 4:  0d4774d45f ! 5:  082fc57288 repo-settings: create feature.manyFiles setting
     @@ -13,6 +13,19 @@
      
          Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
      
     + diff --git a/Documentation/config.txt b/Documentation/config.txt
     + --- a/Documentation/config.txt
     + +++ b/Documentation/config.txt
     +@@
     + 
     + include::config/fastimport.txt[]
     + 
     ++include::config/feature.txt[]
     ++
     + include::config/fetch.txt[]
     + 
     + include::config/format.txt[]
     +
       diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
       --- a/Documentation/config/core.txt
       +++ b/Documentation/config/core.txt
     @@ -29,15 +42,15 @@
       	When missing or is set to `default`, many fields in the stat
      
       diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
     - --- a/Documentation/config/feature.txt
     + new file mode 100644
     + --- /dev/null
       +++ b/Documentation/config/feature.txt
      @@
     - * `core.commitGraph=true` enables reading the commit-graph file.
     - +
     - * `gc.writeCommitGraph=true` enables writing the commit-graph file during
     --garbage collection.
     - \ No newline at end of file
     -+garbage collection.
     ++feature.*::
     ++	The config settings that start with `feature.` modify the defaults of
     ++	a group of other config settings. These groups are created by the Git
     ++	developer community as recommended defaults and are subject to change.
     ++	In particular, new config options may be added with different defaults.
      +
      +feature.manyFiles::
      +	Enable config options that optimize for repos with many files in the
     @@ -48,7 +61,6 @@
      ++
      +* `core.untrackedCache=true` enables the untracked cache. This setting assumes
      +that mtime is working on your machine.
     - \ No newline at end of file
      
       diff --git a/Documentation/config/index.txt b/Documentation/config/index.txt
       --- a/Documentation/config/index.txt
     @@ -63,12 +75,15 @@
       --- a/repo-settings.c
       +++ b/repo-settings.c
      @@
     - 		UPDATE_DEFAULT(r->settings.core_commit_graph, 1);
     - 		UPDATE_DEFAULT(r->settings.gc_write_commit_graph, 1);
     + 		free(strval);
       	}
     + 
     +-
     + 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
     + 		r->settings.pack_use_sparse = value;
      +	if (!repo_config_get_bool(r, "feature.manyfiles", &value) && value) {
     -+		UPDATE_DEFAULT(r->settings.index_version, 4);
     -+		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
     ++		UPDATE_DEFAULT_BOOL(r->settings.index_version, 4);
     ++		UPDATE_DEFAULT_BOOL(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
      +	}
       
       	/* Hack for test programs like test-dump-untracked-cache */
 5:  2e153fac22 ! 6:  f30e25fcd6 repo-settings: create feature.experimental setting
     @@ -10,8 +10,6 @@
      
          * 'pack.useSparse=true'
      
     -    * 'merge.directoryRenames=true'
     -
          * 'fetch.negotiationAlgorithm=skipping'
      
          In the case of fetch.negotiationAlgorithm, the existing logic
     @@ -41,15 +39,12 @@
      +* `pack.useSparse=true` uses a new algorithm when constructing a pack-file
      +which can improve `git push` performance in repos with many files.
      ++
     -+* `merge.directoryRenames=true` uses a new algorithm for detecting renames by
     -+using entire directories at a time instead of single files at a time.
     -++
      +* `fetch.negotiationAlgorithm=skipping` may improve fetch negotiation times by
      +skipping more commits at a time, reducing the number of round trips.
      +
     - feature.manyCommits::
     - 	Enable config options that optimize for repos with many commits. This
     - 	setting is recommended for repos with at least 100,000 commits. The
     + feature.manyFiles::
     + 	Enable config options that optimize for repos with many files in the
     + 	working directory. With many files, commands such as `git status` and
      
       diff --git a/Documentation/config/fetch.txt b/Documentation/config/fetch.txt
       --- a/Documentation/config/fetch.txt
     @@ -65,20 +60,6 @@
       +
       See also the `--negotiation-tip` option for linkgit:git-fetch[1].
      
     - diff --git a/Documentation/config/merge.txt b/Documentation/config/merge.txt
     - --- a/Documentation/config/merge.txt
     - +++ b/Documentation/config/merge.txt
     -@@
     - 	moved into the new directory.  If set to "conflict", a conflict
     - 	will be reported for such paths.  If merge.renames is false,
     - 	merge.directoryRenames is ignored and treated as false.  Defaults
     --	to "conflict".
     -+	to "conflict" unless `feature.experimental` is enabled and the
     -+	default is "true".
     - 
     - merge.renormalize::
     - 	Tell Git that canonical representation of files in the
     -
       diff --git a/Documentation/config/pack.txt b/Documentation/config/pack.txt
       --- a/Documentation/config/pack.txt
       +++ b/Documentation/config/pack.txt
     @@ -214,38 +195,6 @@
       	git_config(fetch_pack_config_cb, NULL);
       }
      
     - diff --git a/merge-recursive.c b/merge-recursive.c
     - --- a/merge-recursive.c
     - +++ b/merge-recursive.c
     -@@
     - 		opt->merge_detect_rename = git_config_rename("merge.renames", value);
     - 		free(value);
     - 	}
     --	if (!git_config_get_string("merge.directoryrenames", &value)) {
     --		int boolval = git_parse_maybe_bool(value);
     --		if (0 <= boolval) {
     --			opt->detect_directory_renames = boolval ? 2 : 0;
     --		} else if (!strcasecmp(value, "conflict")) {
     --			opt->detect_directory_renames = 1;
     --		} /* avoid erroring on values from future versions of git */
     --		free(value);
     --	}
     - 	git_config(git_xmerge_config, NULL);
     - }
     - 
     -@@
     - 	opt->diff_detect_rename = -1;
     - 	opt->merge_detect_rename = -1;
     - 	opt->detect_directory_renames = 1;
     -+
     -+	prepare_repo_settings(repo);
     -+	if (repo->settings.merge_directory_renames >= 0)
     -+		opt->detect_directory_renames = repo->settings.merge_directory_renames;
     -+
     - 	merge_recursive_config(opt);
     - 	merge_verbosity = getenv("GIT_MERGE_VERBOSITY");
     - 	if (merge_verbosity)
     -
       diff --git a/repo-settings.c b/repo-settings.c
       --- a/repo-settings.c
       +++ b/repo-settings.c
     @@ -253,39 +202,31 @@
       		free(strval);
       	}
       
     -+	if (!repo_config_get_maybe_bool(r, "merge.directoryrenames", &value))
     -+		r->settings.merge_directory_renames = value ? MERGE_DIRECTORY_RENAMES_TRUE : 0;
     -+	else if (!repo_config_get_string(r, "merge.directoryrenames", &strval)) {
     -+		if (!strcasecmp(strval, "conflict"))
     -+			r->settings.merge_directory_renames = MERGE_DIRECTORY_RENAMES_CONFLICT;
     -+	}
      +	if (!repo_config_get_string(r, "fetch.negotiationalgorithm", &strval)) {
      +		if (!strcasecmp(strval, "skipping"))
      +			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_SKIPPING;
      +		else
      +			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_DEFAULT;
      +	}
     - 
     ++
       	if (!repo_config_get_bool(r, "pack.usesparse", &value))
       		r->settings.pack_use_sparse = value;
     -@@
     - 		UPDATE_DEFAULT(r->settings.index_version, 4);
     - 		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
     + 	if (!repo_config_get_bool(r, "feature.manyfiles", &value) && value) {
     + 		UPDATE_DEFAULT_BOOL(r->settings.index_version, 4);
     + 		UPDATE_DEFAULT_BOOL(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
       	}
      +	if (!repo_config_get_bool(r, "feature.experimental", &value) && value) {
     -+		UPDATE_DEFAULT(r->settings.pack_use_sparse, 1);
     -+		UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_TRUE);
     -+		UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
     ++		UPDATE_DEFAULT_BOOL(r->settings.pack_use_sparse, 1);
     ++		UPDATE_DEFAULT_BOOL(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
      +	}
       
       	/* Hack for test programs like test-dump-untracked-cache */
       	if (ignore_untracked_cache_config)
     -@@
     + 		r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
       	else
     - 		UPDATE_DEFAULT(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
     - 
     -+	UPDATE_DEFAULT(r->settings.merge_directory_renames, MERGE_DIRECTORY_RENAMES_CONFLICT);
     -+	UPDATE_DEFAULT(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
     + 		UPDATE_DEFAULT_BOOL(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
     ++
     ++	UPDATE_DEFAULT_BOOL(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
       }
      
       diff --git a/repository.h b/repository.h
     @@ -295,13 +236,6 @@
       	UNTRACKED_CACHE_WRITE = 2
       };
       
     -+enum merge_directory_renames_setting {
     -+	MERGE_DIRECTORY_RENAMES_UNSET = -1,
     -+	MERGE_DIRECTORY_RENAMES_NONE = 0,
     -+	MERGE_DIRECTORY_RENAMES_CONFLICT = 1,
     -+	MERGE_DIRECTORY_RENAMES_TRUE = 2,
     -+};
     -+
      +enum fetch_negotiation_setting {
      +	FETCH_NEGOTIATION_UNSET = -1,
      +	FETCH_NEGOTIATION_NONE = 0,
     @@ -316,7 +250,6 @@
       	enum untracked_cache_setting core_untracked_cache;
       
       	int pack_use_sparse;
     -+	enum merge_directory_renames_setting merge_directory_renames;
      +	enum fetch_negotiation_setting fetch_negotiation_algorithm;
       };
       

-- 
gitgitgadget

[PATCH v4 1/6] repo-settings: consolidate some config settings

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

There are a few important config settings that are not loaded
during git_default_config. These are instead loaded on-demand.

Centralize these config options to a single scan, and store
all of the values in a repo_settings struct. The values for
each setting are initialized as negative to indicate "unset".

This centralization will be particularly important in a later
change to introduce "meta" config settings that change the
defaults for these config settings.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Makefile               |  1 +
 builtin/gc.c           | 12 +++++-------
 builtin/pack-objects.c |  8 ++++----
 commit-graph.c         |  6 +++---
 read-cache.c           | 11 ++++++-----
 repo-settings.c        | 25 +++++++++++++++++++++++++
 repository.h           | 14 ++++++++++++++
 7 files changed, 58 insertions(+), 19 deletions(-)
 create mode 100644 repo-settings.c

diff --git a/Makefile b/Makefile
index 11ccea4071..032fe9b5f9 100644
--- a/Makefile
+++ b/Makefile
@@ -964,6 +964,7 @@ LIB_OBJS += refspec.o
 LIB_OBJS += ref-filter.o
 LIB_OBJS += remote.o
 LIB_OBJS += replace-object.o
+LIB_OBJS += repo-settings.o
 LIB_OBJS += repository.o
 LIB_OBJS += rerere.o
 LIB_OBJS += resolve-undo.o
diff --git a/builtin/gc.c b/builtin/gc.c
index c18efadda5..4b8fbb9965 100644
--- a/builtin/gc.c
+++ b/builtin/gc.c
@@ -41,7 +41,6 @@ static int aggressive_depth = 50;
 static int aggressive_window = 250;
 static int gc_auto_threshold = 6700;
 static int gc_auto_pack_limit = 50;
-static int gc_write_commit_graph;
 static int detach_auto = 1;
 static timestamp_t gc_log_expire_time;
 static const char *gc_log_expire = "1.day.ago";
@@ -148,7 +147,6 @@ static void gc_config(void)
 	git_config_get_int("gc.aggressivedepth", &aggressive_depth);
 	git_config_get_int("gc.auto", &gc_auto_threshold);
 	git_config_get_int("gc.autopacklimit", &gc_auto_pack_limit);
-	git_config_get_bool("gc.writecommitgraph", &gc_write_commit_graph);
 	git_config_get_bool("gc.autodetach", &detach_auto);
 	git_config_get_expiry("gc.pruneexpire", &prune_expire);
 	git_config_get_expiry("gc.worktreepruneexpire", &prune_worktrees_expire);
@@ -685,11 +683,11 @@ int cmd_gc(int argc, const char **argv, const char *prefix)
 		clean_pack_garbage();
 	}
 
-	if (gc_write_commit_graph &&
-	    write_commit_graph_reachable(get_object_directory(),
-					 !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
-					 NULL))
-		return 1;
+	prepare_repo_settings(the_repository);
+	if (the_repository->settings.gc_write_commit_graph == 1)
+		write_commit_graph_reachable(get_object_directory(),
+					     !quiet && !daemonized ? COMMIT_GRAPH_PROGRESS : 0,
+					     NULL);
 
 	if (auto_gc && too_many_loose_objects())
 		warning(_("There are too many unreachable loose objects; "
diff --git a/builtin/pack-objects.c b/builtin/pack-objects.c
index 267c562b1f..f9f7181f3f 100644
--- a/builtin/pack-objects.c
+++ b/builtin/pack-objects.c
@@ -2709,10 +2709,6 @@ static int git_pack_config(const char *k, const char *v, void *cb)
 		use_bitmap_index_default = git_config_bool(k, v);
 		return 0;
 	}
-	if (!strcmp(k, "pack.usesparse")) {
-		sparse = git_config_bool(k, v);
-		return 0;
-	}
 	if (!strcmp(k, "pack.threads")) {
 		delta_search_threads = git_config_int(k, v);
 		if (delta_search_threads < 0)
@@ -3332,6 +3328,10 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 	read_replace_refs = 0;
 
 	sparse = git_env_bool("GIT_TEST_PACK_SPARSE", 0);
+	prepare_repo_settings(the_repository);
+	if (!sparse && the_repository->settings.pack_use_sparse != -1)
+		sparse = the_repository->settings.pack_use_sparse;
+
 	reset_pack_idx_option(&pack_idx_opts);
 	git_config(git_pack_config, NULL);
 
diff --git a/commit-graph.c b/commit-graph.c
index b3c4de79b6..7854e491b2 100644
--- a/commit-graph.c
+++ b/commit-graph.c
@@ -466,7 +466,6 @@ static void prepare_commit_graph_one(struct repository *r, const char *obj_dir)
 static int prepare_commit_graph(struct repository *r)
 {
 	struct object_directory *odb;
-	int config_value;
 
 	if (git_env_bool(GIT_TEST_COMMIT_GRAPH_DIE_ON_LOAD, 0))
 		die("dying as requested by the '%s' variable on commit-graph load!",
@@ -476,9 +475,10 @@ static int prepare_commit_graph(struct repository *r)
 		return !!r->objects->commit_graph;
 	r->objects->commit_graph_attempted = 1;
 
+	prepare_repo_settings(r);
+
 	if (!git_env_bool(GIT_TEST_COMMIT_GRAPH, 0) &&
-	    (repo_config_get_bool(r, "core.commitgraph", &config_value) ||
-	    !config_value))
+	    r->settings.core_commit_graph != 1)
 		/*
 		 * This repository is not configured to use commit graphs, so
 		 * do not load one. (But report commit_graph_attempted anyway
diff --git a/read-cache.c b/read-cache.c
index c701f7f8b8..59dbebc15d 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -1599,16 +1599,17 @@ struct cache_entry *refresh_cache_entry(struct index_state *istate,
 
 #define INDEX_FORMAT_DEFAULT 3
 
-static unsigned int get_index_format_default(void)
+static unsigned int get_index_format_default(struct repository *r)
 {
 	char *envversion = getenv("GIT_INDEX_VERSION");
 	char *endp;
-	int value;
 	unsigned int version = INDEX_FORMAT_DEFAULT;
 
 	if (!envversion) {
-		if (!git_config_get_int("index.version", &value))
-			version = value;
+		prepare_repo_settings(r);
+
+		if (r->settings.index_version >= 0)
+			version = r->settings.index_version;
 		if (version < INDEX_FORMAT_LB || INDEX_FORMAT_UB < version) {
 			warning(_("index.version set, but the value is invalid.\n"
 				  "Using version %i"), INDEX_FORMAT_DEFAULT);
@@ -2765,7 +2766,7 @@ static int do_write_index(struct index_state *istate, struct tempfile *tempfile,
 	}
 
 	if (!istate->version) {
-		istate->version = get_index_format_default();
+		istate->version = get_index_format_default(the_repository);
 		if (git_env_bool("GIT_TEST_SPLIT_INDEX", 0))
 			init_split_index(istate);
 	}
diff --git a/repo-settings.c b/repo-settings.c
new file mode 100644
index 0000000000..309577f6bc
--- /dev/null
+++ b/repo-settings.c
@@ -0,0 +1,25 @@
+#include "cache.h"
+#include "config.h"
+#include "repository.h"
+
+void prepare_repo_settings(struct repository *r)
+{
+	int value;
+
+	if (r->settings.initialized)
+		return;
+
+	/* Defaults */
+	memset(&r->settings, -1, sizeof(r->settings));
+
+	if (!repo_config_get_bool(r, "core.commitgraph", &value))
+		r->settings.core_commit_graph = value;
+	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
+		r->settings.gc_write_commit_graph = value;
+
+	if (!repo_config_get_bool(r, "index.version", &value))
+		r->settings.index_version = value;
+
+	if (!repo_config_get_bool(r, "pack.usesparse", &value))
+		r->settings.pack_use_sparse = value;
+}
diff --git a/repository.h b/repository.h
index 4fb6a5885f..cc285ad327 100644
--- a/repository.h
+++ b/repository.h
@@ -11,6 +11,17 @@ struct pathspec;
 struct raw_object_store;
 struct submodule_cache;
 
+struct repo_settings {
+	int initialized;
+
+	int core_commit_graph;
+	int gc_write_commit_graph;
+
+	int index_version;
+
+	int pack_use_sparse;
+};
+
 struct repository {
 	/* Environment */
 	/*
@@ -72,6 +83,8 @@ struct repository {
 	 */
 	char *submodule_prefix;
 
+	struct repo_settings settings;
+
 	/* Subsystems */
 	/*
 	 * Repository's config which contains key-value pairs from the usual
@@ -157,5 +170,6 @@ int repo_read_index_unmerged(struct repository *);
  */
 void repo_update_index_if_able(struct repository *, struct lock_file *);
 
+void prepare_repo_settings(struct repository *r);
 
 #endif /* REPOSITORY_H */
-- 
gitgitgadget

[PATCH v4 2/6] t6501: use 'git gc' in quiet mode

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

t6501-freshen-objects.sh sends the standard error from
'git gc' to a file and verifies that it is empty. This
is intended as a way to ensure no warnings are written
during the operation. However, as the commit-graph is
added as a step to 'git gc', its progress will appear
in the output.

Pass the '-q' argument to avoid a failing test case
when progress is written.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 t/t6501-freshen-objects.sh | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/t/t6501-freshen-objects.sh b/t/t6501-freshen-objects.sh
index 033871ee5f..f30b4849b6 100755
--- a/t/t6501-freshen-objects.sh
+++ b/t/t6501-freshen-objects.sh
@@ -137,7 +137,7 @@ test_expect_success 'do not complain about existing broken links (commit)' '
 	some message
 	EOF
 	commit=$(git hash-object -t commit -w broken-commit) &&
-	git gc 2>stderr &&
+	git gc -q 2>stderr &&
 	verbose git cat-file -e $commit &&
 	test_must_be_empty stderr
 '
@@ -147,7 +147,7 @@ test_expect_success 'do not complain about existing broken links (tree)' '
 	100644 blob 0000000000000000000000000000000000000003	foo
 	EOF
 	tree=$(git mktree --missing <broken-tree) &&
-	git gc 2>stderr &&
+	git gc -q 2>stderr &&
 	git cat-file -e $tree &&
 	test_must_be_empty stderr
 '
@@ -162,7 +162,7 @@ test_expect_success 'do not complain about existing broken links (tag)' '
 	this is a broken tag
 	EOF
 	tag=$(git hash-object -t tag -w broken-tag) &&
-	git gc 2>stderr &&
+	git gc -q 2>stderr &&
 	git cat-file -e $tag &&
 	test_must_be_empty stderr
 '
-- 
gitgitgadget

[PATCH v4 3/6] commit-graph: turn on commit-graph by default

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The commit-graph feature has seen a lot of activity in the past
year or so since it was introduced. The feature is a critical
performance enhancement for medium- to large-sized repos, and
does not significantly hurt small repos.

Change the defaults for core.commitGraph and gc.writeCommitGraph
to true so users benefit from this feature by default.

There are several places in the test suite where the environment
variable GIT_TEST_COMMIT_GRAPH is disabled to avoid reading a
commit-graph, if it exists. The config option overrides the
environment, so swap these. Some GIT_TEST_COMMIT_GRAPH assignments
remain, and those are to avoid writing a commit-graph when a new
commit is created.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config/core.txt       | 2 +-
 Documentation/config/gc.txt         | 2 +-
 repo-settings.c                     | 4 ++++
 t/t0410-partial-clone.sh            | 2 +-
 t/t5307-pack-missing-commit.sh      | 4 ++--
 t/t5324-split-commit-graph.sh       | 2 ++
 t/t6011-rev-list-with-bad-commit.sh | 2 +-
 7 files changed, 12 insertions(+), 6 deletions(-)

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index 75538d27e7..e66d79fd76 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -577,7 +577,7 @@ the `GIT_NOTES_REF` environment variable.  See linkgit:git-notes[1].
 
 core.commitGraph::
 	If true, then git will read the commit-graph file (if it exists)
-	to parse the graph structure of commits. Defaults to false. See
+	to parse the graph structure of commits. Defaults to true. See
 	linkgit:git-commit-graph[1] for more information.
 
 core.useReplaceRefs::
diff --git a/Documentation/config/gc.txt b/Documentation/config/gc.txt
index 02b92b18b5..00ea0a678e 100644
--- a/Documentation/config/gc.txt
+++ b/Documentation/config/gc.txt
@@ -63,7 +63,7 @@ gc.writeCommitGraph::
 	If true, then gc will rewrite the commit-graph file when
 	linkgit:git-gc[1] is run. When using `git gc --auto`
 	the commit-graph will be updated if housekeeping is
-	required. Default is false. See linkgit:git-commit-graph[1]
+	required. Default is true. See linkgit:git-commit-graph[1]
 	for details.
 
 gc.logExpiry::
diff --git a/repo-settings.c b/repo-settings.c
index 309577f6bc..d00b675687 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -2,6 +2,8 @@
 #include "config.h"
 #include "repository.h"
 
+#define UPDATE_DEFAULT_BOOL(s,v) do { if (s == -1) { s = v; } } while(0)
+
 void prepare_repo_settings(struct repository *r)
 {
 	int value;
@@ -16,6 +18,8 @@ void prepare_repo_settings(struct repository *r)
 		r->settings.core_commit_graph = value;
 	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
 		r->settings.gc_write_commit_graph = value;
+	UPDATE_DEFAULT_BOOL(r->settings.core_commit_graph, 1);
+	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);
 
 	if (!repo_config_get_bool(r, "index.version", &value))
 		r->settings.index_version = value;
diff --git a/t/t0410-partial-clone.sh b/t/t0410-partial-clone.sh
index 5bd892f2f7..181ffa44e9 100755
--- a/t/t0410-partial-clone.sh
+++ b/t/t0410-partial-clone.sh
@@ -234,7 +234,7 @@ test_expect_success 'rev-list stops traversal at missing and promised commit' '
 
 	git -C repo config core.repositoryformatversion 1 &&
 	git -C repo config extensions.partialclone "arbitrary string" &&
-	GIT_TEST_COMMIT_GRAPH=0 git -C repo rev-list --exclude-promisor-objects --objects bar >out &&
+	GIT_TEST_COMMIT_GRAPH=0 git -C repo -c core.commitGraph=false rev-list --exclude-promisor-objects --objects bar >out &&
 	grep $(git -C repo rev-parse bar) out &&
 	! grep $FOO out
 '
diff --git a/t/t5307-pack-missing-commit.sh b/t/t5307-pack-missing-commit.sh
index dacb440b27..f4338abb78 100755
--- a/t/t5307-pack-missing-commit.sh
+++ b/t/t5307-pack-missing-commit.sh
@@ -24,11 +24,11 @@ test_expect_success 'check corruption' '
 '
 
 test_expect_success 'rev-list notices corruption (1)' '
-	test_must_fail env GIT_TEST_COMMIT_GRAPH=0 git rev-list HEAD
+	test_must_fail env GIT_TEST_COMMIT_GRAPH=0 git -c core.commitGraph=false rev-list HEAD
 '
 
 test_expect_success 'rev-list notices corruption (2)' '
-	test_must_fail env GIT_TEST_COMMIT_GRAPH=0 git rev-list --objects HEAD
+	test_must_fail env GIT_TEST_COMMIT_GRAPH=0 git -c core.commitGraph=false rev-list --objects HEAD
 '
 
 test_expect_success 'pack-objects notices corruption' '
diff --git a/t/t5324-split-commit-graph.sh b/t/t5324-split-commit-graph.sh
index 03f45a1ed9..19aa40de15 100755
--- a/t/t5324-split-commit-graph.sh
+++ b/t/t5324-split-commit-graph.sh
@@ -8,6 +8,7 @@ GIT_TEST_COMMIT_GRAPH=0
 test_expect_success 'setup repo' '
 	git init &&
 	git config core.commitGraph true &&
+	git config gc.writeCommitGraph false &&
 	infodir=".git/objects/info" &&
 	graphdir="$infodir/commit-graphs" &&
 	test_oid_init
@@ -332,6 +333,7 @@ test_expect_success 'split across alternate where alternate is not split' '
 	git clone --no-hardlinks . alt-split &&
 	(
 		cd alt-split &&
+		rm -f .git/objects/info/commit-graph &&
 		echo "$(pwd)"/../.git/objects >.git/objects/info/alternates &&
 		test_commit 18 &&
 		git commit-graph write --reachable --split &&
diff --git a/t/t6011-rev-list-with-bad-commit.sh b/t/t6011-rev-list-with-bad-commit.sh
index 545b461e51..bad02cf5b8 100755
--- a/t/t6011-rev-list-with-bad-commit.sh
+++ b/t/t6011-rev-list-with-bad-commit.sh
@@ -42,7 +42,7 @@ test_expect_success 'corrupt second commit object' \
    '
 
 test_expect_success 'rev-list should fail' '
-	test_must_fail env GIT_TEST_COMMIT_GRAPH=0 git rev-list --all > /dev/null
+	test_must_fail env GIT_TEST_COMMIT_GRAPH=0 git -c core.commitGraph=false rev-list --all > /dev/null
 '
 
 test_expect_success 'git repack _MUST_ fail' \
-- 
gitgitgadget

[PATCH v4 4/6] repo-settings: parse core.untrackedCache

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The core.untrackedCache config setting is slightly complicated,
so clarify its use and centralize its parsing into the repo
settings.

The default value is "keep" (returned as -1), which persists the
untracked cache if it exists.

If the value is set as "false" (returned as 0), then remove the
untracked cache if it exists.

If the value is set as "true" (returned as 1), then write the
untracked cache and persist it.

Instead of relying on magic values of -1, 0, and 1, split these
options into an enum. This allows the use of "-1" as a
default value. After parsing the config options, if the value is
unset we can initialize it to UNTRACKED_CACHE_KEEP.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 builtin/update-index.c |  6 ++++--
 config.c               | 24 ------------------------
 read-cache.c           | 19 +++++++++----------
 repo-settings.c        | 19 +++++++++++++++++++
 repository.h           |  8 ++++++++
 5 files changed, 40 insertions(+), 36 deletions(-)

diff --git a/builtin/update-index.c b/builtin/update-index.c
index dff2f4b837..49302d98c5 100644
--- a/builtin/update-index.c
+++ b/builtin/update-index.c
@@ -966,6 +966,7 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 	struct parse_opt_ctx_t ctx;
 	strbuf_getline_fn getline_fn;
 	int parseopt_state = PARSE_OPT_UNKNOWN;
+	struct repository *r = the_repository;
 	struct option options[] = {
 		OPT_BIT('q', NULL, &refresh_args.flags,
 			N_("continue refresh even when index needs update"),
@@ -1180,11 +1181,12 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 		remove_split_index(&the_index);
 	}
 
+	prepare_repo_settings(r);
 	switch (untracked_cache) {
 	case UC_UNSPECIFIED:
 		break;
 	case UC_DISABLE:
-		if (git_config_get_untracked_cache() == 1)
+		if (r->settings.core_untracked_cache == UNTRACKED_CACHE_WRITE)
 			warning(_("core.untrackedCache is set to true; "
 				  "remove or change it, if you really want to "
 				  "disable the untracked cache"));
@@ -1196,7 +1198,7 @@ int cmd_update_index(int argc, const char **argv, const char *prefix)
 		return !test_if_untracked_cache_is_supported();
 	case UC_ENABLE:
 	case UC_FORCE:
-		if (git_config_get_untracked_cache() == 0)
+		if (r->settings.core_untracked_cache == UNTRACKED_CACHE_REMOVE)
 			warning(_("core.untrackedCache is set to false; "
 				  "remove or change it, if you really want to "
 				  "enable the untracked cache"));
diff --git a/config.c b/config.c
index faa57e436c..3241dbc54d 100644
--- a/config.c
+++ b/config.c
@@ -2277,30 +2277,6 @@ int git_config_get_expiry_in_days(const char *key, timestamp_t *expiry, timestam
 	return -1; /* thing exists but cannot be parsed */
 }
 
-int git_config_get_untracked_cache(void)
-{
-	int val = -1;
-	const char *v;
-
-	/* Hack for test programs like test-dump-untracked-cache */
-	if (ignore_untracked_cache_config)
-		return -1;
-
-	if (!git_config_get_maybe_bool("core.untrackedcache", &val))
-		return val;
-
-	if (!git_config_get_value("core.untrackedcache", &v)) {
-		if (!strcasecmp(v, "keep"))
-			return -1;
-
-		error(_("unknown core.untrackedCache value '%s'; "
-			"using 'keep' default value"), v);
-		return -1;
-	}
-
-	return -1; /* default value */
-}
-
 int git_config_get_split_index(void)
 {
 	int val;
diff --git a/read-cache.c b/read-cache.c
index 59dbebc15d..7a07286164 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -1845,18 +1845,17 @@ static void check_ce_order(struct index_state *istate)
 
 static void tweak_untracked_cache(struct index_state *istate)
 {
-	switch (git_config_get_untracked_cache()) {
-	case -1: /* keep: do nothing */
-		break;
-	case 0: /* false */
+	struct repository *r = the_repository;
+
+	prepare_repo_settings(r);
+
+	if (r->settings.core_untracked_cache  == UNTRACKED_CACHE_REMOVE) {
 		remove_untracked_cache(istate);
-		break;
-	case 1: /* true */
-		add_untracked_cache(istate);
-		break;
-	default: /* unknown value: do nothing */
-		break;
+		return;
 	}
+
+	if (r->settings.core_untracked_cache == UNTRACKED_CACHE_WRITE)
+		add_untracked_cache(istate);
 }
 
 static void tweak_split_index(struct index_state *istate)
diff --git a/repo-settings.c b/repo-settings.c
index d00b675687..abbc6566f8 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -7,6 +7,7 @@
 void prepare_repo_settings(struct repository *r)
 {
 	int value;
+	char *strval;
 
 	if (r->settings.initialized)
 		return;
@@ -23,7 +24,25 @@ void prepare_repo_settings(struct repository *r)
 
 	if (!repo_config_get_bool(r, "index.version", &value))
 		r->settings.index_version = value;
+	if (!repo_config_get_maybe_bool(r, "core.untrackedcache", &value)) {
+		if (value == 0)
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_REMOVE;
+		else
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_WRITE;
+	} else if (!repo_config_get_string(r, "core.untrackedcache", &strval)) {
+		if (!strcasecmp(strval, "keep"))
+			r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
+
+		free(strval);
+	}
+
 
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
+
+	/* Hack for test programs like test-dump-untracked-cache */
+	if (ignore_untracked_cache_config)
+		r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
+	else
+		UPDATE_DEFAULT_BOOL(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
 }
diff --git a/repository.h b/repository.h
index cc285ad327..cf7ff0778c 100644
--- a/repository.h
+++ b/repository.h
@@ -11,6 +11,13 @@ struct pathspec;
 struct raw_object_store;
 struct submodule_cache;
 
+enum untracked_cache_setting {
+	UNTRACKED_CACHE_UNSET = -1,
+	UNTRACKED_CACHE_REMOVE = 0,
+	UNTRACKED_CACHE_KEEP = 1,
+	UNTRACKED_CACHE_WRITE = 2
+};
+
 struct repo_settings {
 	int initialized;
 
@@ -18,6 +25,7 @@ struct repo_settings {
 	int gc_write_commit_graph;
 
 	int index_version;
+	enum untracked_cache_setting core_untracked_cache;
 
 	int pack_use_sparse;
 };
-- 
gitgitgadget

[PATCH v4 5/6] repo-settings: create feature.manyFiles setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The feature.manyFiles setting is suitable for repos with many
files in the working directory. By setting index.version=4 and
core.untrackedCache=true, commands such as 'git status' should
improve.

While adding this setting, modify the index version precedence
tests to check how this setting overrides the default for
index.version is unset.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config.txt         |  2 ++
 Documentation/config/core.txt    |  4 +++-
 Documentation/config/feature.txt | 15 +++++++++++++++
 Documentation/config/index.txt   |  1 +
 repo-settings.c                  |  5 ++++-
 t/t1600-index.sh                 | 31 ++++++++++++++++++++++++++-----
 6 files changed, 51 insertions(+), 7 deletions(-)
 create mode 100644 Documentation/config/feature.txt

diff --git a/Documentation/config.txt b/Documentation/config.txt
index e3f5bc3396..77f3b1486b 100644
--- a/Documentation/config.txt
+++ b/Documentation/config.txt
@@ -345,6 +345,8 @@ include::config/difftool.txt[]
 
 include::config/fastimport.txt[]
 
+include::config/feature.txt[]
+
 include::config/fetch.txt[]
 
 include::config/format.txt[]
diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index e66d79fd76..852d2ba37a 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -86,7 +86,9 @@ core.untrackedCache::
 	it will automatically be removed, if set to `false`. Before
 	setting it to `true`, you should check that mtime is working
 	properly on your system.
-	See linkgit:git-update-index[1]. `keep` by default.
+	See linkgit:git-update-index[1]. `keep` by default, unless
+	`feature.manyFiles` is enabled which sets this setting to
+	`true` by default.
 
 core.checkStat::
 	When missing or is set to `default`, many fields in the stat
diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
new file mode 100644
index 0000000000..8ea198a642
--- /dev/null
+++ b/Documentation/config/feature.txt
@@ -0,0 +1,15 @@
+feature.*::
+	The config settings that start with `feature.` modify the defaults of
+	a group of other config settings. These groups are created by the Git
+	developer community as recommended defaults and are subject to change.
+	In particular, new config options may be added with different defaults.
+
+feature.manyFiles::
+	Enable config options that optimize for repos with many files in the
+	working directory. With many files, commands such as `git status` and
+	`git checkout` may be slow and these new defaults improve performance:
++
+* `index.version=4` enables path-prefix compression in the index.
++
+* `core.untrackedCache=true` enables the untracked cache. This setting assumes
+that mtime is working on your machine.
diff --git a/Documentation/config/index.txt b/Documentation/config/index.txt
index f181503041..7cb50b37e9 100644
--- a/Documentation/config/index.txt
+++ b/Documentation/config/index.txt
@@ -24,3 +24,4 @@ index.threads::
 index.version::
 	Specify the version with which new index files should be
 	initialized.  This does not affect existing repositories.
+	If `feature.manyFiles` is enabled, then the default is 4.
diff --git a/repo-settings.c b/repo-settings.c
index abbc6566f8..d5bf9069f4 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -36,9 +36,12 @@ void prepare_repo_settings(struct repository *r)
 		free(strval);
 	}
 
-
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
+	if (!repo_config_get_bool(r, "feature.manyfiles", &value) && value) {
+		UPDATE_DEFAULT_BOOL(r->settings.index_version, 4);
+		UPDATE_DEFAULT_BOOL(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
+	}
 
 	/* Hack for test programs like test-dump-untracked-cache */
 	if (ignore_untracked_cache_config)
diff --git a/t/t1600-index.sh b/t/t1600-index.sh
index 42962ed7d4..c77721b580 100755
--- a/t/t1600-index.sh
+++ b/t/t1600-index.sh
@@ -59,17 +59,38 @@ test_expect_success 'out of bounds index.version issues warning' '
 	)
 '
 
-test_expect_success 'GIT_INDEX_VERSION takes precedence over config' '
+test_index_version () {
+	INDEX_VERSION_CONFIG=$1 &&
+	FEATURE_MANY_FILES=$2 &&
+	ENV_VAR_VERSION=$3
+	EXPECTED_OUTPUT_VERSION=$4 &&
 	(
 		rm -f .git/index &&
-		GIT_INDEX_VERSION=4 &&
-		export GIT_INDEX_VERSION &&
-		git config --add index.version 2 &&
+		rm -f .git/config &&
+		if test "$INDEX_VERSION_CONFIG" -ne 0
+		then
+			git config --add index.version $INDEX_VERSION_CONFIG
+		fi &&
+		git config --add feature.manyFiles $FEATURE_MANY_FILES
+		if test "$ENV_VAR_VERSION" -ne 0
+		then
+			GIT_INDEX_VERSION=$ENV_VAR_VERSION &&
+			export GIT_INDEX_VERSION
+		else
+			unset GIT_INDEX_VERSION
+		fi &&
 		git add a 2>&1 &&
-		echo 4 >expect &&
+		echo $EXPECTED_OUTPUT_VERSION >expect &&
 		test-tool index-version <.git/index >actual &&
 		test_cmp expect actual
 	)
+}
+
+test_expect_success 'index version config precedence' '
+	test_index_version 2 false 4 4 &&
+	test_index_version 2 true 0 2 &&
+	test_index_version 0 true 0 4 &&
+	test_index_version 0 true 2 2
 '
 
 test_done
-- 
gitgitgadget

[PATCH v4 6/6] repo-settings: create feature.experimental setting

[Derrick Stolee via GitGitGadget]

From: Derrick Stolee <dstolee@microsoft.com>

The 'feature.experimental' setting includes config options that are
not committed to become defaults, but could use additional testing.

Update the following config settings to take new defaults, and to
use the repo_settings struct if not already using them:

* 'pack.useSparse=true'

* 'fetch.negotiationAlgorithm=skipping'

In the case of fetch.negotiationAlgorithm, the existing logic
would load the config option only when about to use the setting,
so had a die() statement on an unknown string value. This is
removed as now the config is parsed under prepare_repo_settings().
In general, this die() is probably misplaced and not valuable.
A test was removed that checked this die() statement executed.

Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
---
 Documentation/config/feature.txt     | 14 ++++++++++++++
 Documentation/config/fetch.txt       |  3 ++-
 Documentation/config/pack.txt        |  3 ++-
 fetch-negotiator.c                   | 25 +++++++++++++------------
 fetch-negotiator.h                   |  5 +++--
 fetch-pack.c                         | 11 +++++------
 repo-settings.c                      | 13 +++++++++++++
 repository.h                         |  8 ++++++++
 t/t5552-skipping-fetch-negotiator.sh | 23 -----------------------
 9 files changed, 60 insertions(+), 45 deletions(-)

diff --git a/Documentation/config/feature.txt b/Documentation/config/feature.txt
index 8ea198a642..545522f306 100644
--- a/Documentation/config/feature.txt
+++ b/Documentation/config/feature.txt
@@ -4,6 +4,20 @@ feature.*::
 	developer community as recommended defaults and are subject to change.
 	In particular, new config options may be added with different defaults.
 
+feature.experimental::
+	Enable config options that are new to Git, and are being considered for
+	future defaults. Config settings included here may be added or removed
+	with each release, including minor version updates. These settings may
+	have unintended interactions since they are so new. Please enable this
+	setting if you are interested in providing feedback on experimental
+	features. The new default values are:
++
+* `pack.useSparse=true` uses a new algorithm when constructing a pack-file
+which can improve `git push` performance in repos with many files.
++
+* `fetch.negotiationAlgorithm=skipping` may improve fetch negotiation times by
+skipping more commits at a time, reducing the number of round trips.
+
 feature.manyFiles::
 	Enable config options that optimize for repos with many files in the
 	working directory. With many files, commands such as `git status` and
diff --git a/Documentation/config/fetch.txt b/Documentation/config/fetch.txt
index ba890b5884..d402110638 100644
--- a/Documentation/config/fetch.txt
+++ b/Documentation/config/fetch.txt
@@ -59,7 +59,8 @@ fetch.negotiationAlgorithm::
 	effort to converge faster, but may result in a larger-than-necessary
 	packfile; The default is "default" which instructs Git to use the default algorithm
 	that never skips commits (unless the server has acknowledged it or one
-	of its descendants).
+	of its descendants). If `feature.experimental` is enabled, then this
+	setting defaults to "skipping".
 	Unknown values will cause 'git fetch' to error out.
 +
 See also the `--negotiation-tip` option for linkgit:git-fetch[1].
diff --git a/Documentation/config/pack.txt b/Documentation/config/pack.txt
index 9cdcfa7324..1d66f0c992 100644
--- a/Documentation/config/pack.txt
+++ b/Documentation/config/pack.txt
@@ -112,7 +112,8 @@ pack.useSparse::
 	objects. This can have significant performance benefits when
 	computing a pack to send a small change. However, it is possible
 	that extra objects are added to the pack-file if the included
-	commits contain certain types of direct renames.
+	commits contain certain types of direct renames. Default is `false`
+	unless `feature.experimental` is enabled.
 
 pack.writeBitmaps (deprecated)::
 	This is a deprecated synonym for `repack.writeBitmaps`.
diff --git a/fetch-negotiator.c b/fetch-negotiator.c
index d6d685cba0..0a1357dc9d 100644
--- a/fetch-negotiator.c
+++ b/fetch-negotiator.c
@@ -2,19 +2,20 @@
 #include "fetch-negotiator.h"
 #include "negotiator/default.h"
 #include "negotiator/skipping.h"
+#include "repository.h"
 
-void fetch_negotiator_init(struct fetch_negotiator *negotiator,
-			   const char *algorithm)
+void fetch_negotiator_init(struct repository *r,
+			   struct fetch_negotiator *negotiator)
 {
-	if (algorithm) {
-		if (!strcmp(algorithm, "skipping")) {
-			skipping_negotiator_init(negotiator);
-			return;
-		} else if (!strcmp(algorithm, "default")) {
-			/* Fall through to default initialization */
-		} else {
-			die("unknown fetch negotiation algorithm '%s'", algorithm);
-		}
+	prepare_repo_settings(r);
+	switch(r->settings.fetch_negotiation_algorithm) {
+	case FETCH_NEGOTIATION_SKIPPING:
+		skipping_negotiator_init(negotiator);
+		return;
+
+	case FETCH_NEGOTIATION_DEFAULT:
+	default:
+		default_negotiator_init(negotiator);
+		return;
 	}
-	default_negotiator_init(negotiator);
 }
diff --git a/fetch-negotiator.h b/fetch-negotiator.h
index 9e3967ce66..ea78868504 100644
--- a/fetch-negotiator.h
+++ b/fetch-negotiator.h
@@ -2,6 +2,7 @@
 #define FETCH_NEGOTIATOR_H
 
 struct commit;
+struct repository;
 
 /*
  * An object that supplies the information needed to negotiate the contents of
@@ -52,7 +53,7 @@ struct fetch_negotiator {
 	void *data;
 };
 
-void fetch_negotiator_init(struct fetch_negotiator *negotiator,
-			   const char *algorithm);
+void fetch_negotiator_init(struct repository *r,
+			   struct fetch_negotiator *negotiator);
 
 #endif
diff --git a/fetch-pack.c b/fetch-pack.c
index 65be043f2a..d81f47c07b 100644
--- a/fetch-pack.c
+++ b/fetch-pack.c
@@ -36,7 +36,6 @@ static int agent_supported;
 static int server_supports_filtering;
 static struct lock_file shallow_lock;
 static const char *alternate_shallow_file;
-static char *negotiation_algorithm;
 static struct strbuf fsck_msg_types = STRBUF_INIT;
 
 /* Remember to update object flag allocation in object.h */
@@ -892,12 +891,13 @@ static struct ref *do_fetch_pack(struct fetch_pack_args *args,
 				 struct shallow_info *si,
 				 char **pack_lockfile)
 {
+	struct repository *r = the_repository;
 	struct ref *ref = copy_ref_list(orig_ref);
 	struct object_id oid;
 	const char *agent_feature;
 	int agent_len;
 	struct fetch_negotiator negotiator;
-	fetch_negotiator_init(&negotiator, negotiation_algorithm);
+	fetch_negotiator_init(r, &negotiator);
 
 	sort_ref_list(&ref, ref_compare_name);
 	QSORT(sought, nr_sought, cmp_ref_by_name);
@@ -911,7 +911,7 @@ static struct ref *do_fetch_pack(struct fetch_pack_args *args,
 
 	if (server_supports("shallow"))
 		print_verbose(args, _("Server supports %s"), "shallow");
-	else if (args->depth > 0 || is_repository_shallow(the_repository))
+	else if (args->depth > 0 || is_repository_shallow(r))
 		die(_("Server does not support shallow clients"));
 	if (args->depth > 0 || args->deepen_since || args->deepen_not)
 		args->deepen = 1;
@@ -1379,6 +1379,7 @@ static struct ref *do_fetch_pack_v2(struct fetch_pack_args *args,
 				    struct shallow_info *si,
 				    char **pack_lockfile)
 {
+	struct repository *r = the_repository;
 	struct ref *ref = copy_ref_list(orig_ref);
 	enum fetch_state state = FETCH_CHECK_LOCAL;
 	struct oidset common = OIDSET_INIT;
@@ -1386,7 +1387,7 @@ static struct ref *do_fetch_pack_v2(struct fetch_pack_args *args,
 	int in_vain = 0;
 	int haves_to_send = INITIAL_FLUSH;
 	struct fetch_negotiator negotiator;
-	fetch_negotiator_init(&negotiator, negotiation_algorithm);
+	fetch_negotiator_init(r, &negotiator);
 	packet_reader_init(&reader, fd[0], NULL, 0,
 			   PACKET_READ_CHOMP_NEWLINE |
 			   PACKET_READ_DIE_ON_ERR_PACKET);
@@ -1505,8 +1506,6 @@ static void fetch_pack_config(void)
 	git_config_get_bool("repack.usedeltabaseoffset", &prefer_ofs_delta);
 	git_config_get_bool("fetch.fsckobjects", &fetch_fsck_objects);
 	git_config_get_bool("transfer.fsckobjects", &transfer_fsck_objects);
-	git_config_get_string("fetch.negotiationalgorithm",
-			      &negotiation_algorithm);
 
 	git_config(fetch_pack_config_cb, NULL);
 }
diff --git a/repo-settings.c b/repo-settings.c
index d5bf9069f4..3779b85c17 100644
--- a/repo-settings.c
+++ b/repo-settings.c
@@ -36,16 +36,29 @@ void prepare_repo_settings(struct repository *r)
 		free(strval);
 	}
 
+	if (!repo_config_get_string(r, "fetch.negotiationalgorithm", &strval)) {
+		if (!strcasecmp(strval, "skipping"))
+			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_SKIPPING;
+		else
+			r->settings.fetch_negotiation_algorithm = FETCH_NEGOTIATION_DEFAULT;
+	}
+
 	if (!repo_config_get_bool(r, "pack.usesparse", &value))
 		r->settings.pack_use_sparse = value;
 	if (!repo_config_get_bool(r, "feature.manyfiles", &value) && value) {
 		UPDATE_DEFAULT_BOOL(r->settings.index_version, 4);
 		UPDATE_DEFAULT_BOOL(r->settings.core_untracked_cache, UNTRACKED_CACHE_WRITE);
 	}
+	if (!repo_config_get_bool(r, "feature.experimental", &value) && value) {
+		UPDATE_DEFAULT_BOOL(r->settings.pack_use_sparse, 1);
+		UPDATE_DEFAULT_BOOL(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_SKIPPING);
+	}
 
 	/* Hack for test programs like test-dump-untracked-cache */
 	if (ignore_untracked_cache_config)
 		r->settings.core_untracked_cache = UNTRACKED_CACHE_KEEP;
 	else
 		UPDATE_DEFAULT_BOOL(r->settings.core_untracked_cache, UNTRACKED_CACHE_KEEP);
+
+	UPDATE_DEFAULT_BOOL(r->settings.fetch_negotiation_algorithm, FETCH_NEGOTIATION_DEFAULT);
 }
diff --git a/repository.h b/repository.h
index cf7ff0778c..4da275e73f 100644
--- a/repository.h
+++ b/repository.h
@@ -18,6 +18,13 @@ enum untracked_cache_setting {
 	UNTRACKED_CACHE_WRITE = 2
 };
 
+enum fetch_negotiation_setting {
+	FETCH_NEGOTIATION_UNSET = -1,
+	FETCH_NEGOTIATION_NONE = 0,
+	FETCH_NEGOTIATION_DEFAULT = 1,
+	FETCH_NEGOTIATION_SKIPPING = 2,
+};
+
 struct repo_settings {
 	int initialized;
 
@@ -28,6 +35,7 @@ struct repo_settings {
 	enum untracked_cache_setting core_untracked_cache;
 
 	int pack_use_sparse;
+	enum fetch_negotiation_setting fetch_negotiation_algorithm;
 };
 
 struct repository {
diff --git a/t/t5552-skipping-fetch-negotiator.sh b/t/t5552-skipping-fetch-negotiator.sh
index 8a14be51a1..f70cbcc9ca 100755
--- a/t/t5552-skipping-fetch-negotiator.sh
+++ b/t/t5552-skipping-fetch-negotiator.sh
@@ -60,29 +60,6 @@ test_expect_success 'commits with no parents are sent regardless of skip distanc
 	have_not_sent c6 c4 c3
 '
 
-test_expect_success 'unknown fetch.negotiationAlgorithm values error out' '
-	rm -rf server client trace &&
-	git init server &&
-	test_commit -C server to_fetch &&
-
-	git init client &&
-	test_commit -C client on_client &&
-	git -C client checkout on_client &&
-
-	test_config -C client fetch.negotiationAlgorithm invalid &&
-	test_must_fail git -C client fetch "$(pwd)/server" 2>err &&
-	test_i18ngrep "unknown fetch negotiation algorithm" err &&
-
-	# Explicit "default" value
-	test_config -C client fetch.negotiationAlgorithm default &&
-	git -C client -c fetch.negotiationAlgorithm=default fetch "$(pwd)/server" &&
-
-	# Implementation detail: If there is nothing to fetch, we will not error out
-	test_config -C client fetch.negotiationAlgorithm invalid &&
-	git -C client fetch "$(pwd)/server" 2>err &&
-	test_i18ngrep ! "unknown fetch negotiation algorithm" err
-'
-
 test_expect_success 'when two skips collide, favor the larger one' '
 	rm -rf server client trace &&
 	git init server &&
-- 
gitgitgadget

Re: [PATCH v4 0/6] Create 'feature.*' config area and some centralized config parsing

[Junio C Hamano]

"Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:

>  * Update docs and 0 constant in feature.experimental around
>    merge.directoryRenames.

I somehow thought that the concensus was that *.experimental is for
things that should become the future default, and this one is not
something we would want to make the default.

Re: [PATCH v4 0/6] Create 'feature.*' config area and some centralized config parsing

[Junio C Hamano]

Junio C Hamano <gitster@pobox.com> writes:

> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>
>>  * Update docs and 0 constant in feature.experimental around
>>    merge.directoryRenames.
>
> I somehow thought that the concensus was that *.experimental is for
> things that should become the future default, and this one is not
> something we would want to make the default.

Ah, I see merge.directoryRenames is gone.  Sorry for a confusion.

Re: [PATCH v4 0/6] Create 'feature.*' config area and some centralized config parsing

[Derrick Stolee]

On 8/13/2019 5:08 PM, Junio C Hamano wrote:
> Junio C Hamano <gitster@pobox.com> writes:
> 
>> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>
>>>  * Update docs and 0 constant in feature.experimental around
>>>    merge.directoryRenames.
>>
>> I somehow thought that the concensus was that *.experimental is for
>> things that should become the future default, and this one is not
>> something we would want to make the default.
> 
> Ah, I see merge.directoryRenames is gone.  Sorry for a confusion.

Confusion is my fault. This was a terrible way to say, "I cleaned up the
adjustments to merge.directoryRenames AND removed it from features.experimental"

Thanks,
-Stolee

Re: [PATCH v4 0/6] Create 'feature.*' config area and some centralized config parsing

[Derrick Stolee]

On 8/14/2019 6:32 AM, Derrick Stolee wrote:
> On 8/13/2019 5:08 PM, Junio C Hamano wrote:
>> Junio C Hamano <gitster@pobox.com> writes:
>>
>>> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>>
>>>>  * Update docs and 0 constant in feature.experimental around
>>>>    merge.directoryRenames.
>>>
>>> I somehow thought that the concensus was that *.experimental is for
>>> things that should become the future default, and this one is not
>>> something we would want to make the default.
>>
>> Ah, I see merge.directoryRenames is gone.  Sorry for a confusion.
> 
> Confusion is my fault. This was a terrible way to say, "I cleaned up the
> adjustments to merge.directoryRenames AND removed it from features.experimental"

It's too early for me to say things... let's try this again.

1. merge.directoryRenames is removed from this series.

2. The commit where I did the cleanup according to Elijahs' recommendation is 
   available as a tag, as mentioned in [1].

[1] https://public-inbox.org/git/1a2bf8e1-5cc1-c626-9dfc-dd6dc58fa0ad@gmail.com/

Re: [PATCH v4 3/6] commit-graph: turn on commit-graph by default

[Junio C Hamano]

"Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:

> diff --git a/repo-settings.c b/repo-settings.c
> index 309577f6bc..d00b675687 100644
> --- a/repo-settings.c
> +++ b/repo-settings.c
> @@ -2,6 +2,8 @@
>  #include "config.h"
>  #include "repository.h"
>  
> +#define UPDATE_DEFAULT_BOOL(s,v) do { if (s == -1) { s = v; } } while(0)
> +
>  void prepare_repo_settings(struct repository *r)
>  {
>  	int value;
> @@ -16,6 +18,8 @@ void prepare_repo_settings(struct repository *r)
>  		r->settings.core_commit_graph = value;
>  	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
>  		r->settings.gc_write_commit_graph = value;
> +	UPDATE_DEFAULT_BOOL(r->settings.core_commit_graph, 1);
> +	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);


This is a "review comment" that is more than 2 years late X-<, but I
noticed that this is used to muck with a structure that was
initialized by filling it with \0377 bytes.

+           /* Defaults */
+           memset(&r->settings, -1, sizeof(r->settings));

but the structure is is full of "int" and "enum", so apparently this
works only on 2's complement architecture.

        struct repo_settings {
                int initialized;

                int core_commit_graph;
                int commit_graph_read_changed_paths;
                int gc_write_commit_graph;
                int fetch_write_commit_graph;

                int index_version;
                enum untracked_cache_setting core_untracked_cache;

                int pack_use_sparse;
                enum fetch_negotiation_setting fetch_negotiation_algorithm;

                int core_multi_pack_index;

                unsigned command_requires_full_index:1,
                         sparse_index:1;
        };

I see that the earliest iteration of this series [*1*] set the
default explicitly using assignments of the correct types, like
this:


+void prepare_repo_settings(struct repository *r)
+{
+       if (r->settings)
+          return;
+
+       r->settings = xmalloc(sizeof(*r->settings));
+
+       /* Defaults */
+       r->settings->core_commit_graph = -1;
+       r->settings->gc_write_commit_graph = -1;
+       r->settings->pack_use_sparse = -1;
+       r->settings->index_version = -1;
+ ...

which I think should be a reasonable starting point to fix the
current code.

Another thing I noticed is that while it may have been only for
setting the default value for a boolean variable initially, other
changes abuse the macro to set an arbitrary integer values to
integer members of the structure, e.g. c6cc4c5a (repo-settings:
create feature.manyFiles setting, 2019-08-13) sets 4 to the
index_version (naturally, the choice between 0 and 1 does not make
much sense for the member), and ad0fb659 (repo-settings: parse
core.untrackedCache, 2019-08-13) stuffs UNTRACKED_CACHE_* enum to
core_untracked_cache.  The UPDATE_DEFAULT_BOOL() macro should be
renamed to UPDATE_DEFAULT_INT() at least, I would think, to save
readers from confusion.

Thanks.


[Reference]

*1* https://lore.kernel.org/git/72f652b89c71526cc423e7812de66f41a079f181.1563818059.git.gitgitgadget@gmail.com/

Re: [PATCH v4 3/6] commit-graph: turn on commit-graph by default

[Ævar Arnfjörð Bjarmason]

On Sun, Sep 19 2021, Junio C Hamano wrote:

> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>
>> diff --git a/repo-settings.c b/repo-settings.c
>> index 309577f6bc..d00b675687 100644
>> --- a/repo-settings.c
>> +++ b/repo-settings.c
>> @@ -2,6 +2,8 @@
>>  #include "config.h"
>>  #include "repository.h"
>>  
>> +#define UPDATE_DEFAULT_BOOL(s,v) do { if (s == -1) { s = v; } } while(0)
>> +
>>  void prepare_repo_settings(struct repository *r)
>>  {
>>  	int value;
>> @@ -16,6 +18,8 @@ void prepare_repo_settings(struct repository *r)
>>  		r->settings.core_commit_graph = value;
>>  	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
>>  		r->settings.gc_write_commit_graph = value;
>> +	UPDATE_DEFAULT_BOOL(r->settings.core_commit_graph, 1);
>> +	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);
>
>
> This is a "review comment" that is more than 2 years late X-<, but I
> noticed that this is used to muck with a structure that was
> initialized by filling it with \0377 bytes.
>
> +           /* Defaults */
> +           memset(&r->settings, -1, sizeof(r->settings));
>
> but the structure is is full of "int" and "enum", so apparently this
> works only on 2's complement architecture.
>
>         struct repo_settings {
>                 int initialized;
>
>                 int core_commit_graph;
>                 int commit_graph_read_changed_paths;
>                 int gc_write_commit_graph;
>                 int fetch_write_commit_graph;
>
>                 int index_version;
>                 enum untracked_cache_setting core_untracked_cache;
>
>                 int pack_use_sparse;
>                 enum fetch_negotiation_setting fetch_negotiation_algorithm;
>
>                 int core_multi_pack_index;
>
>                 unsigned command_requires_full_index:1,
>                          sparse_index:1;
>         };
>
> I see that the earliest iteration of this series [*1*] set the
> default explicitly using assignments of the correct types, like
> this:
>
>
> +void prepare_repo_settings(struct repository *r)
> +{
> +       if (r->settings)
> +          return;
> +
> +       r->settings = xmalloc(sizeof(*r->settings));
> +
> +       /* Defaults */
> +       r->settings->core_commit_graph = -1;
> +       r->settings->gc_write_commit_graph = -1;
> +       r->settings->pack_use_sparse = -1;
> +       r->settings->index_version = -1;
> + ...
>
> which I think should be a reasonable starting point to fix the
> current code.
>
> Another thing I noticed is that while it may have been only for
> setting the default value for a boolean variable initially, other
> changes abuse the macro to set an arbitrary integer values to
> integer members of the structure, e.g. c6cc4c5a (repo-settings:
> create feature.manyFiles setting, 2019-08-13) sets 4 to the
> index_version (naturally, the choice between 0 and 1 does not make
> much sense for the member), and ad0fb659 (repo-settings: parse
> core.untrackedCache, 2019-08-13) stuffs UNTRACKED_CACHE_* enum to
> core_untracked_cache.  The UPDATE_DEFAULT_BOOL() macro should be
> renamed to UPDATE_DEFAULT_INT() at least, I would think, to save
> readers from confusion.

Yes this is all a bit weird and/or broken, but I'm a bit perplexed at
this reply to a 2+ year old E-Mail given my outstanding series to fix
all these issues you've noted here[1] posted in the last few days, and
you having read (at least part of) it[1].

But then again, the last patch you left a comment on was 3/5. It's 4/5
that fixes all the issues you note above[2] :)

The macro is gone, so is the memset to -1 and other weird emergent
behavior. We can rely on repo_init() having zero'd the structure for us,
and we just proceed to set sensible defaults in a way that doesn't stomp
over the types in the struct.

1. https://lore.kernel.org/git/cover-v3-0.5-00000000000-20210919T084703Z-avarab@gmail.com/
2. https://lore.kernel.org/git/patch-v3-4.5-28286a61162-20210919T084703Z-avarab@gmail.com/

> [Reference]
>
> *1* https://lore.kernel.org/git/72f652b89c71526cc423e7812de66f41a079f181.1563818059.git.gitgitgadget@gmail.com/

Re: [PATCH v4 3/6] commit-graph: turn on commit-graph by default

[brian m. carlson]

[-- Attachment #1: Type: text/plain, Size: 2034 bytes --]

On 2021-09-20 at 00:42:57, Junio C Hamano wrote:
> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
> 
> > diff --git a/repo-settings.c b/repo-settings.c
> > index 309577f6bc..d00b675687 100644
> > --- a/repo-settings.c
> > +++ b/repo-settings.c
> > @@ -2,6 +2,8 @@
> >  #include "config.h"
> >  #include "repository.h"
> >  
> > +#define UPDATE_DEFAULT_BOOL(s,v) do { if (s == -1) { s = v; } } while(0)
> > +
> >  void prepare_repo_settings(struct repository *r)
> >  {
> >  	int value;
> > @@ -16,6 +18,8 @@ void prepare_repo_settings(struct repository *r)
> >  		r->settings.core_commit_graph = value;
> >  	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
> >  		r->settings.gc_write_commit_graph = value;
> > +	UPDATE_DEFAULT_BOOL(r->settings.core_commit_graph, 1);
> > +	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);
> 
> 
> This is a "review comment" that is more than 2 years late X-<, but I
> noticed that this is used to muck with a structure that was
> initialized by filling it with \0377 bytes.
> 
> +           /* Defaults */
> +           memset(&r->settings, -1, sizeof(r->settings));
> 
> but the structure is is full of "int" and "enum", so apparently this
> works only on 2's complement architecture.

This statement is true, but are there systems capable of running Git
which don't use two's complement?  Rust requires two's complement signed
integers, and there's a proposal[0] to the C++ working group to only
support two's complement because "[t]o the author’s knowledge no modern
machine uses both C++ and a signed integer representation other than
two’s complement".  That proposal goes on to note that none of MSVC,
GCC, or LLVM support other options.

Personally I am not aware of any modern processor which provides signed
integer types using other than two's complement.

[0] http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0907r4.html
-- 
brian m. carlson (he/him or they/them)
Toronto, Ontario, CA

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 262 bytes --]

Re: [PATCH v4 3/6] commit-graph: turn on commit-graph by default

[Phillip Wood]

On 20/09/2021 02:25, brian m. carlson wrote:
> On 2021-09-20 at 00:42:57, Junio C Hamano wrote:
>> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>
>>> diff --git a/repo-settings.c b/repo-settings.c
>>> index 309577f6bc..d00b675687 100644
>>> --- a/repo-settings.c
>>> +++ b/repo-settings.c
>>> @@ -2,6 +2,8 @@
>>>   #include "config.h"
>>>   #include "repository.h"
>>>   
>>> +#define UPDATE_DEFAULT_BOOL(s,v) do { if (s == -1) { s = v; } } while(0)
>>> +
>>>   void prepare_repo_settings(struct repository *r)
>>>   {
>>>   	int value;
>>> @@ -16,6 +18,8 @@ void prepare_repo_settings(struct repository *r)
>>>   		r->settings.core_commit_graph = value;
>>>   	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
>>>   		r->settings.gc_write_commit_graph = value;
>>> +	UPDATE_DEFAULT_BOOL(r->settings.core_commit_graph, 1);
>>> +	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);
>>
>>
>> This is a "review comment" that is more than 2 years late X-<, but I
>> noticed that this is used to muck with a structure that was
>> initialized by filling it with \0377 bytes.
>>
>> +           /* Defaults */
>> +           memset(&r->settings, -1, sizeof(r->settings));
>>
>> but the structure is is full of "int" and "enum", so apparently this
>> works only on 2's complement architecture.
> 
> This statement is true, but are there systems capable of running Git
> which don't use two's complement?  Rust requires two's complement signed
> integers, and there's a proposal[0] to the C++ working group to only
> support two's complement because "[t]o the author’s knowledge no modern
> machine uses both C++ and a signed integer representation other than
> two’s complement".  That proposal goes on to note that none of MSVC,
> GCC, or LLVM support other options.

A similar proposal [1] is included in the draft of the next C standard 
[2]. As integer representation is implementation defined I believe this 
code has well defined behavior on 2's complement implementations. If an 
enum has no negative members then the compiler may choose an unsigned 
representation but even then the comparison to -1 is well defined. In 
this case I'm pretty sure the enums all have -1 as a member so are 
signed. Using memset() to initialize the struct eases future maintenance 
when members are added or removed and seems to me to be a sensible 
design choice.

Best Wishes

Phillip

[1] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2412.pdf
[2] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2596.pdf

> Personally I am not aware of any modern processor which provides signed
> integer types using other than two's complement.
> 
> [0] http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0907r4.html
> 

Re: [PATCH v4 3/6] commit-graph: turn on commit-graph by default

[Ævar Arnfjörð Bjarmason]

On Mon, Sep 20 2021, Phillip Wood wrote:

> On 20/09/2021 02:25, brian m. carlson wrote:
>> On 2021-09-20 at 00:42:57, Junio C Hamano wrote:
>>> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>>
>>>> diff --git a/repo-settings.c b/repo-settings.c
>>>> index 309577f6bc..d00b675687 100644
>>>> --- a/repo-settings.c
>>>> +++ b/repo-settings.c
>>>> @@ -2,6 +2,8 @@
>>>>   #include "config.h"
>>>>   #include "repository.h"
>>>>   +#define UPDATE_DEFAULT_BOOL(s,v) do { if (s == -1) { s = v; } }
>>>> while(0)
>>>> +
>>>>   void prepare_repo_settings(struct repository *r)
>>>>   {
>>>>   	int value;
>>>> @@ -16,6 +18,8 @@ void prepare_repo_settings(struct repository *r)
>>>>   		r->settings.core_commit_graph = value;
>>>>   	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
>>>>   		r->settings.gc_write_commit_graph = value;
>>>> +	UPDATE_DEFAULT_BOOL(r->settings.core_commit_graph, 1);
>>>> +	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);
>>>
>>>
>>> This is a "review comment" that is more than 2 years late X-<, but I
>>> noticed that this is used to muck with a structure that was
>>> initialized by filling it with \0377 bytes.
>>>
>>> +           /* Defaults */
>>> +           memset(&r->settings, -1, sizeof(r->settings));
>>>
>>> but the structure is is full of "int" and "enum", so apparently this
>>> works only on 2's complement architecture.
>> This statement is true, but are there systems capable of running Git
>> which don't use two's complement?  Rust requires two's complement signed
>> integers, and there's a proposal[0] to the C++ working group to only
>> support two's complement because "[t]o the author’s knowledge no modern
>> machine uses both C++ and a signed integer representation other than
>> two’s complement".  That proposal goes on to note that none of MSVC,
>> GCC, or LLVM support other options.
>
> A similar proposal [1] is included in the draft of the next C standard
> [2]. As integer representation is implementation defined I believe
> this code has well defined behavior on 2's complement
> implementations. If an enum has no negative members then the compiler
> may choose an unsigned representation but even then the comparison to
> -1 is well defined.

That's informative, thanks.

> In this case I'm pretty sure the enums all have -1
> as a member so are signed. Using memset() to initialize the struct
> eases future maintenance when members are added or removed and seems
> to me to be a sensible design choice.

It's really not sensible at all in this particular case, as I think my
[1] which gets rid of the pattern convincingly argues.

I.e. the only reason it had a memset() of -1 after we'd already memset
it to 0 was because the function was tripping over itself and setting
defaults in the wrong order for no good reason.

I.e. it was doing things like (pseudocode);

    memset(&data, -1, ...)
    if_config_is_true_set("x.y", data.x_y);
    if (data.x_y == -1)
        data.x_y = x_y_default();

When we can instead just do:

    data.x_y = x_y_default();
    set_if_cfg_key_exists("x.y", &data.x_y);

Which is how we e.g. handle options parsing, we have hardcoded defaults,
then read defaults from config, then set options, in that order.

We don't set options, then check if each value is still -1, if so read
config etc. Just read them in priority order, doing it any other way is
just make-work for something that's the equivalent of a simple
short-circuit || operation.

Anyway, there are other cases where we need to read something and
distinguish e.g. false/true from "unset", and there a -1,0,1 tri-state
serves us well.

But even in those cases what repo-settings.c was doing of memsetting the
entire struct to -1 (2's compliment aside) just makes for needlessly
hard to read code.

If we've got some members that need -1 defaults we should instead have
that in an *_INIT macro or equivalent. The pre-[1] repo-settings.c also
has code like this pseudocode:

    data.a_b = -1; /* default for a bi-state, not tri-state variable */
    set_if_cfg_key_exists("a.b", &data.a_b);
    if (data.a_b == -1)
        data.a_b = 1; /* on by default */

Which, urm, you can just do as:

    data.a_b = 1; /* on by default */
    set_if_cfg_key_exists("a.b", &data.a_b);

I.e. the setup for things that never wanted or cared about being set to
-1 was complicated by them needing to un-set themselves from a -1
default they never wanted.

Thus the anti-pattern, yes set defaults for some members to -1, but not
the entire struct. The only value we should memset a whole bag-of-stuff
config struct to is 0, as that's the only sensible default & plays well
with other C semantics.

1. https://lore.kernel.org/git/patch-v3-4.5-28286a61162-20210919T084703Z-avarab@gmail.com/

>
> Best Wishes
>
> Phillip
>
> [1] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2412.pdf
> [2] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2596.pdf
>
>> Personally I am not aware of any modern processor which provides signed
>> integer types using other than two's complement.
>> [0]
>> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0907r4.html
>> 

Re: [PATCH v4 3/6] commit-graph: turn on commit-graph by default

[Phillip Wood]

On 20/09/2021 14:30, Ævar Arnfjörð Bjarmason wrote:
> 
> On Mon, Sep 20 2021, Phillip Wood wrote:
> 
>> On 20/09/2021 02:25, brian m. carlson wrote:
>>> On 2021-09-20 at 00:42:57, Junio C Hamano wrote:
>>>> "Derrick Stolee via GitGitGadget" <gitgitgadget@gmail.com> writes:
>>>>
>>>>> diff --git a/repo-settings.c b/repo-settings.c
>>>>> index 309577f6bc..d00b675687 100644
>>>>> --- a/repo-settings.c
>>>>> +++ b/repo-settings.c
>>>>> @@ -2,6 +2,8 @@
>>>>>    #include "config.h"
>>>>>    #include "repository.h"
>>>>>    +#define UPDATE_DEFAULT_BOOL(s,v) do { if (s == -1) { s = v; } }
>>>>> while(0)
>>>>> +
>>>>>    void prepare_repo_settings(struct repository *r)
>>>>>    {
>>>>>    	int value;
>>>>> @@ -16,6 +18,8 @@ void prepare_repo_settings(struct repository *r)
>>>>>    		r->settings.core_commit_graph = value;
>>>>>    	if (!repo_config_get_bool(r, "gc.writecommitgraph", &value))
>>>>>    		r->settings.gc_write_commit_graph = value;
>>>>> +	UPDATE_DEFAULT_BOOL(r->settings.core_commit_graph, 1);
>>>>> +	UPDATE_DEFAULT_BOOL(r->settings.gc_write_commit_graph, 1);
>>>>
>>>>
>>>> This is a "review comment" that is more than 2 years late X-<, but I
>>>> noticed that this is used to muck with a structure that was
>>>> initialized by filling it with \0377 bytes.
>>>>
>>>> +           /* Defaults */
>>>> +           memset(&r->settings, -1, sizeof(r->settings));
>>>>
>>>> but the structure is is full of "int" and "enum", so apparently this
>>>> works only on 2's complement architecture.
>>> This statement is true, but are there systems capable of running Git
>>> which don't use two's complement?  Rust requires two's complement signed
>>> integers, and there's a proposal[0] to the C++ working group to only
>>> support two's complement because "[t]o the author’s knowledge no modern
>>> machine uses both C++ and a signed integer representation other than
>>> two’s complement".  That proposal goes on to note that none of MSVC,
>>> GCC, or LLVM support other options.
>>
>> A similar proposal [1] is included in the draft of the next C standard
>> [2]. As integer representation is implementation defined I believe
>> this code has well defined behavior on 2's complement
>> implementations. If an enum has no negative members then the compiler
>> may choose an unsigned representation but even then the comparison to
>> -1 is well defined.
> 
> That's informative, thanks.
> 
>> In this case I'm pretty sure the enums all have -1
>> as a member so are signed. Using memset() to initialize the struct
>> eases future maintenance when members are added or removed and seems
>> to me to be a sensible design choice.
> 
> It's really not sensible at all in this particular case, as I think my
> [1] which gets rid of the pattern convincingly argues.

I meant that it was a sensible way to initialize all the struct members 
to -1, I did not mean to comment either way on whether such an 
initialization was sensible. If we can just store the default and then 
update from the user's config that sounds sensible.

Best Wishes

Phillip

> I.e. the only reason it had a memset() of -1 after we'd already memset
> it to 0 was because the function was tripping over itself and setting
> defaults in the wrong order for no good reason.
> 
> I.e. it was doing things like (pseudocode);
> 
>      memset(&data, -1, ...)
>      if_config_is_true_set("x.y", data.x_y);
>      if (data.x_y == -1)
>          data.x_y = x_y_default();
> 
> When we can instead just do:
> 
>      data.x_y = x_y_default();
>      set_if_cfg_key_exists("x.y", &data.x_y);
> 
> Which is how we e.g. handle options parsing, we have hardcoded defaults,
> then read defaults from config, then set options, in that order.
> 
> We don't set options, then check if each value is still -1, if so read
> config etc. Just read them in priority order, doing it any other way is
> just make-work for something that's the equivalent of a simple
> short-circuit || operation.
> 
> Anyway, there are other cases where we need to read something and
> distinguish e.g. false/true from "unset", and there a -1,0,1 tri-state
> serves us well.
> 
> But even in those cases what repo-settings.c was doing of memsetting the
> entire struct to -1 (2's compliment aside) just makes for needlessly
> hard to read code.
> 
> If we've got some members that need -1 defaults we should instead have
> that in an *_INIT macro or equivalent. The pre-[1] repo-settings.c also
> has code like this pseudocode:
> 
>      data.a_b = -1; /* default for a bi-state, not tri-state variable */
>      set_if_cfg_key_exists("a.b", &data.a_b);
>      if (data.a_b == -1)
>          data.a_b = 1; /* on by default */
> 
> Which, urm, you can just do as:
> 
>      data.a_b = 1; /* on by default */
>      set_if_cfg_key_exists("a.b", &data.a_b);
> 
> I.e. the setup for things that never wanted or cared about being set to
> -1 was complicated by them needing to un-set themselves from a -1
> default they never wanted.
> 
> Thus the anti-pattern, yes set defaults for some members to -1, but not
> the entire struct. The only value we should memset a whole bag-of-stuff
> config struct to is 0, as that's the only sensible default & plays well
> with other C semantics.
> 
> 1. https://lore.kernel.org/git/patch-v3-4.5-28286a61162-20210919T084703Z-avarab@gmail.com/
> 
>>
>> Best Wishes
>>
>> Phillip
>>
>> [1] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2412.pdf
>> [2] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n2596.pdf
>>
>>> Personally I am not aware of any modern processor which provides signed
>>> integer types using other than two's complement.
>>> [0]
>>> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2018/p0907r4.html
>>>
> 

Re: [PATCH v4 3/6] commit-graph: turn on commit-graph by default

[Jeff King]

On Mon, Sep 20, 2021 at 03:30:38PM +0200, Ævar Arnfjörð Bjarmason wrote:

> Thus the anti-pattern, yes set defaults for some members to -1, but not
> the entire struct. The only value we should memset a whole bag-of-stuff
> config struct to is 0, as that's the only sensible default & plays well
> with other C semantics.

FWIW, I agree. I had to scratch my head for a moment at why a memset of
"-1" would work at all on multi-byte types. I think it's better avoided
in the name of readability and obviousness, not to mention the trap it
leaves for items that don't sensibly initialize with it (like, say,
pointers).

As an aside, memset to 0 is _also_ undefined for pointers, but we long
ago decided not to care, as no real-world systems have a problem with
this.

-Peff