Skip to contentSkip to conversation

environment: clean up repository config handling

PATCH Superseded Tian Yuchen 3 participants 29 messages Aug 5 – Nov 3

lore
Share & tools
lore
#0 lore raw

cover environment: clean up repository config handling

Hi all,

This series contains several cleanup patches for repository configuration handling.

No functional changes are intended. The patches make the related code more consistent and easier to maintain by improving documentation, formatting, and the organization of repo_config_values.

RFC:

If there are other small cleanups in this area that would be useful to include, suggestions are welcome.

Regards, yuchen

Tian Yuchen (3):
environment: simplify repository config getters
environment: clarify repository config getter documentation
environment: reorder variables in repo_config_values structure

 environment.c | 49 ++++++++++++++++++++++++++++++-------------------
 environment.h | 31 ++++++++++++++++---------------
 2 files changed, 46 insertions(+), 34 deletions(-)
Signature · 1 line
2.43.0
#3 lore raw

PATCH 2/3 environment: clarify repository config getter documentation

Update the comment above repository config getters to describe their common behavior.

The getters handle repositories that are not fully initialized by returning the corresponding default values.

---
 environment.h | 11 +++--------
 1 file changed, 3 insertions(+), 8 deletions(-)
environment.h +3 -8
index e7ec5b0437..30678257b5 100644
@@ -175,18 +175,13 @@ int git_default_core_config(const char *var, const char *value,
 			    const struct config_context *ctx, void *cb);
 
 /*
- * Getters for the `protect_hfs` and `protect_ntfs` fields of `struct repo_config_values`.
- * They check `repo->initialized` to prevent calling `repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
+ * Getters for configuration variables in `struct repo_config_values`.
+ * These functions handle uninitialized repositories or non-git
+ * environments by returning appropriate default values.
  */
 int repo_protect_hfs(struct repository *repo);
 int repo_protect_ntfs(struct repository *repo);
 
-/*
- * Getter for the `ignore_case` field of `struct repo_config_values`.
- * It checks `repo->initialized` to prevent calling repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
- */
 int repo_ignore_case(struct repository *repo);
 
 int repo_trust_executable_bit(struct repository *repo);
Signature · 1 line
2.43.0
#1 lore raw

PATCH 1/3 environment: simplify repository config getters

In reply to Tian Yuchen

Drop unnecessary parentheses and NULL checks in repository config getters.

These getters are only used with non-NULL repositories, so the extra checks do not match their current callers.

---
 environment.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
environment.c +9 -9
index 76ee65e62b..f5628b6758 100644
@@ -119,23 +119,23 @@ int is_bare_repository(struct repository *repo)
 
 int repo_protect_ntfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_ntfs :
-		PROTECT_NTFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_ntfs
+		: PROTECT_NTFS_DEFAULT;
 }
 
 int repo_protect_hfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_hfs :
-		PROTECT_HFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_hfs
+		: PROTECT_HFS_DEFAULT;
 }
 
 int repo_ignore_case(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->ignore_case :
-		0;
+	return repo->initialized
+		? repo_config_values(repo)->ignore_case
+		: 0;
 }
 
 int repo_trust_executable_bit(struct repository *repo)
Signature · 1 line
2.43.0
#6 lore raw

PATCH 3/3 environment: reorder variables in repo_config_values structure

In reply to Tian Yuchen

Reorder the fields in struct repo_config_values and its initialization function to follow the order of configuration sections.

Keeping the declaration and initialization order aligned makes the structure easier to review and maintain.

---
 environment.c | 31 +++++++++++++++++++++----------
 environment.h | 20 +++++++++++++-------
 2 files changed, 34 insertions(+), 17 deletions(-)
environment.c +21 -10
index f5628b6758..918d8b50b8 100644
@@ -745,31 +745,42 @@ int git_default_config(const char *var, const char *value,
 
 void repo_config_values_init(struct repo_config_values *cfg)
 {
+	/* core */
 	cfg->attributes_file = NULL;
 	cfg->excludes_file = NULL;
 	cfg->editor_program = NULL;
 	cfg->pager_program = NULL;
 	cfg->askpass_program = NULL;
-	cfg->apply_default_whitespace = NULL;
-	cfg->apply_default_ignorewhitespace = NULL;
-	cfg->push_default = PUSH_DEFAULT_UNSPECIFIED;
-	cfg->autorebase = AUTOREBASE_NEVER;
 	cfg->object_creation_mode = OBJECT_CREATION_MODE;
 	cfg->apply_sparse_checkout = 0;
+	cfg->trust_ctime = 1;
+	cfg->check_stat = 1;
+	cfg->zlib_compression_level = Z_BEST_SPEED;
+	cfg->precomposed_unicode = -1;
+	cfg->core_sparse_checkout_cone = 0;
+	cfg->warn_on_object_refname_ambiguity = 1;
 	cfg->protect_hfs = PROTECT_HFS_DEFAULT;
 	cfg->protect_ntfs = PROTECT_NTFS_DEFAULT;
 	cfg->ignore_case = 0;
 	cfg->trust_executable_bit = 1;
 	cfg->has_symlinks = platform_has_symlinks();
+
+	/* apply */
+	cfg->apply_default_whitespace = NULL;
+	cfg->apply_default_ignorewhitespace = NULL;
+
+	/* branch */
+	cfg->autorebase = AUTOREBASE_NEVER;
 	cfg->branch_track = BRANCH_TRACK_REMOTE;
-	cfg->trust_ctime = 1;
-	cfg->check_stat = 1;
-	cfg->zlib_compression_level = Z_BEST_SPEED;
+
+	/* pack */
 	cfg->pack_compression_level = Z_DEFAULT_COMPRESSION;
-	cfg->precomposed_unicode = -1; /* see probe_utf8_pathname_composition() */
-	cfg->core_sparse_checkout_cone = 0;
+
+	/* push */
+	cfg->push_default = PUSH_DEFAULT_UNSPECIFIED;
+
+	/* sparse */
 	cfg->sparse_expect_files_outside_of_patterns = 0;
-	cfg->warn_on_object_refname_ambiguity = 1;
 }
 
 void repo_config_values_clear(struct repo_config_values *cfg)
environment.h +13 -7
index 30678257b5..52ed13c0fc 100644
@@ -121,16 +121,11 @@ struct repo_config_values {
 	char *editor_program;
 	char *pager_program;
 	char *askpass_program;
-	char *apply_default_whitespace;
-	char *apply_default_ignorewhitespace;
-	enum push_default_type push_default;
-	enum rebase_setup_type autorebase;
 	enum object_creation_mode object_creation_mode;
 	int apply_sparse_checkout;
 	int trust_ctime;
 	int check_stat;
 	int zlib_compression_level;
-	int pack_compression_level;
 	int precomposed_unicode;
 	int core_sparse_checkout_cone;
 	int warn_on_object_refname_ambiguity;
@@ -140,11 +135,22 @@ struct repo_config_values {
 	int trust_executable_bit;
 	int has_symlinks;
 
-	/* section "sparse" config values */
-	int sparse_expect_files_outside_of_patterns;
+	/* section "apply" config values */
+	char *apply_default_whitespace;
+	char *apply_default_ignorewhitespace;
 
 	/* section "branch" config values */
+	enum rebase_setup_type autorebase;
 	enum branch_track branch_track;
+
+	/* section "pack" config values */
+	int pack_compression_level;
+
+	/* section "push" config values */
+	enum push_default_type push_default;
+
+	/* section "sparse" config values */
+	int sparse_expect_files_outside_of_patterns;
 };
 
 struct repo_config_values *repo_config_values(struct repository *repo);
Signature · 1 line
2.43.0
#4 lore raw

Re: [PATCH 2/3] environment: clarify repository config getter documentation

In reply to Tian Yuchen
Quoted 23 lines · Tian Yuchen

Update the comment above repository config getters to describe their common behavior.

The getters handle repositories that are not fully initialized by returning the corresponding default values.

---
 environment.h | 11 +++--------
 1 file changed, 3 insertions(+), 8 deletions(-)
environment.h +0 -1
index e7ec5b0437..30678257b5 100644
@@ -175,18 +175,13 @@ int git_default_core_config(const char *var, const char *value,
 			    const struct config_context *ctx, void *cb);
 
 /*
- * Getters for the `protect_hfs` and `protect_ntfs` fields of `struct repo_config_values`.

environment.h int git_default_core_config(const char *var, const char *value, · line 185

Quoted 10 lines
environment.h +3 -3
- * They check `repo->initialized` to prevent calling `repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
+ * Getters for configuration variables in `struct repo_config_values`.
+ * These functions handle uninitialized repositories or non-git
+ * environments by returning appropriate default values.
  */
 int repo_protect_hfs(struct repository *repo);
 int repo_protect_ntfs(struct repository *repo);
 
-/*

Two puzzlements.

* Is the above comment block meant to apply to repo_ignore_case() in addition to repo_protect_ntfs() and repo_protect_hfs()? If so, the blank line before repo_ignore_case() is a bit misleading.

* The phrase "uninitialized repositories or non-Git environments" strongly hints that I can pass NULL to indicate that we are running in a non-Git environment. However, the change in [PATCH 1/3] we just saw means I would get a segfault if I did so, does it not?

Quoted 7 lines
environment.h +0 -4
- * Getter for the `ignore_case` field of `struct repo_config_values`.
- * It checks `repo->initialized` to prevent calling repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
- */
 int repo_ignore_case(struct repository *repo);
 
 int repo_trust_executable_bit(struct repository *repo);
#7 lore raw

Re: [PATCH 3/3] environment: reorder variables in repo_config_values structure

In reply to Tian Yuchen
Quoted 5 lines · Tian Yuchen

Reorder the fields in struct repo_config_values and its initialization function to follow the order of configuration sections.

Keeping the declaration and initialization order aligned makes the structure easier to review and maintain.

Really?

Do you have some automated tool to make sure these initialization assignments in the environment.c file and declaration in the environment.h file match the order in Documentation/config/*.adoc or something else? Have you designated some list as the authoritative source of truth to check these against? Without such a list to check the code against and a mechanism to enforce the ordering, I find it hard to agree with such a claim that this makes it easier to maintain.

It is typical to list the structure members in the order of stricter to looser alignment requirement of their types. I do not know how strictly it is followed for "struct repo_config_values", but by spreading pointer valued members more widely with smaller enums in between, the change certainly is making the overall structure size larger by requiring more padding between the members with different alignment requirements. Not that we would have 100s of instances of these structures.

Quoted 112 lines
---
 environment.c | 31 +++++++++++++++++++++----------
 environment.h | 20 +++++++++++++-------
 2 files changed, 34 insertions(+), 17 deletions(-)
environment.c +21 -10
index f5628b6758..918d8b50b8 100644
@@ -745,31 +745,42 @@ int git_default_config(const char *var, const char *value,
 
 void repo_config_values_init(struct repo_config_values *cfg)
 {
+	/* core */
 	cfg->attributes_file = NULL;
 	cfg->excludes_file = NULL;
 	cfg->editor_program = NULL;
 	cfg->pager_program = NULL;
 	cfg->askpass_program = NULL;
-	cfg->apply_default_whitespace = NULL;
-	cfg->apply_default_ignorewhitespace = NULL;
-	cfg->push_default = PUSH_DEFAULT_UNSPECIFIED;
-	cfg->autorebase = AUTOREBASE_NEVER;
 	cfg->object_creation_mode = OBJECT_CREATION_MODE;
 	cfg->apply_sparse_checkout = 0;
+	cfg->trust_ctime = 1;
+	cfg->check_stat = 1;
+	cfg->zlib_compression_level = Z_BEST_SPEED;
+	cfg->precomposed_unicode = -1;
+	cfg->core_sparse_checkout_cone = 0;
+	cfg->warn_on_object_refname_ambiguity = 1;
 	cfg->protect_hfs = PROTECT_HFS_DEFAULT;
 	cfg->protect_ntfs = PROTECT_NTFS_DEFAULT;
 	cfg->ignore_case = 0;
 	cfg->trust_executable_bit = 1;
 	cfg->has_symlinks = platform_has_symlinks();
+
+	/* apply */
+	cfg->apply_default_whitespace = NULL;
+	cfg->apply_default_ignorewhitespace = NULL;
+
+	/* branch */
+	cfg->autorebase = AUTOREBASE_NEVER;
 	cfg->branch_track = BRANCH_TRACK_REMOTE;
-	cfg->trust_ctime = 1;
-	cfg->check_stat = 1;
-	cfg->zlib_compression_level = Z_BEST_SPEED;
+
+	/* pack */
 	cfg->pack_compression_level = Z_DEFAULT_COMPRESSION;
-	cfg->precomposed_unicode = -1; /* see probe_utf8_pathname_composition() */
-	cfg->core_sparse_checkout_cone = 0;
+
+	/* push */
+	cfg->push_default = PUSH_DEFAULT_UNSPECIFIED;
+
+	/* sparse */
 	cfg->sparse_expect_files_outside_of_patterns = 0;
-	cfg->warn_on_object_refname_ambiguity = 1;
 }
 
 void repo_config_values_clear(struct repo_config_values *cfg)
environment.h +13 -7
index 30678257b5..52ed13c0fc 100644
@@ -121,16 +121,11 @@ struct repo_config_values {
 	char *editor_program;
 	char *pager_program;
 	char *askpass_program;
-	char *apply_default_whitespace;
-	char *apply_default_ignorewhitespace;
-	enum push_default_type push_default;
-	enum rebase_setup_type autorebase;
 	enum object_creation_mode object_creation_mode;
 	int apply_sparse_checkout;
 	int trust_ctime;
 	int check_stat;
 	int zlib_compression_level;
-	int pack_compression_level;
 	int precomposed_unicode;
 	int core_sparse_checkout_cone;
 	int warn_on_object_refname_ambiguity;
@@ -140,11 +135,22 @@ struct repo_config_values {
 	int trust_executable_bit;
 	int has_symlinks;
 
-	/* section "sparse" config values */
-	int sparse_expect_files_outside_of_patterns;
+	/* section "apply" config values */
+	char *apply_default_whitespace;
+	char *apply_default_ignorewhitespace;
 
 	/* section "branch" config values */
+	enum rebase_setup_type autorebase;
 	enum branch_track branch_track;
+
+	/* section "pack" config values */
+	int pack_compression_level;
+
+	/* section "push" config values */
+	enum push_default_type push_default;
+
+	/* section "sparse" config values */
+	int sparse_expect_files_outside_of_patterns;
 };
 
 struct repo_config_values *repo_config_values(struct repository *repo);
#2 lore raw

Re: [PATCH 1/3] environment: simplify repository config getters

In reply to Tian Yuchen
Quoted 5 lines · Tian Yuchen

Drop unnecessary parentheses and NULL checks in repository config getters.

These getters are only used with non-NULL repositories, so the extra checks do not match their current callers.

OK. If repo MUST always be non-NULL, even when we haven't fully initialized them, then not punting on repo==NULL case like the original code is definitely an improvement.

Looking good.

Quoted 46 lines
---
 environment.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
environment.c +9 -9
index 76ee65e62b..f5628b6758 100644
@@ -119,23 +119,23 @@ int is_bare_repository(struct repository *repo)
 
 int repo_protect_ntfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_ntfs :
-		PROTECT_NTFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_ntfs
+		: PROTECT_NTFS_DEFAULT;
 }
 
 int repo_protect_hfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_hfs :
-		PROTECT_HFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_hfs
+		: PROTECT_HFS_DEFAULT;
 }
 
 int repo_ignore_case(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->ignore_case :
-		0;
+	return repo->initialized
+		? repo_config_values(repo)->ignore_case
+		: 0;
 }
 
 int repo_trust_executable_bit(struct repository *repo)
#8 lore raw

Re: [PATCH 3/3] environment: reorder variables in repo_config_values structure

In reply to Junio C Hamano
Quoted 10 lines · Junio C Hamano
Quoted 5 lines · Tian Yuchen

Reorder the fields in struct repo_config_values and its initialization function to follow the order of configuration sections.

Keeping the declaration and initialization order aligned makes the structure easier to review and maintain.

Really?

Quoted 8 lines

Do you have some automated tool to make sure these initialization assignments in the environment.c file and declaration in the environment.h file match the order in Documentation/config/*.adoc or something else? Have you designated some list as the authoritative source of truth to check these against? Without such a list to check the code against and a mechanism to enforce the ordering, I find it hard to agree with such a claim that this makes it easier to maintain.

I see.

Quoted 10 lines

It is typical to list the structure members in the order of stricter to looser alignment requirement of their types. I do not know how strictly it is followed for "struct repo_config_values", but by spreading pointer valued members more widely with smaller enums in between, the change certainly is making the overall structure size larger by requiring more padding between the members with different alignment requirements. Not that we would have 100s of instances of these structures.

Oh, I overlooked the size issue. Thanks for pointing out.

I think I will drop this commit. However, the original comments:

struct repo_config_values {
	/* section "core" config values */
	char *attributes_file;
	char *excludes_file;
	char *editor_program;
	char *pager_program;
	char *askpass_program;
	char *apply_default_whitespace;
	char *apply_default_ignorewhitespace;
	enum push_default_type push_default;
	enum rebase_setup_type autorebase;
	enum object_creation_mode object_creation_mode;
	int apply_sparse_checkout;
	int trust_ctime;
	int check_stat;
	int zlib_compression_level;
	int pack_compression_level;
	int precomposed_unicode;
	int core_sparse_checkout_cone;
	int warn_on_object_refname_ambiguity;
	int protect_hfs;
	int protect_ntfs;
	int ignore_case;
	int trust_executable_bit;
	int has_symlinks;

	/* section "sparse" config values */
	int sparse_expect_files_outside_of_patterns;

	/* section "branch" config values */
	enum branch_track branch_track;
};

still do not accurately reflect the grouping of the members, right? Can we remove them directly instead?

Thanks, yuchen

Quoted 112 lines
Quoted 112 lines
---
  environment.c | 31 +++++++++++++++++++++----------
  environment.h | 20 +++++++++++++-------
  2 files changed, 34 insertions(+), 17 deletions(-)
environment.c +21 -10
index f5628b6758..918d8b50b8 100644
@@ -745,31 +745,42 @@ int git_default_config(const char *var, const char *value,
 
  void repo_config_values_init(struct repo_config_values *cfg)
  {
+	/* core */
  	cfg->attributes_file = NULL;
  	cfg->excludes_file = NULL;
  	cfg->editor_program = NULL;
  	cfg->pager_program = NULL;
  	cfg->askpass_program = NULL;
-	cfg->apply_default_whitespace = NULL;
-	cfg->apply_default_ignorewhitespace = NULL;
-	cfg->push_default = PUSH_DEFAULT_UNSPECIFIED;
-	cfg->autorebase = AUTOREBASE_NEVER;
  	cfg->object_creation_mode = OBJECT_CREATION_MODE;
  	cfg->apply_sparse_checkout = 0;
+	cfg->trust_ctime = 1;
+	cfg->check_stat = 1;
+	cfg->zlib_compression_level = Z_BEST_SPEED;
+	cfg->precomposed_unicode = -1;
+	cfg->core_sparse_checkout_cone = 0;
+	cfg->warn_on_object_refname_ambiguity = 1;
  	cfg->protect_hfs = PROTECT_HFS_DEFAULT;
  	cfg->protect_ntfs = PROTECT_NTFS_DEFAULT;
  	cfg->ignore_case = 0;
  	cfg->trust_executable_bit = 1;
  	cfg->has_symlinks = platform_has_symlinks();
+
+	/* apply */
+	cfg->apply_default_whitespace = NULL;
+	cfg->apply_default_ignorewhitespace = NULL;
+
+	/* branch */
+	cfg->autorebase = AUTOREBASE_NEVER;
  	cfg->branch_track = BRANCH_TRACK_REMOTE;
-	cfg->trust_ctime = 1;
-	cfg->check_stat = 1;
-	cfg->zlib_compression_level = Z_BEST_SPEED;
+
+	/* pack */
  	cfg->pack_compression_level = Z_DEFAULT_COMPRESSION;
-	cfg->precomposed_unicode = -1; /* see probe_utf8_pathname_composition() */
-	cfg->core_sparse_checkout_cone = 0;
+
+	/* push */
+	cfg->push_default = PUSH_DEFAULT_UNSPECIFIED;
+
+	/* sparse */
  	cfg->sparse_expect_files_outside_of_patterns = 0;
-	cfg->warn_on_object_refname_ambiguity = 1;
  }
 
  void repo_config_values_clear(struct repo_config_values *cfg)
environment.h +13 -7
index 30678257b5..52ed13c0fc 100644
@@ -121,16 +121,11 @@ struct repo_config_values {
  	char *editor_program;
  	char *pager_program;
  	char *askpass_program;
-	char *apply_default_whitespace;
-	char *apply_default_ignorewhitespace;
-	enum push_default_type push_default;
-	enum rebase_setup_type autorebase;
  	enum object_creation_mode object_creation_mode;
  	int apply_sparse_checkout;
  	int trust_ctime;
  	int check_stat;
  	int zlib_compression_level;
-	int pack_compression_level;
  	int precomposed_unicode;
  	int core_sparse_checkout_cone;
  	int warn_on_object_refname_ambiguity;
@@ -140,11 +135,22 @@ struct repo_config_values {
  	int trust_executable_bit;
  	int has_symlinks;
 
-	/* section "sparse" config values */
-	int sparse_expect_files_outside_of_patterns;
+	/* section "apply" config values */
+	char *apply_default_whitespace;
+	char *apply_default_ignorewhitespace;
 
  	/* section "branch" config values */
+	enum rebase_setup_type autorebase;
  	enum branch_track branch_track;
+
+	/* section "pack" config values */
+	int pack_compression_level;
+
+	/* section "push" config values */
+	enum push_default_type push_default;
+
+	/* section "sparse" config values */
+	int sparse_expect_files_outside_of_patterns;
  };
 
  struct repo_config_values *repo_config_values(struct repository *repo);
#5 lore raw

Re: [PATCH 2/3] environment: clarify repository config getter documentation

In reply to Junio C Hamano
Quoted 34 lines · Junio C Hamano
Quoted 32 lines · Tian Yuchen

Update the comment above repository config getters to describe their common behavior.

The getters handle repositories that are not fully initialized by returning the corresponding default values.

---
  environment.h | 11 +++--------
  1 file changed, 3 insertions(+), 8 deletions(-)
environment.h +3 -3
index e7ec5b0437..30678257b5 100644
@@ -175,18 +175,13 @@ int git_default_core_config(const char *var, const char *value,
  			    const struct config_context *ctx, void *cb);
 
  /*
- * Getters for the `protect_hfs` and `protect_ntfs` fields of `struct repo_config_values`.
- * They check `repo->initialized` to prevent calling `repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
+ * Getters for configuration variables in `struct repo_config_values`.
+ * These functions handle uninitialized repositories or non-git
+ * environments by returning appropriate default values.
   */
  int repo_protect_hfs(struct repository *repo);
  int repo_protect_ntfs(struct repository *repo);
 
Quoted 8 lines

environment.h int git_default_core_config(const char *var, const char *value, · line 185

Quoted 1 line
environment.h +0 -1
-/*

Two puzzlements.

* Is the above comment block meant to apply to repo_ignore_case() in addition to repo_protect_ntfs() and repo_protect_hfs()? If so, the blank line before repo_ignore_case() is a bit misleading.

Not really, they are meant to apply to all getters below. I will remove the blank lines.

Quoted 6 lines

* The phrase "uninitialized repositories or non-Git environments" strongly hints that I can pass NULL to indicate that we are running in a non-Git environment. However, the change in [PATCH 1/3] we just saw means I would get a segfault if I did so, does it not?

This is a mistake. I meant "these getters can handle repositories, even when they are not fully initailzed" but not "these getters can handle whatever we pass in". So I will change it in the next reroll.

Quoted 7 lines

environment.h int git_default_core_config(const char *var, const char *value, · line 187

Quoted 7 lines
environment.h +0 -4
- * Getter for the `ignore_case` field of `struct repo_config_values`.
- * It checks `repo->initialized` to prevent calling repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
- */
  int repo_ignore_case(struct repository *repo);
 
  int repo_trust_executable_bit(struct repository *repo);

Thanks! yuchen

#11 lore raw

v2 cover environment: clean up repository config handling

In reply to Tian Yuchen

Hi all,

This series contains several cleanup patches for repository configuration handling.

No functional changes are intended. The patches make the related code more consistent and easier to maintain by improving documentation, formatting, and the organization of repo_config_values.

RFC:
If there are other small cleanups in this area that would be useful to include, suggestions are welcome.

Regards, yuchen

Changes since v1:

- in commit 2/3, drop several blank lines to group the getters under the comment. Note that the comment does not apply to repo_excludes_file.

- in commit 3/3, do not change the order of the members. Instead, drop the original comments that do not accurately categorize them.

Tian Yuchen (3):
environment: simplify repository config getters
environment: clarify repository config getter documentation
environment: remove inaccurate repo_config_values comments

 environment.c | 18 +++++++++---------
 environment.h | 19 +++----------------
 2 files changed, 12 insertions(+), 25 deletions(-)
Signature · 1 line
2.43.0
#12 lore raw

PATCH v2 1/3 environment: simplify repository config getters

Drop unnecessary parentheses and NULL checks in repository config getters.

These getters are only used with non-NULL repositories, so the extra checks do not match their current callers.

---
 environment.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
environment.c +9 -9
index 76ee65e62b..f5628b6758 100644
@@ -119,23 +119,23 @@ int is_bare_repository(struct repository *repo)
 
 int repo_protect_ntfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_ntfs :
-		PROTECT_NTFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_ntfs
+		: PROTECT_NTFS_DEFAULT;
 }
 
 int repo_protect_hfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_hfs :
-		PROTECT_HFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_hfs
+		: PROTECT_HFS_DEFAULT;
 }
 
 int repo_ignore_case(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->ignore_case :
-		0;
+	return repo->initialized
+		? repo_config_values(repo)->ignore_case
+		: 0;
 }
 
 int repo_trust_executable_bit(struct repository *repo)
Signature · 1 line
2.43.0
#15 lore raw

PATCH v2 2/3 environment: clarify repository config getter documentation

In reply to Tian Yuchen

Update the comment above repository config getters to describe their common behavior.

The getters handle repositories that are not fully initialized by returning the corresponding default values.

---
 environment.h | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)
environment.h +3 -11
index e7ec5b0437..1a58b553b5 100644
@@ -175,22 +175,14 @@ int git_default_core_config(const char *var, const char *value,
 			    const struct config_context *ctx, void *cb);
 
 /*
- * Getters for the `protect_hfs` and `protect_ntfs` fields of `struct repo_config_values`.
- * They check `repo->initialized` to prevent calling `repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
+ * Getters for configuration variables in `struct repo_config_values`.
+ * These functions handle repositories that are not fully initialized
+ * by returning appropriate default values.
  */
 int repo_protect_hfs(struct repository *repo);
 int repo_protect_ntfs(struct repository *repo);
-
-/*
- * Getter for the `ignore_case` field of `struct repo_config_values`.
- * It checks `repo->initialized` to prevent calling repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
- */
 int repo_ignore_case(struct repository *repo);
-
 int repo_trust_executable_bit(struct repository *repo);
-
 int repo_has_symlinks(struct repository *repo);
 
 const char *repo_excludes_file(struct repository *repo);
Signature · 1 line
2.43.0
#17 lore raw

PATCH v2 3/3 environment: remove inaccurate repo_config_values comments

In reply to Tian Yuchen

The section comments in struct repo_config_values do not accurately describe all members grouped under them. Remove them rather than implying a relationship that does not exist.

---
 environment.h | 5 -----
 1 file changed, 5 deletions(-)
environment.h +0 -5
index 1a58b553b5..ab52330159 100644
@@ -115,7 +115,6 @@ enum object_creation_mode {
 };
 
 struct repo_config_values {
-	/* section "core" config values */
 	char *attributes_file;
 	char *excludes_file;
 	char *editor_program;
@@ -139,11 +138,7 @@ struct repo_config_values {
 	int ignore_case;
 	int trust_executable_bit;
 	int has_symlinks;
-
-	/* section "sparse" config values */
 	int sparse_expect_files_outside_of_patterns;
-
-	/* section "branch" config values */
 	enum branch_track branch_track;
 };
 
Signature · 1 line
2.43.0
#9 lore raw

Re: [PATCH 3/3] environment: reorder variables in repo_config_values structure

In reply to Tian Yuchen
Quoted 33 lines · Tian Yuchen
Quoted 10 lines · Junio C Hamano
Quoted 5 lines · Tian Yuchen

Reorder the fields in struct repo_config_values and its initialization function to follow the order of configuration sections.

Keeping the declaration and initialization order aligned makes the structure easier to review and maintain.

Really?

Quoted 8 lines

Do you have some automated tool to make sure these initialization assignments in the environment.c file and declaration in the environment.h file match the order in Documentation/config/*.adoc or something else? Have you designated some list as the authoritative source of truth to check these against? Without such a list to check the code against and a mechanism to enforce the ordering, I find it hard to agree with such a claim that this makes it easier to maintain.

I see.

Quoted 10 lines

It is typical to list the structure members in the order of stricter to looser alignment requirement of their types. I do not know how strictly it is followed for "struct repo_config_values", but by spreading pointer valued members more widely with smaller enums in between, the change certainly is making the overall structure size larger by requiring more padding between the members with different alignment requirements. Not that we would have 100s of instances of these structures.

Quoted 1 line

Oh, I overlooked the size issue. Thanks for pointing out.

I didn't mean to "point out" any size issue. As I said, it is not like we have hundreds of these, so padding bloat here and there would not matter and if we get a readability boost by reordering into a sensible order, that by itself could be a win.

#13 lore raw

Re: [PATCH v2 1/3] environment: simplify repository config getters

In reply to Tian Yuchen
Quoted 5 lines · Tian Yuchen

Drop unnecessary parentheses and NULL checks in repository config getters.

These getters are only used with non-NULL repositories, so the extra checks do not match their current callers.

You would need to explain why it is sensible to enforce on future callers the same rule that current callers honor, or why it is unlikely that we will gain any more callers in the future (which would justify catering only to current callers).

Quoted 45 lines
---
 environment.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
environment.c +9 -9
index 76ee65e62b..f5628b6758 100644
@@ -119,23 +119,23 @@ int is_bare_repository(struct repository *repo)
 
 int repo_protect_ntfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_ntfs :
-		PROTECT_NTFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_ntfs
+		: PROTECT_NTFS_DEFAULT;
 }
 
 int repo_protect_hfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_hfs :
-		PROTECT_HFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_hfs
+		: PROTECT_HFS_DEFAULT;
 }
 
 int repo_ignore_case(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->ignore_case :
-		0;
+	return repo->initialized
+		? repo_config_values(repo)->ignore_case
+		: 0;
 }
 
 int repo_trust_executable_bit(struct repository *repo)
#16 lore raw

Re: [PATCH v2 2/3] environment: clarify repository config getter documentation

In reply to Tian Yuchen
Quoted 19 lines · Tian Yuchen

Update the comment above repository config getters to describe their common behavior.

The getters handle repositories that are not fully initialized by returning the corresponding default values.

---
 environment.h | 14 +++-----------
 1 file changed, 3 insertions(+), 11 deletions(-)
environment.h +0 -0
index e7ec5b0437..1a58b553b5 100644

environment.h int git_default_core_config(const char *var, const char *value, · line 181

Quoted 10 lines
environment.h +3 -3
@@ -175,22 +175,14 @@ int git_default_core_config(const char *var, const char *value,
 			    const struct config_context *ctx, void *cb);
 
 /*
- * Getters for the `protect_hfs` and `protect_ntfs` fields of `struct repo_config_values`.
- * They check `repo->initialized` to prevent calling `repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
+ * Getters for configuration variables in `struct repo_config_values`.
+ * These functions handle repositories that are not fully initialized
+ * by returning appropriate default values.
  */

Do we also want to mention that calling them when the caller is outside a repository is an error, or is it obvious enough?

Quoted 15 lines
environment.h +0 -8
 int repo_protect_hfs(struct repository *repo);
 int repo_protect_ntfs(struct repository *repo);
-
-/*
- * Getter for the `ignore_case` field of `struct repo_config_values`.
- * It checks `repo->initialized` to prevent calling repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
- */
 int repo_ignore_case(struct repository *repo);
-
 int repo_trust_executable_bit(struct repository *repo);
-
 int repo_has_symlinks(struct repository *repo);
 
 const char *repo_excludes_file(struct repository *repo);
#10 lore raw

Re: [PATCH 3/3] environment: reorder variables in repo_config_values structure

In reply to Junio C Hamano
Quoted 35 lines · Junio C Hamano
Quoted 33 lines · Tian Yuchen
Quoted 10 lines · Junio C Hamano
Quoted 5 lines · Tian Yuchen

Reorder the fields in struct repo_config_values and its initialization function to follow the order of configuration sections.

Keeping the declaration and initialization order aligned makes the structure easier to review and maintain.

Really?

Quoted 8 lines

Do you have some automated tool to make sure these initialization assignments in the environment.c file and declaration in the environment.h file match the order in Documentation/config/*.adoc or something else? Have you designated some list as the authoritative source of truth to check these against? Without such a list to check the code against and a mechanism to enforce the ordering, I find it hard to agree with such a claim that this makes it easier to maintain.

I see.

Quoted 10 lines

It is typical to list the structure members in the order of stricter to looser alignment requirement of their types. I do not know how strictly it is followed for "struct repo_config_values", but by spreading pointer valued members more widely with smaller enums in between, the change certainly is making the overall structure size larger by requiring more padding between the members with different alignment requirements. Not that we would have 100s of instances of these structures.

Quoted 7 lines
Quoted 1 line

Oh, I overlooked the size issue. Thanks for pointing out.

I didn't mean to "point out" any size issue. As I said, it is not like we have hundreds of these, so padding bloat here and there would not matter and if we get a readability boost by reordering into a sensible order, that by itself could be a win.

Okay.

As you said before, the boost on readability seems to be limited. Reordering by config section is not a strong maintenance rule without an authoritative source. So let's don't reorder them anyways.

Thanks, yuchen

#14 lore raw

Re: [PATCH v2 1/3] environment: simplify repository config getters

In reply to Junio C Hamano
Quoted 57 lines · Junio C Hamano
Quoted 5 lines · Tian Yuchen

Drop unnecessary parentheses and NULL checks in repository config getters.

These getters are only used with non-NULL repositories, so the extra checks do not match their current callers.

You would need to explain why it is sensible to enforce on future callers the same rule that current callers honor, or why it is unlikely that we will gain any more callers in the future (which would justify catering only to current callers).

Quoted 44 lines
---
  environment.c | 18 +++++++++---------
  1 file changed, 9 insertions(+), 9 deletions(-)
environment.c +9 -9
index 76ee65e62b..f5628b6758 100644
@@ -119,23 +119,23 @@ int is_bare_repository(struct repository *repo)
 
  int repo_protect_ntfs(struct repository *repo)
  {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_ntfs :
-		PROTECT_NTFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_ntfs
+		: PROTECT_NTFS_DEFAULT;
  }
 
  int repo_protect_hfs(struct repository *repo)
  {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_hfs :
-		PROTECT_HFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_hfs
+		: PROTECT_HFS_DEFAULT;
  }
 
  int repo_ignore_case(struct repository *repo)
  {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->ignore_case :
-		0;
+	return repo->initialized
+		? repo_config_values(repo)->ignore_case
+		: 0;
  }
 
Quoted 1 line

environment.c int is_bare_repository(struct repository *repo) · line 141

Quoted 1 line
environment.c +0 -0
  int repo_trust_executable_bit(struct repository *repo)

I see, will change the commit message then ;)

Thanks, yuchen

#18 lore raw

v3 cover environment: clean up repository config handling

In reply to Tian Yuchen

Hi all,

This series contains several cleanup patches for repository configuration handling.

No functional changes are intended. The patches make the related code more consistent and easier to maintain by improving documentation, formatting, and the organization of repo_config_values.

RFC:
If there are other small cleanups in this area that would be useful to include, suggestions are welcome.

Regards, yuchen

Changes since v2:

- in the commit message of patch 1/3, explain why NULL repository is not allowed.

- in patch 2/3, mention in the comment that NULL repository shouldn't be passed in.

Tian Yuchen (3):
environment: drop redundant NULL checks in config getters
environment: clarify repository config getter documentation
environment: remove inaccurate repo_config_values comments

 environment.c | 18 +++++++++---------
 environment.h | 20 ++++----------------
 2 files changed, 13 insertions(+), 25 deletions(-)
Signature · 1 line
2.43.0
#19 lore raw

PATCH v3 1/3 environment: drop redundant NULL checks in config getters

These repository config getters require a valid repository pointer. While an uninitialized repository is a valid state and is handled by returning default values, passing NULL is a programming error.

Drop the NULL checks so that invalid callers are not silently accepted.

---
 environment.c | 18 +++++++++---------
 1 file changed, 9 insertions(+), 9 deletions(-)
environment.c +9 -9
index 76ee65e62b..f5628b6758 100644
@@ -119,23 +119,23 @@ int is_bare_repository(struct repository *repo)
 
 int repo_protect_ntfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_ntfs :
-		PROTECT_NTFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_ntfs
+		: PROTECT_NTFS_DEFAULT;
 }
 
 int repo_protect_hfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_hfs :
-		PROTECT_HFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_hfs
+		: PROTECT_HFS_DEFAULT;
 }
 
 int repo_ignore_case(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->ignore_case :
-		0;
+	return repo->initialized
+		? repo_config_values(repo)->ignore_case
+		: 0;
 }
 
 int repo_trust_executable_bit(struct repository *repo)
Signature · 1 line
2.43.0
#21 lore raw

PATCH v3 2/3 environment: clarify repository config getter documentation

In reply to Tian Yuchen

Update the comment above repository config getters to describe their common behavior.

The getters handle repositories that are not fully initialized by returning the corresponding default values.

---
 environment.h | 15 ++++-----------
 1 file changed, 4 insertions(+), 11 deletions(-)
environment.h +4 -11
index e7ec5b0437..6f864c1635 100644
@@ -175,22 +175,15 @@ int git_default_core_config(const char *var, const char *value,
 			    const struct config_context *ctx, void *cb);
 
 /*
- * Getters for the `protect_hfs` and `protect_ntfs` fields of `struct repo_config_values`.
- * They check `repo->initialized` to prevent calling `repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
+ * Getters for configuration variables in `struct repo_config_values`.
+ * These functions require a non-NULL repository pointer and handle
+ * repositories that are not fully initialized by returning appropriate
+ * default values.
  */
 int repo_protect_hfs(struct repository *repo);
 int repo_protect_ntfs(struct repository *repo);
-
-/*
- * Getter for the `ignore_case` field of `struct repo_config_values`.
- * It checks `repo->initialized` to prevent calling repo_config_values()`
- * before the repository setup is fully complete or in non-git environments.
- */
 int repo_ignore_case(struct repository *repo);
-
 int repo_trust_executable_bit(struct repository *repo);
-
 int repo_has_symlinks(struct repository *repo);
 
 const char *repo_excludes_file(struct repository *repo);
Signature · 1 line
2.43.0
#22 lore raw

PATCH v3 3/3 environment: remove inaccurate repo_config_values comments

In reply to Tian Yuchen

The section comments in struct repo_config_values do not accurately describe all members grouped under them. Remove them rather than implying a relationship that does not exist.

---
 environment.h | 5 -----
 1 file changed, 5 deletions(-)
environment.h +0 -5
index 6f864c1635..67fd387d35 100644
@@ -115,7 +115,6 @@ enum object_creation_mode {
 };
 
 struct repo_config_values {
-	/* section "core" config values */
 	char *attributes_file;
 	char *excludes_file;
 	char *editor_program;
@@ -139,11 +138,7 @@ struct repo_config_values {
 	int ignore_case;
 	int trust_executable_bit;
 	int has_symlinks;
-
-	/* section "sparse" config values */
 	int sparse_expect_files_outside_of_patterns;
-
-	/* section "branch" config values */
 	enum branch_track branch_track;
 };
 
Signature · 1 line
2.43.0
#23 lore raw
In reply to Tian Yuchen
Quoted 5 lines · Tian Yuchen

Hi all,

This series contains several cleanup patches for repository configuration handling.

Quoted 7 lines

No functional changes are intended. The patches make the related code more consistent and easier to maintain by improving documentation, formatting, and the organization of repo_config_values.

RFC:
If there are other small cleanups in this area that would be useful to include, suggestions are welcome.

Somewhat unrelated to this patch series, but I was wondering whether you plan to drop the limitation in repo_config_values() that requires that the passed-in repository is the_repository. This limitation is starting to create problems as more and more of our infrastructure is migrating into struct repo_config_values, so using a different repo than the_repository is starting to become harder and harder in our codebase.

Thanks!

Patrick

#24 lore raw
Quoted 22 lines · Patrick Steinhardt
Quoted 5 lines · Tian Yuchen

Hi all,

This series contains several cleanup patches for repository configuration handling.

Quoted 7 lines

No functional changes are intended. The patches make the related code more consistent and easier to maintain by improving documentation, formatting, and the organization of repo_config_values.

RFC:
If there are other small cleanups in this area that would be useful to include, suggestions are welcome.

Somewhat unrelated to this patch series, but I was wondering whether you plan to drop the limitation in repo_config_values() that requires that the passed-in repository is the_repository. This limitation is starting to create problems as more and more of our infrastructure is migrating into struct repo_config_values, so using a different repo than the_repository is starting to become harder and harder in our codebase.

Quoted 3 lines

Thanks!

Patrick

Hmph, that is an interesting point. What is our plan to really enable the use of repository instances other than 'the_repository' here? They of course need to be initialized with repo_init(), but is that enough to sensibly use the embedded 'repo_settings' and 'repo_config_values' structures? (By the way, it is not entirely clear to me why we need both and how we sift variables between them.) Some code paths need to work outside a repository and still need to know about per-user or per-system settings. We were perfectly happy reading from global variables when we had the majority of them there. It is my understanding that they are now found in 'repo_config_values' or 'repo_settings' associated with 'the_repository', which I think is something we cannot really avoid doing. Unless we try to get rid of 'the_repository' and instead have free-standing 'repo_settings' and 'repo_config_values' structures that are not tied to any repository instance, we are back to depending on a set of global variables. 😞

In any case, all of that has little to do with this series, I suspect, unless we are redesigning these configurations and settings in such a way that they are not necessarily tied to any repository instance. While I do not know the exact details, I can imagine a hierarchical system where system- and user-wide sets of setting values are known independently of any repository, only to be overridden by per-repository settings using a last-one-wins strategy at lookup time.

#25 lore raw
Quoted 28 lines · Junio C Hamano
Quoted 22 lines · Patrick Steinhardt
Quoted 5 lines · Tian Yuchen

Hi all,

This series contains several cleanup patches for repository configuration handling.

Quoted 7 lines

No functional changes are intended. The patches make the related code more consistent and easier to maintain by improving documentation, formatting, and the organization of repo_config_values.

RFC:
If there are other small cleanups in this area that would be useful to include, suggestions are welcome.

Somewhat unrelated to this patch series, but I was wondering whether you plan to drop the limitation in repo_config_values() that requires that the passed-in repository is the_repository. This limitation is starting to create problems as more and more of our infrastructure is migrating into struct repo_config_values, so using a different repo than the_repository is starting to become harder and harder in our codebase.

Quoted 3 lines

Thanks!

Patrick

Quoted 7 lines

Hmph, that is an interesting point. What is our plan to really enable the use of repository instances other than 'the_repository' here? They of course need to be initialized with repo_init(), but is that enough to sensibly use the embedded 'repo_settings' and 'repo_config_values' structures? (By the way, it is not entirely clear to me why we need both and how we sift variables between them.)

Yeah, this split is adding to the confusion indeed. I think that we should make it a goal to unify those going forward.

[snip]

Quoted 8 lines

In any case, all of that has little to do with this series, I suspect, unless we are redesigning these configurations and settings in such a way that they are not necessarily tied to any repository instance. While I do not know the exact details, I can imagine a hierarchical system where system- and user-wide sets of setting values are known independently of any repository, only to be overridden by per-repository settings using a last-one-wins strategy at lookup time.

I've been wondering for a while whether we're operating at the wrong level here. Both repo_settings and repo_config_values indicates that we're operating in the context of a repository, but as you mention that may not even be the case.

I don't think the approach is inherently flawed though. From my point of view, the best way forward is to merge those two and then generalize them into something like git_config_values or git_settings, depending on which of both variants we want to retain. We would then have two levels:

- One on the repository level as we have it today.

- One truly global variable, because that stuff in fact _is_ global.

We'd then adapt repo_config_values() so that it knows to populate either of those variables depending on whether or not the user passes a valid repository, and returns a constant pointer to the respective structure. Callers MUST NOT modify that structure -- if they want to, they'll have to make a copy and pass it down the calling stack.

The last part about not modifying that structure could be quite a bit painful though, as it would mean that we might have to adapt call chains to pass down a struct git_config_values instead of a struct repository. But arguably, that's the right thing to do anyway for at least some subsystems that are independent of repositories.

As you say though, none of this is really related to this patch series at hand, and I don't think we need to resolve this discussion before we can merge it. I just want to make sure that we have a plan for how to get rid of the_repository instead of only shuffling stuff around.

Patrick

#26 lore raw
Quoted 40 lines · Patrick Steinhardt
Quoted 24 lines · Junio C Hamano
Quoted 22 lines · Patrick Steinhardt
Quoted 5 lines · Tian Yuchen

Hi all,

This series contains several cleanup patches for repository configuration handling.

Quoted 7 lines

No functional changes are intended. The patches make the related code more consistent and easier to maintain by improving documentation, formatting, and the organization of repo_config_values.

RFC:
If there are other small cleanups in this area that would be useful to include, suggestions are welcome.

Somewhat unrelated to this patch series, but I was wondering whether you plan to drop the limitation in repo_config_values() that requires that the passed-in repository is the_repository. This limitation is starting to create problems as more and more of our infrastructure is migrating into struct repo_config_values, so using a different repo than the_repository is starting to become harder and harder in our codebase.

Quoted 6 lines
Quoted 3 lines

Thanks!

Patrick

Hmph, that is an interesting point. What is our plan to really
...

Yeah, this split is adding to the confusion indeed. I think that we should make it a goal to unify those going forward.
...
The last part about not modifying that structure could be quite a bit painful though, as it would mean that we might have to adapt call chains to pass down a struct git_config_values instead of a struct repository. But arguably, that's the right thing to do anyway for at least some subsystems that are independent of repositories.

Quoted 4 lines

As you say though, none of this is really related to this patch series at hand, and I don't think we need to resolve this discussion before we can merge it. I just want to make sure that we have a plan for how to get rid of the_repository instead of only shuffling stuff around.

Well, after this sort-of offtopic exchange, the thread went dark.

Is anybody interested in reviewing these patches and move the topic forward?

Thanks.

#20 lore raw

Re: [PATCH v3 1/3] environment: drop redundant NULL checks in config getters

In reply to Tian Yuchen
Quoted 5 lines · Tian Yuchen

These repository config getters require a valid repository pointer. While an uninitialized repository is a valid state and is handled by returning default values, passing NULL is a programming error.

Drop the NULL checks so that invalid callers are not silently accepted.

I'm not quite convinced that having these checks in the first place is a good idea. The single biggest problem is that we silently ignore the settings in case the repository just happens to be uninitialized, and we wouldn't ever notice.

On top of that, we even fall back to the wrong value: if we don't have a repository, we shouldn't fall back to the default values. Instead, shouldn't we fall back to the global- or system-level configuration?

I'm not convinced that this design is correct. What I think we should be doing is:

- Have the functions accept an optional repository.

- If a repository is passed, then we verify that it is initialized. If not, we BUG.

- If we haven't yet read the configuration for that repository, then we automatically do it so that we can also pass a repository other than the_repository.

- If no repository is passed, then we populate a global variable that contains the system- and global-level configuration and return that value instead.

That'd work both in the context where we have a repository and where we don't have one, and we'd detect the edge case where we have a repository that is uninitialized.

Quoted 5 lines
environment.c +0 -0
index 76ee65e62b..f5628b6758 100644

environment.c int is_bare_repository(struct repository *repo) · line 125

Quoted 10 lines
environment.c +3 -3
@@ -119,23 +119,23 @@ int is_bare_repository(struct repository *repo)
 
 int repo_protect_ntfs(struct repository *repo)
 {
-	return (repo && repo->initialized) ?
-		repo_config_values(repo)->protect_ntfs :
-		PROTECT_NTFS_DEFAULT;
+	return repo->initialized
+		? repo_config_values(repo)->protect_ntfs
+		: PROTECT_NTFS_DEFAULT;
 }

So I think if we want to lose these checks, we should lose both of them and require the repository to be initialized. But I feel like this whole subsystem needs a bit of a redesign before we can continue iterating on it.

Patrick

#27 lore raw

Re*: [PATCH v3 0/3] environment: clean up repository config handling

In reply to Patrick Steinhardt
Quoted 45 lines · Patrick Steinhardt
Quoted 6 lines · Junio C Hamano
Quoted 4 lines · Patrick Steinhardt

On Fri, Aug 07, 2026 at 04:59:29PM +0800, Tian Yuchen wrote:
Somewhat unrelated to this patch series, but I was wondering whether you plan to drop the limitation in repo_config_values() that requires that the passed-in repository is the_repository. ...

Quoted 7 lines

Hmph, that is an interesting point. What is our plan to really enable the use of repository instances other than 'the_repository' here? They of course need to be initialized with repo_init(), but is that enough to sensibly use the embedded 'repo_settings' and 'repo_config_values' structures? (By the way, it is not entirely clear to me why we need both and how we sift variables between them.)

Yeah, this split is adding to the confusion indeed. I think that we should make it a goal to unify those going forward.
...
I've been wondering for a while whether we're operating at the wrong level here. Both repo_settings and repo_config_values indicates that we're operating in the context of a repository, but as you mention that may not even be the case.

I don't think the approach is inherently flawed though. From my point of view, the best way forward is to merge those two and then generalize them into something like git_config_values or git_settings, depending on which of both variants we want to retain. We would then have two levels:

- One on the repository level as we have it today.

- One truly global variable, because that stuff in fact _is_ global.

We'd then adapt repo_config_values() so that it knows to populate either of those variables depending on whether or not the user passes a valid repository, and returns a constant pointer to the respective structure. Callers MUST NOT modify that structure -- if they want to, they'll have to make a copy and pass it down the calling stack.

The last part about not modifying that structure could be quite a bit painful though, as it would mean that we might have to adapt call chains to pass down a struct git_config_values instead of a struct repository. But arguably, that's the right thing to do anyway for at least some subsystems that are independent of repositories.

Quoted 4 lines

As you say though, none of this is really related to this patch series at hand, and I don't think we need to resolve this discussion before we can merge it. I just want to make sure that we have a plan for how to get rid of the_repository instead of only shuffling stuff around.

Even after we punted on this and stopped the side discussion in early August, there still are topics that want to move things into repo_settings, or even worse, repo_config_values. Until we have a plan to fix repo_config_values() to take an arbitrary repository, should we freeze this part of the code, perhaps like this?

----- >8 -----
Subject: repo_config_values: freeze the structure for now

The configuration and settings per repository is currently in an unfortunate mess. Especially given that repo_config_values() cannot work on any repository instance other than the_repository, moving more members into this structure is unwelcome.

---
 environment.h | 7 +++++++
 1 file changed, 7 insertions(+)
environment.h +7 -0
index 9e175e410e..03b1b35d6a 100644
@@ -124,6 +124,13 @@ enum object_creation_mode {
 	OBJECT_CREATION_USES_RENAMES = 1
 };
 
+/*
+ * See https://lore.kernel.org/git/[email protected]/ and later
+ * discussions.  Until we figure out what to do with this and
+ * repo_settings, and allow them to be truly usable on a repository
+ * instance that is not the_repository, do *NOT* add any new members
+ * to this structure.
+ */
 struct repo_config_values {
 	/* section "core" config values */
 	char *attributes_file;
#28 lore raw

Re: Re*: [PATCH v3 0/3] environment: clean up repository config handling

Quoted 52 lines · Junio C Hamano
Quoted 45 lines · Patrick Steinhardt
Quoted 6 lines · Junio C Hamano
Quoted 4 lines · Patrick Steinhardt

On Fri, Aug 07, 2026 at 04:59:29PM +0800, Tian Yuchen wrote:
Somewhat unrelated to this patch series, but I was wondering whether you plan to drop the limitation in repo_config_values() that requires that the passed-in repository is the_repository. ...

Quoted 7 lines

Hmph, that is an interesting point. What is our plan to really enable the use of repository instances other than 'the_repository' here? They of course need to be initialized with repo_init(), but is that enough to sensibly use the embedded 'repo_settings' and 'repo_config_values' structures? (By the way, it is not entirely clear to me why we need both and how we sift variables between them.)

Yeah, this split is adding to the confusion indeed. I think that we should make it a goal to unify those going forward.
...
I've been wondering for a while whether we're operating at the wrong level here. Both repo_settings and repo_config_values indicates that we're operating in the context of a repository, but as you mention that may not even be the case.

I don't think the approach is inherently flawed though. From my point of view, the best way forward is to merge those two and then generalize them into something like git_config_values or git_settings, depending on which of both variants we want to retain. We would then have two levels:

- One on the repository level as we have it today.

- One truly global variable, because that stuff in fact _is_ global.

We'd then adapt repo_config_values() so that it knows to populate either of those variables depending on whether or not the user passes a valid repository, and returns a constant pointer to the respective structure. Callers MUST NOT modify that structure -- if they want to, they'll have to make a copy and pass it down the calling stack.

The last part about not modifying that structure could be quite a bit painful though, as it would mean that we might have to adapt call chains to pass down a struct git_config_values instead of a struct repository. But arguably, that's the right thing to do anyway for at least some subsystems that are independent of repositories.

Quoted 4 lines

As you say though, none of this is really related to this patch series at hand, and I don't think we need to resolve this discussion before we can merge it. I just want to make sure that we have a plan for how to get rid of the_repository instead of only shuffling stuff around.

Quoted 5 lines

Even after we punted on this and stopped the side discussion in early August, there still are topics that want to move things into repo_settings, or even worse, repo_config_values. Until we have a plan to fix repo_config_values() to take an arbitrary repository, should we freeze this part of the code, perhaps like this?

Ah, thanks for tackling this. I had it on my todo list but didn't get to it yet.

Quoted 21 lines

----- >8 -----
Subject: repo_config_values: freeze the structure for now

The configuration and settings per repository is currently in an unfortunate mess. Especially given that repo_config_values() cannot work on any repository instance other than the_repository, moving more members into this structure is unwelcome.

---
 environment.h | 7 +++++++
 1 file changed, 7 insertions(+)
environment.h +0 -0
index 9e175e410e..03b1b35d6a 100644
@@ -124,6 +124,13 @@ enum object_creation_mode {
 	OBJECT_CREATION_USES_RENAMES = 1
 };
 

environment.h enum object_creation_mode { · line 136

Quoted 10 lines
environment.h +7 -0
+/*
+ * See https://lore.kernel.org/git/[email protected]/ and later
+ * discussions.  Until we figure out what to do with this and
+ * repo_settings, and allow them to be truly usable on a repository
+ * instance that is not the_repository, do *NOT* add any new members
+ * to this structure.
+ */
 struct repo_config_values {
 	/* section "core" config values */
 	char *attributes_file;

Thanks, this looks good to me as-is. I'm very much in favor of merging this so that we don't need to push back so often, and so that newcomers don't waste their time and get frustrated because of it.

Patrick