builtin/fmt-merge-msg: remove dependency on global variables and 'the_repository'
cover builtin/fmt-merge-msg: remove dependency on global variables and 'the_repository'
The aim of this patch series is to remove the definition '#define USE_THE_REPOSITORY_VARIABLE' from "builtin/fmt-merge-msg.c" by removing global variable 'merge_log_config' and the global 'the_repository'
This patch series contains two patches:
1 - Remove the global varaible 'merge_log_config' and add a function 'adjust_shortlog_len()'
in fmt-merge-msg.{c,h} to replicate the variable's usage.
2 - Remove the dependency of 'the_repository' in "builtin/fmt-merge-msg.c", allowing the removal
of the definition '#define USE_THE_REPOSITORY_VARIABLE'. Also add a test to make sure that "git fmt-merge-msg -h" can be called with repository being NULL.
Ayush Chandekar (2):
environment: remove the global variable 'merge_log_config' builtin/fmt-merge-msg: stop depending on 'the_repository'
builtin/fmt-merge-msg.c | 9 ++++----- builtin/merge.c | 4 ++-- environment.c | 2 -- fmt-merge-msg.c | 30 ++++++++++++++++++++++-------- fmt-merge-msg.h | 3 ++- t/t1517-outside-repo.sh | 7 +++++++ 6 files changed, 37 insertions(+), 18 deletions(-)
Signature · 1 line
2.49.0
PATCH 1/2 environment: remove the global variable 'merge_log_config'
The global variable 'merge_log_config', set via the "merge.log" or "merge.summary" settings, is only used in 'cmd_fmt_merge_msg()' and 'cmd_merge()' to adjust the 'shortlog_len' variable.
Remove 'merge_log_config' and introduce a function 'adjust_shortlog_len()' in fmt-merge-msg.c to handle the 'shortlog_len' variable.
This change is part of an ongoing effort to eliminate global variables, improve modularity and help libify the codebase.
--- builtin/fmt-merge-msg.c | 4 ++-- builtin/merge.c | 4 ++-- environment.c | 2 -- fmt-merge-msg.c | 30 ++++++++++++++++++++++-------- fmt-merge-msg.h | 3 ++- 5 files changed, 28 insertions(+), 15 deletions(-)
builtin/fmt-merge-msg.c +2 -2
@@ -58,8 +58,8 @@ int cmd_fmt_merge_msg(int argc, 0);if (argc >0 )usage_with_options (fmt_merge_msg_usage, options); -if (shortlog_len <0 ) - shortlog_len = (merge_log_config >0 ) ? merge_log_config :0 ; + +adjust_shortlog_len (the_repository, &shortlog_len);if (inpath &&strcmp (inpath,"-" )) { in =fopen (inpath,"r" );
builtin/merge.c +2 -2
@@ -1403,8 +1403,8 @@ int cmd_merge(int argc,parse_branch_merge_options (branch_mergeoptions); argc =parse_options (argc, argv, prefix, builtin_merge_options, builtin_merge_usage,0 ); -if (shortlog_len <0 ) - shortlog_len = (merge_log_config >0 ) ? merge_log_config :0 ; + +adjust_shortlog_len (the_repository, &shortlog_len);if (verbosity <0 && show_progress ==-1 ) show_progress =0 ;
environment.c +0 -2
@@ -20,7 +20,6 @@#include "repository.h" #include "config.h" #include "refs.h" -#include "fmt-merge-msg.h" #include "commit.h" #include "strvec.h" #include "path.h"
@@ -66,7 +65,6 @@ int grafts_keep_true_parents;int core_apply_sparse_checkout;int core_sparse_checkout_cone;int sparse_expect_files_outside_of_patterns; -int merge_log_config =-1 ;int precomposed_unicode =-1 ;/* see probe_utf8_pathname_composition() */ unsigned long pack_size_limit_cfg; int max_allowed_tree_depth =
fmt-merge-msg.c +22 -8
@@ -26,14 +26,7 @@ static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb) { -if (!strcmp (key,"merge.log" ) || !strcmp (key,"merge.summary" )) { -int is_bool; - merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); -if (!is_bool && merge_log_config <0 ) -return error ("%s: negative length %s" , key, value); -if (is_bool && merge_log_config) - merge_log_config =DEFAULT_MERGE_LOG_LEN ; - }else if (!strcmp (key,"merge.branchdesc" )) { +if (!strcmp (key,"merge.branchdesc" )) { use_branch_desc =git_config_bool (key, value); }else if (!strcmp (key,"merge.suppressdest" )) {if (!value)
@@ -645,6 +638,27 @@ static void find_merge_parents(struct merge_parents *result, result->nr = j; } +void adjust_shortlog_len (struct repository *r,int *shortlog_len) +{ +const char *keys[] = {"merge.log" ,"merge.summary" ,NULL }; + +if (*shortlog_len >=0 ) +return ; + +for (const char **key = keys; *key; ++key) { +int is_bool, value; +if (!repo_config_get_bool_or_int (r, *key, &is_bool, &value)) { +if (!is_bool && value <0 ) { +error ("%s: negative length %d" , *key, value); +return ; + } + *shortlog_len = (is_bool && value) ?DEFAULT_MERGE_LOG_LEN : value; +return ; + } + } + + *shortlog_len =0 ; +}int fmt_merge_msg (struct strbuf *in,struct strbuf *out,struct fmt_merge_msg_opts *opts)
fmt-merge-msg.h +2 -1
@@ -2,6 +2,7 @@#define FMT_MERGE_MSG_H #include "strbuf.h" +#include "repository.h" #define DEFAULT_MERGE_LOG_LEN 20
@@ -12,9 +13,9 @@ struct fmt_merge_msg_opts {const char *into_name; }; -extern int merge_log_config;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb); +void adjust_shortlog_len (struct repository *r,int *shortlog_len);int fmt_merge_msg (struct strbuf *in,struct strbuf *out,struct fmt_merge_msg_opts *);
Signature · 1 line
2.49.0
Re: [GSOC PATCH 1/2] environment: remove the global variable 'merge_log_config'
Quoted 10 lines · Ayush Chandekar
The global variable '
merge_log_config', set via the "merge.log" or "merge.summary" settings, is only used in 'cmd_fmt_merge_msg()' and 'cmd_merge()' to adjust the 'shortlog_len' variable.Remove '
merge_log_config' and introduce a function 'adjust_shortlog_len()' infmt-merge-msg.cto handle the 'shortlog_len' variable.This change is part of an ongoing effort to eliminate global variables, improve modularity and help libify the codebase.
And the downsides of this change are...?
One obvious behaviour change I can see can happen when you have an invalid value set to merge.summary and run the command with command line override with the "--log" option. In the current code, the config callback barfs when it notices an invalid merge.summary setting, even though it won't be used because the valid value given via the "--log" option would override it. In the updated code, adjust_shortlog_len() would short-circuit and does not even bother reading from the configuration, so the user will not be notified of a broken configuration.
It is not immediately obvious if this particular behaviour change is a regression or an improvement, but it probably deserves to be noted somewhere to help future developers what our thinking was.
Quoted 66 lines
fmt-merge-msg.c +22
-8@@ -26,14 +26,7 @@ static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb) { -if (!strcmp (key,"merge.log" ) || !strcmp (key,"merge.summary" )) { -int is_bool; - merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); -if (!is_bool && merge_log_config <0 ) -return error ("%s: negative length %s" , key, value); -if (is_bool && merge_log_config) - merge_log_config =DEFAULT_MERGE_LOG_LEN ; - }else if (!strcmp (key,"merge.branchdesc" )) { +if (!strcmp (key,"merge.branchdesc" )) { use_branch_desc =git_config_bool (key, value); }else if (!strcmp (key,"merge.suppressdest" )) {if (!value)@@ -645,6 +638,27 @@ static void find_merge_parents(struct merge_parents *result, result->nr = j; } +void adjust_shortlog_len (struct repository *r,int *shortlog_len) +{ +const char *keys[] = {"merge.log" ,"merge.summary" ,NULL }; + +if (*shortlog_len >=0 ) +return ; + +for (const char **key = keys; *key; ++key) { +int is_bool, value; +if (!repo_config_get_bool_or_int (r, *key, &is_bool, &value)) { +if (!is_bool && value <0 ) { +error ("%s: negative length %d" , *key, value); +return ; + } + *shortlog_len = (is_bool && value) ?DEFAULT_MERGE_LOG_LEN : value; +return ; + } + } + + *shortlog_len =0 ; +}int fmt_merge_msg (struct strbuf *in,struct strbuf *out,struct fmt_merge_msg_opts *opts)fmt-merge-msg.h +2
-1@@ -2,6 +2,7 @@#define FMT_MERGE_MSG_H #include "strbuf.h" +#include "repository.h" #define DEFAULT_MERGE_LOG_LEN 20@@ -12,9 +13,9 @@ struct fmt_merge_msg_opts {const char *into_name; }; -extern int merge_log_config;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb); +void adjust_shortlog_len (struct repository *r,int *shortlog_len);int fmt_merge_msg (struct strbuf *in,struct strbuf *out,struct fmt_merge_msg_opts *);
Re: [GSOC PATCH 1/2] environment: remove the global variable 'merge_log_config'
Hi Junio,
Quoted 26 lines · Junio C Hamano
Quoted 10 lines · Ayush Chandekar
The global variable '
merge_log_config', set via the "merge.log" or "merge.summary" settings, is only used in 'cmd_fmt_merge_msg()' and 'cmd_merge()' to adjust the 'shortlog_len' variable.Remove '
merge_log_config' and introduce a function 'adjust_shortlog_len()' infmt-merge-msg.cto handle the 'shortlog_len' variable.This change is part of an ongoing effort to eliminate global variables, improve modularity and help libify the codebase.
And the downsides of this change are...?
One obvious behaviour change I can see can happen when you have an invalid value set to merge.summary and run the command with command line override with the "--log" option. In the current code, the config callback barfs when it notices an invalid merge.summary setting, even though it won't be used because the valid value given via the "--log" option would override it. In the updated code,
adjust_shortlog_len()would short-circuit and does not even bother reading from the configuration, so the user will not be notified of a broken configuration.
Quoted 3 lines
It is not immediately obvious if this particular behaviour change is a regression or an improvement, but it probably deserves to be noted somewhere to help future developers what our thinking was.
Oh right, I did not mention this in the commit message. I am not sure if this behaviour is good or not.
Technically, if the user wants to use the "--log" option, they would not care about the config. Whereas, if the user wants to use the config, they would be notified in case of an invalid one.
I will mention this in the commit message, but do you think this behaviour is fine?
Thanks
Ayush
Re: [GSOC PATCH 1/2] environment: remove the global variable 'merge_log_config'
Quoted 2 lines · Ayush Chandekar
I will mention this in the commit message, but do you think this behaviour is fine?
I cannot answer that one immediately after saying that I do not know if this is a regression or an improvement.
If you still pushed me to answer it immediately, you'd get my default position, a conservative "any changes in behaviour caused by an internal code clean-up is bad and is a serious regression" ;-).
Re: [GSOC PATCH 1/2] environment: remove the global variable 'merge_log_config'
In reply to Ayush ChandekarHi Ayush
fmt-merge-msg.c static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP; · line 36
Quoted 13 lines · Ayush Chandekar
fmt-merge-msg.c +0
-8@@ -26,14 +26,7 @@ static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb) { -if (!strcmp (key,"merge.log" ) || !strcmp (key,"merge.summary" )) { -int is_bool; - merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); -if (!is_bool && merge_log_config <0 ) -return error ("%s: negative length %s" , key, value); -if (is_bool && merge_log_config) - merge_log_config =DEFAULT_MERGE_LOG_LEN ; - }else if (!strcmp (key,"merge.branchdesc" )) {
In the old code if both "merge.log" and "merge.summary" are set in the config file the last one wins
Quoted 6 lines
fmt-merge-msg.c +6
-0+void adjust_shortlog_len (struct repository *r,int *shortlog_len) +{ +const char *keys[] = {"merge.log" ,"merge.summary" ,NULL }; + +if (*shortlog_len >=0 ) +return ;
fmt-merge-msg.c static void find_merge_parents(struct merge_parents *result, · line 656
Quoted 10 lines
fmt-merge-msg.c +10
-0+ +for (const char **key = keys; *key; ++key) { +int is_bool, value; +if (!repo_config_get_bool_or_int (r, *key, &is_bool, &value)) { +if (!is_bool && value <0 ) { +error ("%s: negative length %d" , *key, value); +return ; + } + *shortlog_len = (is_bool && value) ?DEFAULT_MERGE_LOG_LEN : value; +return ;
In the new code "merge.log" is always used in preference to "merge.summary" even if "merge.summary" appears later in the config file. When you have two keys setting the same variable I think the only way to preserve the last one wins behavior is to keep using a callback that updates the value as the config files are parsed.
Thanks
Phillip
Re: [GSOC PATCH 1/2] environment: remove the global variable 'merge_log_config'
Hi Phillip,
Quoted 35 lines · Phillip Wood
Hi Ayush
fmt-merge-msg.c static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP; · line 36
Quoted 13 lines · Ayush Chandekar
fmt-merge-msg.c +0
-8@@ -26,14 +26,7 @@ static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb) { -if (!strcmp (key,"merge.log" ) || !strcmp (key,"merge.summary" )) { -int is_bool; - merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); -if (!is_bool && merge_log_config <0 ) -return error ("%s: negative length %s" , key, value); -if (is_bool && merge_log_config) - merge_log_config =DEFAULT_MERGE_LOG_LEN ; - }else if (!strcmp (key,"merge.branchdesc" )) {In the old code if both "merge.log" and "merge.summary" are set in the config file the last one wins
Quoted 14 lines
fmt-merge-msg.c +14
-0+void adjust_shortlog_len (struct repository *r,int *shortlog_len) +{ +const char *keys[] = {"merge.log" ,"merge.summary" ,NULL }; + +if (*shortlog_len >=0 ) +return ; + +for (const char **key = keys; *key; ++key) { +int is_bool, value; +if (!repo_config_get_bool_or_int (r, *key, &is_bool, &value)) { +if (!is_bool && value <0 ) { +error ("%s: negative length %d" , *key, value); +return ; + }
Quoted 9 lines
fmt-merge-msg.c static void find_merge_parents(struct merge_parents *result, · line 656
Quoted 2 lines
fmt-merge-msg.c +2
-0+ *shortlog_len = (is_bool && value) ?DEFAULT_MERGE_LOG_LEN : value; +return ;In the new code "merge.log" is always used in preference to "merge.summary" even if "merge.summary" appears later in the config file. When you have two keys setting the same variable I think the only way to preserve the last one wins behavior is to keep using a callback that updates the value as the config files are parsed.
Sorry for not mentioning this in the commit message.
I had looked at the documentation which says:
Documentation/git-fmt-merge-msg.adoc
merge.summary::
Synonym to merge.log; this is deprecated and will be removed in the future.
So I thought that I should give precedence to "merge.log" as "merge.summary" is deprecated.
Quoted 4 lines
Thanks
Phillip
Thanks
Ayush
Re: [GSOC PATCH 1/2] environment: remove the global variable 'merge_log_config'
Hi Ayush
Quoted 51 lines · Ayush Chandekar
Quoted 35 lines · Phillip Wood
Hi Ayush
fmt-merge-msg.c static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP; · line 36
Quoted 13 lines · Ayush Chandekar
fmt-merge-msg.c +0
-8@@ -26,14 +26,7 @@ static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb) { -if (!strcmp (key,"merge.log" ) || !strcmp (key,"merge.summary" )) { -int is_bool; - merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); -if (!is_bool && merge_log_config <0 ) -return error ("%s: negative length %s" , key, value); -if (is_bool && merge_log_config) - merge_log_config =DEFAULT_MERGE_LOG_LEN ; - }else if (!strcmp (key,"merge.branchdesc" )) {In the old code if both "merge.log" and "merge.summary" are set in the config file the last one wins
Quoted 14 lines
fmt-merge-msg.c +14
-0+void adjust_shortlog_len (struct repository *r,int *shortlog_len) +{ +const char *keys[] = {"merge.log" ,"merge.summary" ,NULL }; + +if (*shortlog_len >=0 ) +return ; + +for (const char **key = keys; *key; ++key) { +int is_bool, value; +if (!repo_config_get_bool_or_int (r, *key, &is_bool, &value)) { +if (!is_bool && value <0 ) { +error ("%s: negative length %d" , *key, value); +return ; + }Quoted 9 lines
fmt-merge-msg.c static void find_merge_parents(struct merge_parents *result, · line 656
Quoted 2 lines
fmt-merge-msg.c +2
-0+ *shortlog_len = (is_bool && value) ?DEFAULT_MERGE_LOG_LEN : value; +return ;In the new code "merge.log" is always used in preference to "merge.summary" even if "merge.summary" appears later in the config file. When you have two keys setting the same variable I think the only way to preserve the last one wins behavior is to keep using a callback that updates the value as the config files are parsed.
Sorry for not mentioning this in the commit message.
I had looked at the documentation which says:
Quoted 7 lines
Documentation/git-fmt-merge-msg.adoc
merge.summary::
Synonym tomerge.log; this is deprecated and will be removed in the future.So I thought that I should give precedence to "merge.log" as "merge.summary" is deprecated.
If it is deprecated we still need to support it until it is removed (maybe we should do that in Git 3.0?). We cannot change the behavior just because a setting is deprecated.
Thanks
Phillip
Quoted 8 lines
Quoted 4 lines
Thanks
Phillip
Thanks
Ayush
PATCH 2/2 builtin/fmt-merge-msg: stop depending on 'the_repository'
In reply to Ayush ChandekarRefactor builtin/fmt-merge-msg.c to remove the dependancy on the global 'the_repository'. Replace all the occurrences of 'the_repository' with 'repo', where 'repo' is a pointer to 'struct repository' passed to the function 'cmd_fmt_merge_msg()' and thus remove the definition '#define USE_THE_REPOSITORY_VARIABLE'. Also, add a test to make sure that "git fmt-merge-msg -h" can be called outside a repository.
--- builtin/fmt-merge-msg.c | 7 +++---- t/t1517-outside-repo.sh | 7 +++++++ 2 files changed, 10 insertions(+), 4 deletions(-)
builtin/fmt-merge-msg.c +3 -4
@@ -1,4 +1,3 @@ -#define USE_THE_REPOSITORY_VARIABLE #include "builtin.h" #include "config.h" #include "fmt-merge-msg.h"
@@ -13,7 +12,7 @@ static const char * const fmt_merge_msg_usage[] = {int cmd_fmt_merge_msg (int argc,const char **argv,const char *prefix, - struct repository *repo UNUSED) +struct repository *repo) {char *inpath =NULL ;const char *message =NULL ;
@@ -53,13 +52,13 @@ int cmd_fmt_merge_msg(int argc,int ret;struct fmt_merge_msg_opts opts; -git_config (fmt_merge_msg_config,NULL ); argc =parse_options (argc, argv, prefix, options, fmt_merge_msg_usage,0 );if (argc >0 )usage_with_options (fmt_merge_msg_usage, options); +repo_config (repo, fmt_merge_msg_config,NULL ); -adjust_shortlog_len (the_repository, &shortlog_len); +adjust_shortlog_len (repo, &shortlog_len);if (inpath &&strcmp (inpath,"-" )) { in =fopen (inpath,"r" );
t/t1517-outside-repo.sh +7 -0
@@ -121,4 +121,11 @@ test_expect_success 'prune does not crash with -h' ' test_grep "[Uu]sage: git prune " usage ' +test_expect_success 'fmt-merge-msg does not crash with -h' ' + test_expect_code 129 git fmt-merge-msg -h >usage && + test_grep "[Uu]sage: git fmt-merge-msg " usage && + test_expect_code 129 nongit git fmt-merge-msg -h >usage && + test_grep "[Uu]sage: git fmt-merge-msg " usage +' + test_done
Signature · 1 line
2.49.0
Re: [GSOC PATCH 2/2] builtin/fmt-merge-msg: stop depending on 'the_repository'
Quoted 6 lines · Ayush Chandekar
Refactor
builtin/fmt-merge-msg.cto remove the dependancy on the global 'the_repository'. Replace all the occurrences of 'the_repository' with 'repo', where 'repo' is a pointer to 'struct repository' passed to the function 'cmd_fmt_merge_msg()' and thus remove the definition '#defineUSE_THE_REPOSITORY_VARIABLE'. Also, add a test to make sure that "git fmt-merge-msg -h" can be called outside a repository.
This also moves the call to git_config()/repo_config() after parse_options().
It generally is a bad idea to read command line options first and then read the configuration (it is a bug if such a flow causes values from configuration to overwrite values from command line). THe current set of options and configuration variables may not overlap, in which case such a questionable arrangement happen to be without bug right now, but it would prevent future developers from adding new options and configuration variables and make them interact with each other in the most natural way.
In any case, the reason for this change of the order between config and parse-options is not explained at all in the proposed log message.
Quoted 58 lines
Mentored-by Christian Couder <[email protected]>Mentored-by Ghanshyam Thakkar <[email protected]>Signed-off-by Ayush Chandekar <[email protected]>--- builtin/fmt-merge-msg.c | 7 +++---- t/t1517-outside-repo.sh | 7 +++++++ 2 files changed, 10 insertions(+), 4 deletions(-)builtin/fmt-merge-msg.c +3
-4@@ -1,4 +1,3 @@ -#define USE_THE_REPOSITORY_VARIABLE #include "builtin.h" #include "config.h" #include "fmt-merge-msg.h" @@ -13,7 +12,7 @@ static const char * const fmt_merge_msg_usage[] = {int cmd_fmt_merge_msg (int argc,const char **argv,const char *prefix, - struct repository *repo UNUSED) +struct repository *repo) {char *inpath =NULL ;const char *message =NULL ;@@ -53,13 +52,13 @@ int cmd_fmt_merge_msg(int argc,int ret;struct fmt_merge_msg_opts opts; -git_config (fmt_merge_msg_config,NULL ); argc =parse_options (argc, argv, prefix, options, fmt_merge_msg_usage,0 );if (argc >0 )usage_with_options (fmt_merge_msg_usage, options); +repo_config (repo, fmt_merge_msg_config,NULL ); -adjust_shortlog_len (the_repository, &shortlog_len); +adjust_shortlog_len (repo, &shortlog_len);if (inpath &&strcmp (inpath,"-" )) { in =fopen (inpath,"r" );t/t1517-outside-repo.sh +7
-0@@ -121,4 +121,11 @@ test_expect_success 'prune does not crash with -h' ' test_grep "[Uu]sage: git prune " usage ' +test_expect_success 'fmt-merge-msg does not crash with -h' ' + test_expect_code 129 git fmt-merge-msg -h >usage && + test_grep "[Uu]sage: git fmt-merge-msg " usage && + test_expect_code 129 nongit git fmt-merge-msg -h >usage && + test_grep "[Uu]sage: git fmt-merge-msg " usage +' + test_done
Re: [GSOC PATCH 2/2] builtin/fmt-merge-msg: stop depending on 'the_repository'
Quoted 13 lines · Junio C Hamano
Quoted 6 lines · Ayush Chandekar
Refactor
builtin/fmt-merge-msg.cto remove the dependancy on the global 'the_repository'. Replace all the occurrences of 'the_repository' with 'repo', where 'repo' is a pointer to 'struct repository' passed to the function 'cmd_fmt_merge_msg()' and thus remove the definition '#defineUSE_THE_REPOSITORY_VARIABLE'. Also, add a test to make sure that "git fmt-merge-msg -h" can be called outside a repository.This also moves the call to git_config()/repo_config() after
parse_options().
Quoted 9 lines
It generally is a bad idea to read command line options first and then read the configuration (it is a bug if such a flow causes values from configuration to overwrite values from command line). THe current set of options and configuration variables may not overlap, in which case such a questionable arrangement happen to be without bug right now, but it would prevent future developers from adding new options and configuration variables and make them interact with each other in the most natural way.
I understand it, but how do we tackle if NULL repository is passed.
Quoted 4 lines
In any case, the reason for this change of the order between config and parse-options is not explained at all in the proposed log message.
Apologies, I will mention it.
Re: [GSOC PATCH 2/2] builtin/fmt-merge-msg: stop depending on 'the_repository'
Quoted 10 lines · Ayush Chandekar
Quoted 8 lines
It generally is a bad idea to read command line options first and then read the configuration (it is a bug if such a flow causes values from configuration to overwrite values from command line). THe current set of options and configuration variables may not overlap, in which case such a questionable arrangement happen to be without bug right now, but it would prevent future developers from adding new options and configuration variables and make them interact with each other in the most natural way.
I understand it, but how do we tackle if NULL repository is passed.
Perhaps you want to study the problem space and related past changes before going forward. The first place to look at is what happens when you call repo_config(), outside a repository or a working tree and repository is NULL.
f29f1990 (config: teach repo_config to allow repo to be NULL, 2025-03-08) and what it calls "the following commits" may be illuminating.
Just an update, I'm still working on this patch series.
Thanks,
Ayush
v2 cover builtin/fmt-merge-msg: remove dependency on global variables and 'the_repository'
In reply to Ayush ChandekarThe aim of this patch series is to remove the definition '#define USE_THE_REPOSITORY_VARIABLE' from "builtin/fmt-merge-msg.c" by removing global variable 'merge_log_config' and the global 'the_repository'
This patch series contains two patches:
1 - Remove the global varaible 'merge_log_config' and localize it in
'cmd_fmt_merge_msg()' and 'cmd_merge()'. Set its value by passing it in 'fmt_merge_msg_config()' by passing its pointer to the function via the callback parameter.
2 - Remove the dependency of 'the_repository' in "builtin/fmt-merge-msg.c", allowing the removal
of the definition '#define USE_THE_REPOSITORY_VARIABLE'. Also add a test to make sure that "git fmt-merge-msg -h" can be called with repository being NULL.
Thanks to Junio and Phillip for reviewing my patch series and Christian for mentoring me!
Ayush Chandekar (2):
environment: remove the global variable 'merge_log_config' builtin/fmt-merge-msg: stop depending on 'the_repository'
builtin/fmt-merge-msg.c | 6 +++--- builtin/merge.c | 3 ++- environment.c | 1 - fmt-merge-msg.c | 10 ++++++---- fmt-merge-msg.h | 1 - t/t1517-outside-repo.sh | 7 +++++++ 6 files changed, 18 insertions(+), 10 deletions(-)
Range-diff against v1:
1: c82620a1f5 < -: ---------- environment: remove the global variable 'merge_log_config' -: ---------- > 1: 3aa014ed46 environment: remove the global variable 'merge_log_config' 2: 04d6f682a6 ! 2: 8e55516cda builtin/fmt-merge-msg: stop depending on 'the_repository'
@@ Commit message
builtin/fmt-merge-msg: stop depending on 'the_repository'
Refactor builtin/fmt-merge-msg.c to remove the dependancy on the global
- 'the_repository'. Replace all the occurrences of 'the_repository' with
- 'repo', where 'repo' is a pointer to 'struct repository' passed to the
- function 'cmd_fmt_merge_msg()' and thus remove the definition '#define
- USE_THE_REPOSITORY_VARIABLE'. Also, add a test to make sure that "git
- fmt-merge-msg -h" can be called outside a repository.
+ 'the_repository'. Remove the 'UNUSED' macro from the 'struct repository'
+ parameter and replace 'git_config()' with 'repo_config()' so that
+ configuration is read from the passed repository. Also, add a test to
+ make sure that "git fmt-merge-msg -h" can be called outside a
+ repository.
@@ builtin/fmt-merge-msg.c: int cmd_fmt_merge_msg(int argc,
int ret;
struct fmt_merge_msg_opts opts;
-- git_config(fmt_merge_msg_config, NULL);
+- git_config(fmt_merge_msg_config, &merge_log_config);
++ repo_config(repo, fmt_merge_msg_config, &merge_log_config);
argc = parse_options(argc, argv, prefix, options, fmt_merge_msg_usage,
0);
if (argc > 0)
- usage_with_options(fmt_merge_msg_usage, options);
-+ repo_config(repo, fmt_merge_msg_config, NULL);
-
-- adjust_shortlog_len(the_repository, &shortlog_len);
-+ adjust_shortlog_len(repo, &shortlog_len);
-
- if (inpath && strcmp(inpath, "-")) {
- in = fopen(inpath, "r");
## t/t1517-outside-repo.sh ##
@@ t/t1517-outside-repo.sh: test_expect_success 'prune does not crash with -h' '
Signature · 1 line
2.49.0
PATCH v2 1/2 environment: remove the global variable 'merge_log_config'
The global variable 'merge_log_config', set via the "merge.log" or "merge.summary" settings, is only used in 'cmd_fmt_merge_msg()' and 'cmd_merge()' to adjust the 'shortlog_len' variable.
Remove 'merge_log_config' globally and localize it in 'cmd_fmt_merge_msg()' and 'cmd_merge()'. Set its value by passing it in 'fmt_merge_msg_config()' by passing its pointer to the function via the callback parameter.
This change is part of an ongoing effort to eliminate global variables, improve modularity and help libify the codebase.
--- builtin/fmt-merge-msg.c | 3 ++- builtin/merge.c | 3 ++- environment.c | 1 - fmt-merge-msg.c | 10 ++++++---- fmt-merge-msg.h | 1 - 5 files changed, 10 insertions(+), 8 deletions(-)
builtin/fmt-merge-msg.c +2 -1
@@ -19,6 +19,7 @@ int cmd_fmt_merge_msg(int argc,const char *message =NULL ;char *into_name =NULL ;int shortlog_len =-1 ; +int merge_log_config =-1 ;struct option options[] = { { .type =OPTION_INTEGER ,
@@ -53,7 +54,7 @@ int cmd_fmt_merge_msg(int argc,int ret;struct fmt_merge_msg_opts opts; -git_config (fmt_merge_msg_config,NULL ); +git_config (fmt_merge_msg_config, &merge_log_config); argc =parse_options (argc, argv, prefix, options, fmt_merge_msg_usage,0 );if (argc >0 )
builtin/merge.c +2 -1
@@ -1374,6 +1374,7 @@ int cmd_merge(int argc,struct commit_list *remoteheads =NULL , *p;void *branch_to_free;int orig_argc = argc; +int merge_log_config =-1 ;show_usage_with_options_if_asked (argc, argv, builtin_merge_usage, builtin_merge_options);
@@ -1392,7 +1393,7 @@ int cmd_merge(int argc,skip_prefix (branch,"refs/heads/" , &branch);init_diff_ui_defaults (); -git_config (git_merge_config,NULL ); +git_config (git_merge_config, &merge_log_config);if (!branch ||is_null_oid (&head_oid)) head_commit =NULL ;
environment.c +0 -1
@@ -66,7 +66,6 @@ int grafts_keep_true_parents;int core_apply_sparse_checkout;int core_sparse_checkout_cone;int sparse_expect_files_outside_of_patterns; -int merge_log_config =-1 ;int precomposed_unicode =-1 ;/* see probe_utf8_pathname_composition() */ unsigned long pack_size_limit_cfg; int max_allowed_tree_depth =
fmt-merge-msg.c +6 -4
@@ -26,13 +26,15 @@ static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb) { +int *merge_log_config = cb; +if (!strcmp (key,"merge.log" ) || !strcmp (key,"merge.summary" )) {int is_bool; - merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); -if (!is_bool && merge_log_config <0 ) + *merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); +if (!is_bool && *merge_log_config <0 )return error ("%s: negative length %s" , key, value); -if (is_bool && merge_log_config) - merge_log_config =DEFAULT_MERGE_LOG_LEN ; +if (is_bool && *merge_log_config) + *merge_log_config =DEFAULT_MERGE_LOG_LEN ; }else if (!strcmp (key,"merge.branchdesc" )) { use_branch_desc =git_config_bool (key, value); }else if (!strcmp (key,"merge.suppressdest" )) {
fmt-merge-msg.h +0 -1
@@ -12,7 +12,6 @@ struct fmt_merge_msg_opts {const char *into_name; }; -extern int merge_log_config;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb); int fmt_merge_msg(struct strbuf *in, struct strbuf *out,
Signature · 1 line
2.49.0
Re: [GSOC PATCH v2 1/2] environment: remove the global variable 'merge_log_config'
Hi Ayush
Quoted 8 lines · Ayush Chandekar
The global variable '
merge_log_config', set via the "merge.log" or "merge.summary" settings, is only used in 'cmd_fmt_merge_msg()' and 'cmd_merge()' to adjust the 'shortlog_len' variable.Remove '
merge_log_config' globally and localize it in 'cmd_fmt_merge_msg()' and 'cmd_merge()'. Set its value by passing it in 'fmt_merge_msg_config()' by passing its pointer to the function via the callback parameter.
This looks like a good solution
Thanks
Phillip
Quoted 104 lines
This change is part of an ongoing effort to eliminate global variables, improve modularity and help libify the codebase.
Mentored-by Christian Couder <[email protected]>Mentored-by Ghanshyam Thakkar <[email protected]>Signed-off-by Ayush Chandekar <[email protected]>--- builtin/fmt-merge-msg.c | 3 ++- builtin/merge.c | 3 ++- environment.c | 1 - fmt-merge-msg.c | 10 ++++++---- fmt-merge-msg.h | 1 - 5 files changed, 10 insertions(+), 8 deletions(-)builtin/fmt-merge-msg.c +2
-1@@ -19,6 +19,7 @@ int cmd_fmt_merge_msg(int argc,const char *message =NULL ;char *into_name =NULL ;int shortlog_len =-1 ; +int merge_log_config =-1 ;struct option options[] = { { .type =OPTION_INTEGER ,@@ -53,7 +54,7 @@ int cmd_fmt_merge_msg(int argc,int ret;struct fmt_merge_msg_opts opts; -git_config (fmt_merge_msg_config,NULL ); +git_config (fmt_merge_msg_config, &merge_log_config); argc =parse_options (argc, argv, prefix, options, fmt_merge_msg_usage,0 );if (argc >0 )builtin/merge.c +2
-1@@ -1374,6 +1374,7 @@ int cmd_merge(int argc,struct commit_list *remoteheads =NULL , *p;void *branch_to_free;int orig_argc = argc; +int merge_log_config =-1 ;show_usage_with_options_if_asked (argc, argv, builtin_merge_usage, builtin_merge_options);@@ -1392,7 +1393,7 @@ int cmd_merge(int argc,skip_prefix (branch,"refs/heads/" , &branch);init_diff_ui_defaults (); -git_config (git_merge_config,NULL ); +git_config (git_merge_config, &merge_log_config);if (!branch ||is_null_oid (&head_oid)) head_commit =NULL ;environment.c +0
-1@@ -66,7 +66,6 @@ int grafts_keep_true_parents;int core_apply_sparse_checkout;int core_sparse_checkout_cone;int sparse_expect_files_outside_of_patterns; -int merge_log_config =-1 ;int precomposed_unicode =-1 ;/* see probe_utf8_pathname_composition() */ unsigned long pack_size_limit_cfg; int max_allowed_tree_depth =fmt-merge-msg.c +6
-4@@ -26,13 +26,15 @@ static struct string_list suppress_dest_patterns = STRING_LIST_INIT_DUP;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb) { +int *merge_log_config = cb; +if (!strcmp (key,"merge.log" ) || !strcmp (key,"merge.summary" )) {int is_bool; - merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); -if (!is_bool && merge_log_config <0 ) + *merge_log_config =git_config_bool_or_int (key, value, ctx->kvi, &is_bool); +if (!is_bool && *merge_log_config <0 )return error ("%s: negative length %s" , key, value); -if (is_bool && merge_log_config) - merge_log_config =DEFAULT_MERGE_LOG_LEN ; +if (is_bool && *merge_log_config) + *merge_log_config =DEFAULT_MERGE_LOG_LEN ; }else if (!strcmp (key,"merge.branchdesc" )) { use_branch_desc =git_config_bool (key, value); }else if (!strcmp (key,"merge.suppressdest" )) {fmt-merge-msg.h +0
-1@@ -12,7 +12,6 @@ struct fmt_merge_msg_opts {const char *into_name; }; -extern int merge_log_config;int fmt_merge_msg_config (const char *key,const char *value,const struct config_context *ctx,void *cb); int fmt_merge_msg(struct strbuf *in, struct strbuf *out,
Re: [GSOC PATCH v2 1/2] environment: remove the global variable 'merge_log_config'
Quoted 11 lines · Phillip Wood
Hi Ayush
Quoted 7 lines · Ayush Chandekar
The global variable '
merge_log_config', set via the "merge.log" or "merge.summary" settings, is only used in 'cmd_fmt_merge_msg()' and 'cmd_merge()' to adjust the 'shortlog_len' variable.
Remove 'merge_log_config' globally and localize it in 'cmd_fmt_merge_msg()' and 'cmd_merge()'. Set its value by passing it in 'fmt_merge_msg_config()' by passing its pointer to the function via the callback parameter.
Quoted 1 line
This looks like a good solution
When fmt_merge_msg_config() needs to read more stuff, the callback parameter may have to be updated, but this will do for now.
Thanks.
Re: [GSOC PATCH v2 1/2] environment: remove the global variable 'merge_log_config'
Quoted 14 lines · Junio C Hamano
Quoted 11 lines · Phillip Wood
Hi Ayush
Quoted 7 lines · Ayush Chandekar
The global variable '
merge_log_config', set via the "merge.log" or "merge.summary" settings, is only used in 'cmd_fmt_merge_msg()' and 'cmd_merge()' to adjust the 'shortlog_len' variable.
Remove 'merge_log_config' globally and localize it in 'cmd_fmt_merge_msg()' and 'cmd_merge()'. Set its value by passing it in 'fmt_merge_msg_config()' by passing its pointer to the function via the callback parameter.
Quoted 6 lines
Quoted 1 line
This looks like a good solution
When
fmt_merge_msg_config()needs to read more stuff, the callback parameter may have to be updated, but this will do for now.Thanks.
Yes, then we can create a struct and pass the struct instead, maybe.
Thanks,
Ayush
PATCH v2 2/2 builtin/fmt-merge-msg: stop depending on 'the_repository'
In reply to Ayush ChandekarRefactor builtin/fmt-merge-msg.c to remove the dependancy on the global 'the_repository'. Remove the 'UNUSED' macro from the 'struct repository' parameter and replace 'git_config()' with 'repo_config()' so that configuration is read from the passed repository. Also, add a test to make sure that "git fmt-merge-msg -h" can be called outside a repository.
--- builtin/fmt-merge-msg.c | 5 ++--- t/t1517-outside-repo.sh | 7 +++++++ 2 files changed, 9 insertions(+), 3 deletions(-)
builtin/fmt-merge-msg.c +2 -3
@@ -1,4 +1,3 @@ -#define USE_THE_REPOSITORY_VARIABLE #include "builtin.h" #include "config.h" #include "fmt-merge-msg.h"
@@ -13,7 +12,7 @@ static const char * const fmt_merge_msg_usage[] = {int cmd_fmt_merge_msg (int argc,const char **argv,const char *prefix, - struct repository *repo UNUSED) +struct repository *repo) {char *inpath =NULL ;const char *message =NULL ;
@@ -54,7 +53,7 @@ int cmd_fmt_merge_msg(int argc,int ret;struct fmt_merge_msg_opts opts; -git_config (fmt_merge_msg_config, &merge_log_config); +repo_config (repo, fmt_merge_msg_config, &merge_log_config); argc =parse_options (argc, argv, prefix, options, fmt_merge_msg_usage,0 );if (argc >0 )
t/t1517-outside-repo.sh +7 -0
@@ -121,4 +121,11 @@ test_expect_success 'prune does not crash with -h' ' test_grep "[Uu]sage: git prune " usage ' +test_expect_success 'fmt-merge-msg does not crash with -h' ' + test_expect_code 129 git fmt-merge-msg -h >usage && + test_grep "[Uu]sage: git fmt-merge-msg " usage && + test_expect_code 129 nongit git fmt-merge-msg -h >usage && + test_grep "[Uu]sage: git fmt-merge-msg " usage +' + test_done
Signature · 1 line
2.49.0