[PATCH] config.c: Remove unused git_config_global() function

[PATCH] config.c: Remove unused git_config_global() function

[Ramsay Jones]

Commit 8f323c00 (drop support for GIT_CONFIG_NOGLOBAL, 15-03-2011)
removed the git_config_global() function, among other things, since
it is no longer required. Unfortunately, this function has since
been unintentionally restored; I suspect by merge commit 25b86e45
(Merge branch 'jk/maint-config-alias-fix' into next, 25-05-2011).

Signed-off-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
---

Hi Junio,

A few days ago, when the next branch was @ d10669c, I noticed that
"make sparse" on next was complaining thus:

    config.c:828:5: warning: symbol 'git_config_global' was not \
        declared. Should it be static?

However, I remembered that this function had recently been removed, so
I took a quick look ...

ATB,
Ramsay Jones

 config.c |    5 -----
 1 files changed, 0 insertions(+), 5 deletions(-)

diff --git a/config.c b/config.c
index a8267e9..e0b3b80 100644
--- a/config.c
+++ b/config.c
@@ -825,11 +825,6 @@ int git_config_system(void)
 	return !git_env_bool("GIT_CONFIG_NOSYSTEM", 0);
 }
 
-int git_config_global(void)
-{
-	return !git_env_bool("GIT_CONFIG_NOGLOBAL", 0);
-}
-
 int git_config_early(config_fn_t fn, void *data, const char *repo_config)
 {
 	int ret = 0, found = 0;
-- 
1.7.5

Re: [PATCH] config.c: Remove unused git_config_global() function

[Junio C Hamano]

Ramsay Jones <ramsay@ramsay1.demon.co.uk> writes:

> Commit 8f323c00 (drop support for GIT_CONFIG_NOGLOBAL, 15-03-2011)
> removed the git_config_global() function, among other things, since
> it is no longer required. Unfortunately, this function has since
> been unintentionally restored; I suspect by merge commit 25b86e45
> (Merge branch 'jk/maint-config-alias-fix' into next, 25-05-2011).

Yeah, I think it was 1f9a980636 which was a merge between fb674d7 and
73546c0.  I just compared the output between these two:

   $ git diff fb674d7...73546c0 -- config.c ;# what the topic did
   $ git diff fb674d7 1f9a980 -- config.c   ;# what was merged

and indeed I can see these four lines snuck in by accidental faulty
conflict resolution. My bad.

Thanks for spotting.

Effectively navigating branch history (was: Re: [PATCH] config.c: Remove unused git_config_global() function)

[Stephen Bash]

----- Original Message -----
> From: "Junio C Hamano" <gitster@pobox.com>
> Sent: Tuesday, May 31, 2011 1:51:06 PM
> Subject: Re: [PATCH] config.c: Remove unused git_config_global() function
> 
> > Commit 8f323c00 (drop support for GIT_CONFIG_NOGLOBAL, 15-03-2011)
> > removed the git_config_global() function, among other things, since
> > it is no longer required. Unfortunately, this function has since
> > been unintentionally restored; I suspect by merge commit 25b86e45
> > (Merge branch 'jk/maint-config-alias-fix' into next, 25-05-2011).
> 
> Yeah, I think it was 1f9a980636 which was a merge between fb674d7 and
> 73546c0. I just compared the output between these two:
> 
> $ git diff fb674d7...73546c0 -- config.c ;# what the topic did
> $ git diff fb674d7 1f9a980 -- config.c ;# what was merged

Junio - I'd like to take a chance to learn from a master here if you don't mind.  I often find myself wanting to do the "what the topic did" operation but once the branch is merged and deleted, I have difficulty finding useful SHAs to diff.

So in an attempt to educate myself, I tried to track down fb674d7 and 1f9a980 (and failed miserably).  I can easily find 8f323c00 using 'git log -S git_config_global', but conceptually I want to trace children of 8f323c00 to see where it merged, and that seems ... complicated.  Do you mind walking through your thought process on this problem?

Thanks!
Stephen

Re: Effectively navigating branch history

[Junio C Hamano]

Stephen Bash <bash@genarts.com> writes:

>> Yeah, I think it was 1f9a980636 which was a merge between fb674d7 and
>> 73546c0. I just compared the output between these two:
>> 
>> $ git diff fb674d7...73546c0 -- config.c ;# what the topic did
>> $ git diff fb674d7 1f9a980 -- config.c ;# what was merged
>
> I often find myself wanting to do the "what the topic did" operation but
> once the branch is merged and deleted, I have difficulty finding useful
> SHAs to diff.
>
> So in an attempt to educate myself, I tried to track down fb674d7 and
> 1f9a980 (and failed miserably).

I just did

    $ git log --first-parent master -- config.c

which was sufficient for my workflow as most of the commits on master are
merges from topics and I can see what each topic did from the merge commit
summary message.  


If I were doing this on a codebase I am not very familiar with, I probably
I would have first narrowed things down a bit by

    $ git grep -n -e NOGLOBAL -- config.c ;# to learn the line number 830
    $ git blame -L825,835 -- config.c

to learn where the offending NOGLOBAL came from (it is from ab88c363) so
that I can say

    $ git log -m -p -GNOGLOBAL ab88c363.. -- config.c

to find that 1f9a980 reintroduced it by mistake.

Re: Effectively navigating branch history

[Stephen Bash]

----- Original Message -----
> From: "Junio C Hamano" <gitster@pobox.com>
> Sent: Tuesday, May 31, 2011 2:40:50 PM
> Subject: Re: Effectively navigating branch history
> 
> >> Yeah, I think it was 1f9a980636 which was a merge between fb674d7
> >> and 73546c0. I just compared the output between these two:
> >>
> >> $ git diff fb674d7...73546c0 -- config.c ;# what the topic did
> >> $ git diff fb674d7 1f9a980 -- config.c ;# what was merged
> >
> > I often find myself wanting to do the "what the topic did" operation
> > but once the branch is merged and deleted, I have difficulty finding
> > useful SHAs to diff.
> >
> > So in an attempt to educate myself, I tried to track down fb674d7
> > and 1f9a980 (and failed miserably).
> 
> I just did
> 
> $ git log --first-parent master -- config.c
> 
> which was sufficient for my workflow as most of the commits on master
> are merges from topics and I can see what each topic did from the merge
> commit summary message.

Ah, if only our workflow made it that easy... unfortunately it does not ;)

> If I were doing this on a codebase I am not very familiar with, I
> probably I would have first narrowed things down a bit by
> 
> $ git grep -n -e NOGLOBAL -- config.c ;# to learn the line number 830
> $ git blame -L825,835 -- config.c
> 
> to learn where the offending NOGLOBAL came from (it is from ab88c363)
> so that I can say
> 
> $ git log -m -p -GNOGLOBAL ab88c363.. -- config.c
> 
> to find that 1f9a980 reintroduced it by mistake.

Ah, I was missing the -m!  At one point I realized pickaxe was only showing me non-merge commits and skimmed the log manpage to see if there was an option that would enable it but apparently missed -m.

Thanks!

Stephen