[Question] builtin/branch.c

[Question] builtin/branch.c

[Tao Qingyun]

Hi, I am learning `builtin/branch.c`. I find that it will call `branch_get`
before create and [un]set upstream, and die with "no such branch" if failed.
but `branch_get` seems never fail, it is a get_or_create. Also, it was
confused that getting a branch before it has created.

builtin/branch.c #811

    } else if (argc > 0 && argc <= 2) {
        struct branch *branch = branch_get(argv[0]);

        if (!branch)
            die(_("no such branch '%s'"), argv[0]);

        if (filter.kind != FILTER_REFS_BRANCHES)
            die(_("-a and -r options to 'git branch' do not make sense with a branch name"));

        if (track == BRANCH_TRACK_OVERRIDE)
            die(_("the '--set-upstream' option is no longer supported. Please use '--track' or '--set-upstream-to' instead."));

        create_branch(argv[0], (argc == 2) ? argv[1] : head,
                  force, 0, reflog, quiet, track);

Re: [Question] builtin/branch.c

[Ævar Arnfjörð Bjarmason]

On Sat, Oct 13, 2018 at 10:12 AM Tao Qingyun <taoqy@ls-a.me> wrote:
> Hi, I am learning `builtin/branch.c`. I find that it will call `branch_get`
> before create and [un]set upstream, and die with "no such branch" if failed.
> but `branch_get` seems never fail, it is a get_or_create. Also, it was
> confused that getting a branch before it has created.
>
> builtin/branch.c #811
>
>     } else if (argc > 0 && argc <= 2) {
>         struct branch *branch = branch_get(argv[0]);
>
>         if (!branch)
>             die(_("no such branch '%s'"), argv[0]);

From my reading of the source you're correct. That !branch case is
pointless. The only way that function can fail is in the x*() family
of functions, which'll make the function die instead of returning
NULL.

[PATCH] builtin/branch.c: remove useless branch_get

[Tao Qingyun]

Signed-off-by: Tao Qingyun <taoqy@ls-a.me>
---
 builtin/branch.c | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/builtin/branch.c b/builtin/branch.c
index c396c41533..2367703034 100644
--- a/builtin/branch.c
+++ b/builtin/branch.c
@@ -809,11 +809,6 @@ int cmd_branch(int argc, const char **argv, const char *prefix)
 		git_config_set_multivar(buf.buf, NULL, NULL, 1);
 		strbuf_release(&buf);
 	} else if (argc > 0 && argc <= 2) {
-		struct branch *branch = branch_get(argv[0]);
-
-		if (!branch)
-			die(_("no such branch '%s'"), argv[0]);
-
 		if (filter.kind != FILTER_REFS_BRANCHES)
 			die(_("-a and -r options to 'git branch' do not make sense with a branch name"));
 
-- 
2.19.0

Re: [Question] builtin/branch.c

[Jeff King]

On Sun, Oct 14, 2018 at 12:19:35PM +0200, Ævar Arnfjörð Bjarmason wrote:

> On Sat, Oct 13, 2018 at 10:12 AM Tao Qingyun <taoqy@ls-a.me> wrote:
> > Hi, I am learning `builtin/branch.c`. I find that it will call `branch_get`
> > before create and [un]set upstream, and die with "no such branch" if failed.
> > but `branch_get` seems never fail, it is a get_or_create. Also, it was
> > confused that getting a branch before it has created.
> >
> > builtin/branch.c #811
> >
> >     } else if (argc > 0 && argc <= 2) {
> >         struct branch *branch = branch_get(argv[0]);
> >
> >         if (!branch)
> >             die(_("no such branch '%s'"), argv[0]);
> 
> From my reading of the source you're correct. That !branch case is
> pointless. The only way that function can fail is in the x*() family
> of functions, which'll make the function die instead of returning
> NULL.

It sometimes returns current_branch, which can be NULL (e.g., if you're
on a detached HEAD). Try:

  $ git branch HEAD
  fatal: no such branch 'HEAD'

  $ git branch ''
  fatal: no such branch ''

However, it seems weird that we'd check those cases here (and provide
such lousy messages). And indeed, dropping that and letting us
eventually hit create_branch() gives a much better message:

  $ git branch HEAD
  fatal: 'HEAD' is not a valid branch name.

  $ git branch ''
  fatal: '' is not a valid branch name.

I think we'd want to see that reasoning in the commit message.

-Peff

Re: [PATCH] builtin/branch.c: remove useless branch_get

[Jeff King]

On Mon, Oct 15, 2018 at 10:08:39PM +0800, Tao Qingyun wrote:

> Signed-off-by: Tao Qingyun <taoqy@ls-a.me>

The commit message should describe the "why" here. I gave some reasoning
nearby in:

  https://public-inbox.org/git/20181015171417.GA1301@sigill.intra.peff.net/

From your initial message, it sounds like this might also be fixing a
bug ("confusing that getting a branch before it has created"). Can you
describe that (and ideally show the fix with a test)?

> diff --git a/builtin/branch.c b/builtin/branch.c
> index c396c41533..2367703034 100644
> --- a/builtin/branch.c
> +++ b/builtin/branch.c
> @@ -809,11 +809,6 @@ int cmd_branch(int argc, const char **argv, const char *prefix)
>  		git_config_set_multivar(buf.buf, NULL, NULL, 1);
>  		strbuf_release(&buf);
>  	} else if (argc > 0 && argc <= 2) {
> -		struct branch *branch = branch_get(argv[0]);
> -
> -		if (!branch)
> -			die(_("no such branch '%s'"), argv[0]);
> -

From what I can tell, the patch itself _is_ an improvement. I just think
we need to explain why for the record.

-Peff

[PATCH v2] builtin/branch.c: remove useless branch_get

[Tao Qingyun]

branch_get sometimes returns current_branch, which can be NULL (e.g., if
you're on a detached HEAD). Try:

  $ git branch HEAD
  fatal: no such branch 'HEAD'

  $ git branch ''
  fatal: no such branch ''

However, it seems weird that we'd check those cases here (and provide
such lousy messages). And indeed, dropping that and letting us
eventually hit create_branch() gives a much better message:

  $ git branch HEAD
  fatal: 'HEAD' is not a valid branch name.

  $ git branch ''
  fatal: '' is not a valid branch name.

Signed-off-by: Tao Qingyun <taoqy@ls-a.me>
---
 builtin/branch.c | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/builtin/branch.c b/builtin/branch.c
index c396c41533..2367703034 100644
--- a/builtin/branch.c
+++ b/builtin/branch.c
@@ -809,11 +809,6 @@ int cmd_branch(int argc, const char **argv, const char *prefix)
 		git_config_set_multivar(buf.buf, NULL, NULL, 1);
 		strbuf_release(&buf);
 	} else if (argc > 0 && argc <= 2) {
-		struct branch *branch = branch_get(argv[0]);
-
-		if (!branch)
-			die(_("no such branch '%s'"), argv[0]);
-
 		if (filter.kind != FILTER_REFS_BRANCHES)
 			die(_("-a and -r options to 'git branch' do not make sense with a branch name"));
 
-- 
2.19.0

Re: [PATCH v2] builtin/branch.c: remove useless branch_get

[Jeff King]

On Tue, Oct 16, 2018 at 10:54:28PM +0800, Tao Qingyun wrote:

> branch_get sometimes returns current_branch, which can be NULL (e.g., if
> you're on a detached HEAD). Try:
> 
>   $ git branch HEAD
>   fatal: no such branch 'HEAD'
> 
>   $ git branch ''
>   fatal: no such branch ''
> 
> However, it seems weird that we'd check those cases here (and provide
> such lousy messages). And indeed, dropping that and letting us
> eventually hit create_branch() gives a much better message:
> 
>   $ git branch HEAD
>   fatal: 'HEAD' is not a valid branch name.
> 
>   $ git branch ''
>   fatal: '' is not a valid branch name.

This explanation is perfect, of course. ;)

I still wondered if you had another motivation hinted at in your
original mail, though (some weirdness with running branch_get early).
It's OK if there isn't one, but I just want to make sure we capture all
of the details.

Other than that question, the patch looks good to me.

-Peff

Re: [PATCH v2] builtin/branch.c: remove useless branch_get

[Tao Qingyun]

> > branch_get sometimes returns current_branch, which can be NULL (e.g., if
> > you're on a detached HEAD). Try:
> > 
> >   $ git branch HEAD
> >   fatal: no such branch 'HEAD'
> > 
> >   $ git branch ''
> >   fatal: no such branch ''
> > 
> > However, it seems weird that we'd check those cases here (and provide
> > such lousy messages). And indeed, dropping that and letting us
> > eventually hit create_branch() gives a much better message:
> > 
> >   $ git branch HEAD
> >   fatal: 'HEAD' is not a valid branch name.
> > 
> >   $ git branch ''
> >   fatal: '' is not a valid branch name.
> 
> This explanation is perfect, of course. ;)
> 
> I still wondered if you had another motivation hinted at in your
> original mail, though (some weirdness with running branch_get early).
> It's OK if there isn't one, but I just want to make sure we capture all
> of the details.
> 

Yes, this explanation is perfect. ;)

> Other than that question, the patch looks good to me.
> 
> -Peff