Re: bug report: pre-push hook

From: Emily Shaffer <emilyshaffer@google.com>
Date: Mar 24, 2022, 22:47 UTC
Message-ID: <CAJoAoZmfLaxTGGQRaR8D0-L3pH563z6_04UUwBo7GY1A14Ebww@mail.gmail.com>
In-Reply-To: <Yjziu8nU/SZ7SyH0@nand.local>

On Thu, Mar 24, 2022 at 2:29 PM Taylor Blau <me@ttaylorr.com> wrote:

Show 26 quoted lines

>
> On Thu, Mar 24, 2022 at 11:30:16AM -0700, Emily Shaffer wrote:
> > > What happened instead? (Actual behavior)
> > > Tests seem to execute on the background, but I do not see output
> >
> > I wonder - are you printing your test output to stdout, or stderr?
> > stderr of the hook is printed to stdout of the Git process, so that
> > may be why.
>
> I didn't think that we redirected the pre-push (or any) hook's stderr to
> stdout in the parent process. I wrote a small pre-push hook which is
> just:
>
>     $ cat .git/hooks/pre-push
>     #!/bin/sh
>
>     echo >&2 "hi!"
>
> and then:
>
>     $ git.compile push ttaylorr --dry-run 2>foo
>     $ grep hi foo
>     hi!
>
> But I might be holding it wrong, since I am not a frequent user of
> hooks.

Ah, it seems I'm completely mistaken :) Sorry for the misinfo. Urk.

Having a look in 'master', run_pre_push_hook() does not use hook.c machinery yet at all; it also seems like it doesn't try to capture stdout from the pre-push child process anyways. That makes me ask "how did this ever work?" which brings me back to wanting a small hook we can try to reproduce with from Honza :)

 - Emily