From: Jeff King <peff@peff.net>
Date: Thu, 27 May 2021 13:48:11 GMT
Subject: Re: Runaway sed memory use in test on older sed+glibc (was "Re: [PATCH v6 1/3] test: add helper functions for git-bundle")
Message-ID: <YK+jGxlHHZ63o94N@coredump.intra.peff.net>
In-Reply-To: <87tumol4tg.fsf@evledraar.gmail.com>

On Thu, May 27, 2021 at 02:19:04PM +0200, Ævar Arnfjörð Bjarmason wrote:

> Anyway, looking at this whole test file with fresh eyes this pattern
> seems very strange. You duplicated most of test_commit with this
> test_commit_setvar. It's a bit more verbosity but why not just use:
> 
>     test_commit ...
>     A=$(git rev-parse HEAD)
> 
> Or teach test_commit a --rev-parse option or something and:
> 
>     A=$(test_commit ...)

The latter would be nice (and maybe even could be the default, as the
stdout is usually just going to --verbose output anyway). But I don't
think it works, because the $() is a subshell, so we lose the side
effect of test_tick incrementing the timestamp counter.

I wasn't following this thread carefully, but your suggestion to just
do "A=$(git rev-parse HEAD)" seems quite readable to me.

-Peff