Re: [PATCH v4 5/7] tests: don't swallow Git errors upstream of pipes

From: Matthew DeVore <matvore@google.com>
Date: Oct 5, 2018, 17:54 UTC
Message-ID: <CAMfpvhJasrpDxgt+LATsSQf68LB97hCQnCCnRMJP51iXDc34SQ@mail.gmail.com>
In-Reply-To: <xmqqd0sowda4.fsf@gitster-ct.c.googlers.com>

On Fri, Oct 5, 2018 at 9:48 AM Junio C Hamano <gitster@pobox.com> wrote:

Show 7 quoted lines

>
> Hopefully this is not a blind mechanical patch, as introduction of
> unexpected temporary files in the working tree could interfere with
> later tests (e.g. they may expect exact set of untracked files, and
> these new temporary files would b unexpected additions).
>
> Thanks.

I reviewed the files modified in this patch. For the svn one, there are other scratch files being created in the cwd before mine, so my additions should be fine, I admit I don't fully understand the commands being run in that test...after the temp files are created, I see git svn proplist and git svn propget being executed... the public documentation for them does not seem to contradict my assumption. (http://svnbook.red-bean.com/en/1.6/svn.ref.svn.c.propget.html)

As for the other test files in this patch, all git invocations are either "git clone" or "git init <dir>" or "git -C ..." which means none of them care about extra files in the cwd. After I searched for Git commands I read through the files somewhat briskly and looked for any globs or suspicious invocations of awk, grep, or other commands (we really do use a limited number of commands in the tests I saw :) and did not find anything that depended on existence or non-existence of unnamed files.