From: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Date: Fri, 30 Jul 2010 20:20:40 GMT
Subject: Re: When to use test_commit (Re: [PATCH] checkout: add a test for creating a new branch with regexp as a starting point)
Message-ID: <AANLkTimeQA2_fQ_f2pQchbo1coj3=mV9TRPnbET8SQ4F@mail.gmail.com>
In-Reply-To: <20100730191357.GA2448@burratino>

On Fri, Jul 30, 2010 at 19:13, Jonathan Nieder <jrnieder@gmail.com> wrote:
> Ævar Arnfjörð Bjarmason wrote:
>
>> And actually, we should probably have a
>> test_commit_notag() then.
>
> I don’t know.  What’s so bad about using "git commit" directly?

Nothing, it's just easier to write and read the tests when we have and
use functions for these common operations.

> I often find myself using "git commit" in tests because test_commit
> imposes all the usual restrictions for a ref name on the commit
> message.  I would happily use an abbreviation for
>
>  test_tick &&
>  git commit -m "something" &&
>  git tag something-else
>
> if available because I don’t like typing, but would that help the
> reader and test runner any?

Maybe just introduce a fourth argument to test_commit, to make it
<message> [<file> [<contents> [<tagname>]] instead of <message>
[<file> [<contents>]] ?