Re: [PATCH v4 0/6] Support triangular workflows

From: Jeff King <peff@peff.net>
Date: Mar 31, 2013, 21:43 UTC
Message-ID: <20130331214318.GA24646@sigill.intra.peff.net>
In-Reply-To: <CALkWK0=a0wmUe-rjK8Gg8izc0FdXYzChSrgp+W-JNsECsDf+yA@mail.gmail.com>

On Mon, Apr 01, 2013 at 02:21:22AM +0530, Ramkumar Ramachandra wrote:

Show 6 quoted lines

> Jeff King wrote:
> > [...]
> 
> So, you're saying: don't test compound statements for failure, since
> anything in the chain could fail and propagate failure.  I should only
> test simple git-foo commands for failure?

Right.

Show 9 quoted lines

> > Sometimes it's annoyingly verbose to break down a compound function. But
> > I think in this case, you can make your tests more robust by just
> > checking the affirmative that the ref is still where we expect it to be,
> > like:
> >
> >   check_push_result up_repo $the_first_commit heads/master
> 
> Doesn't that change the meaning of the test though?  I really like how
> the original tests read.

Does it? I thought the original was:

  test_must_fail check_push_result up_repo $the_commit heads/master

which is checking that we did _not_ push $the_commit to up_repo. Checking that without a negative means confirming that what _used_ to be there is still there, which is $the_first_commit.

But I didn't actually run it, so I might be wrong about what is supposed to be there after the (lack of) push.

-Peff