From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Mon, 18 Feb 2008 13:35:18 GMT
Subject: Re: [PATCH 2/5] Add platform-independent .git "symlink"
Message-ID: <alpine.LSU.1.00.0802181334010.30505@racer.site>
In-Reply-To: <8c5c35580802180518x5c392bddya21c48ea4db9a7cd@mail.gmail.com>

Hi,

On Mon, 18 Feb 2008, Lars Hjemli wrote:

> On Feb 18, 2008 1:34 PM, Johannes Schindelin 
> <Johannes.Schindelin@gmx.de> wrote:
> >
> > [* snip *]
> >
> > So maybe we can just scratch all that, and I agree that an invalid 
> > .git file means "no repository" (as opposed to "no valid repository").
> >
> > In that case, you might want to test for that, too...
> 
> I think I do, did you find a loophole in the testing?

I meant writing a readable, but invalid .git file.  Then testing that Git 
die()s with the correct message.

> > Speaking about tests:
> >
> > > +test_expect_success 'setup' '
> > > +       REAL="$(pwd)/.real" &&
> > > +       mv .git "$REAL" &&
> > > +       echo "gitdir: $REAL" >.git
> > > +'
> >
> > Let's not do this.  It would clutter the t/ directory unnecessarily.
> 
> What do you mean? The test just moves t/trash/.git to t/trash/.real...

Oh, well.  I'll go and buy me some new glasses.

Thanks,
Dscho