Re: [PATCH 3/3] t3600: test rm of path with changed leading symlinks

From: Jeff King <peff@peff.net>
Date: Apr 4, 2013, 23:29 UTC
Message-ID: <20130404232903.GA27128@sigill.intra.peff.net>
In-Reply-To: <7vhajlgabi.fsf@alter.siamese.dyndns.org>

On Thu, Apr 04, 2013 at 04:12:01PM -0700, Junio C Hamano wrote:

Show 10 quoted lines

> Jeff King <peff@peff.net> writes:
> 
> > Deleting across symlinks inside the repo can be brushed aside with "eh,
> > well, it is just another way to mention the same name in the
> > filesystem". But deleting anything outside of the repo seems actively
> > wrong.
> 
> Yup, you finally caught up ;-) IIRC, such an outside repository
> target was the case people realized that "git add" shouldn't see
> across symlinks.

It would help if you spelled it out rather than making me come to it while arguing against you. ;P

Show 7 quoted lines

> > Hmm. I think you have convinced me (or perhaps I have convinced myself)
> > that we should generally not be crossing symlink boundaries in
> > translating names between the filesystem and index. I still don't want
> > to work on it, though. :)
> 
> That is OK.  Just let's not etch a wrong behaviour in stone with
> that test.

So let's drop patch 3. Do we want instead to have an expect_failure documenting the correct behavior?

-Peff