Re: Puzzled by a t9500 test failure

From: Shawn O. Pearce <spearce@spearce.org>
Date: Oct 14, 2008, 00:22 UTC
Message-ID: <20081014002246.GA4856@spearce.org>
In-Reply-To: <7vhc7g8257.fsf@gitster.siamese.dyndns.org>

Junio C Hamano <gitster@pobox.com> wrote:

Show 27 quoted lines

> "Shawn O. Pearce" <spearce@spearce.org> writes:
> 
> > Junio C Hamano <gitster@pobox.com> wrote:
> >> With recent tip of 'master':
> >> 
> >>     $ make
> >>     $ mkdir t/trash
> >>     $ cd t && sh t9500-*.sh -i
> >> 
> >> fails at the very first test.  Can anybody figure out why?
> >
> > Hmmph.  Near as I can tell its because gitweb.log has this in it:
> >
> > $ cat gitweb.log
> > [Mon Oct 13 23:43:36 2008] gitweb.perl: Use of uninitialized value in scalar chomp at /home/sop/local/maint-git/t/../gitweb/gitweb.perl line 1811.
> > [Mon Oct 13 23:43:36 2008] gitweb.perl: Use of uninitialized value in addition (+) at /home/sop/local/maint-git/t/../gitweb/gitweb.perl line 3772.
> > HASH..
> >
> > The test does a grep for '[[]' and if it matches, fails.  I'm not
> > sure why this started showing up now.  Obviously I merged something
> > that failed the test suite, but I was pretty sure I had run the full
> > set before publishing anything.
> 
> Nothing as far as I can tell changed while you were the pumpking.
> 
> I was just wondering why the presense of that extra, should-be-unused,
> t/trash directory affects the outcome of the test.

Seriously?

Oh.  Some time with strace later...

Its because Perl totally borked the $projectroot path when it was looking at $GIT_DIR/ctags. $GIT_DIR has a space in the path and Perl bound the first aprt, opened it, but cannot open the second part of the $GIT_DIR/ctags string because it doesn't exist.

The code in question is the new tag cloud code in gitweb that Peter added. So bisect is still correct.

My guess is its the code in git_get_project_ctags(), around l.1804-1816.

-- 
Shawn.