Re: Public repro case! Re: [PATCH/RFC] Allow writing loose objects that are corrupted in a pack file

From: Shawn O. Pearce <spearce@spearce.org>
Date: Jan 8, 2009, 06:04 UTC
Message-ID: <20090108060413.GA17728@spearce.org>
In-Reply-To: <7v4p0a8if2.fsf@gitster.siamese.dyndns.org>

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

Show 14 quoted lines

> "Shawn O. Pearce" <spearce@spearce.org> writes:
> > For those following along at home, Linus' 2.6 tree:
> >
> > $ ulimit -v `echo '150 * 1024'|bc -l`
> > $ git co 56d18e9932ebf4e8eca42d2ce509450e6c9c1666
> 
> Hmm, without any "wrap zlib to die on error" patch, this step already
> fails with:
> 
>     $ git checkout 56d18e9932ebf4e8eca42d2ce509450e6c9c1666
>     fatal: Out of memory? mmap failed: Cannot allocate memory
> 
> I guess that is because our test repositories are packed differently.
> I'll retry after repacking..

Yup. I actually did something more like this to get the test repository:

  git clone git://android.git.kernel.org/kernel/common.git
  git fetch git://kernel.org/pub/.../torvalds/linux-2.6.git master

The android kernel repository I had handy on my local system was quite a bit away from Linus' so I wound up with two different but sizable packs. I thought android was closer to upstream, but its apparently not. I started from there because it was local and I thought it would be a quick way to get a test environment, but sadly it didn't even have the base 56d18e we were talking about.

-- 
Shawn.