Re: [WIP] Shift rev-list enumeration from upload-pack to pack-objects

From: Nicolas Pitre <nico@cam.org>
Date: Jun 7, 2009, 16:41 UTC
Message-ID: <alpine.LFD.2.00.0906071225060.3906@xanadu.home>
In-Reply-To: <c77435a80906070625i4daaa69bi134df765d7a77cdf@mail.gmail.com>

On Sun, 7 Jun 2009, Nick Edelen wrote:

Show 6 quoted lines

> I'm using the --revs flag in pack-objects, which causes it to use
> get_object_list().  you'll notice, regardless of whether --thin is
> set, this function still calls
> 	mark_edges_uninteresting(revs.commits, &revs, show_edge);
> which sets uninteresting objects as preferred bases, which I'd think
> would create a thin pack.  I could be wrong though...

Look at the arguments passed to setup_revisions(). When --thin is set, the --objects-edge flag is passed instead of --objects. Now see what effect this has on the third argument of mark_edges_uninteresting().

Show 5 quoted lines

> as I mentioned in the comment and above, it's an easy fix, but even 
> then I wasn't sure what to do with commit grafts.  as use_thin_pack 
> seemed to be predominantly set on shallow interactions, I just didn't 
> bother seperating the cases 'normal but thick pack' and 'shallow 
> stuff'.

Please do separate them. In theory you could use thin packs with a relative deepening of a shallow clone. In other words, !thin and shallow is a wrong association to make.

Nicolas