From: Shawn Pearce <spearce@spearce.org>
Date: Sun, 29 Oct 2006 03:38:29 GMT
Subject: Re: fetching packs and storing them as packs
Message-ID: <20061029033829.GA3435@spearce.org>
In-Reply-To: <4543DA2E.9030300@tromer.org>

Eran Tromer <git2eran@tromer.org> wrote:
> Hi Shawn,
> 
> On 2006-10-28 09:21, Shawn Pearce wrote:
> > Lets say that a pack X is NOT eligible to be repacked if
> > "$GIT_DIR/objects/pack/pack-X.keep" exists.
> > 
> > Thus we want to have the new ".keep" file for historical packs and
> > incoming receive-pack between steps c and g.  In the former case
> > the historical pack is already "very large" and thus one additional
> > empty file to indicate we want to retain that pack as-is is trivial
> > overhead (relatively speaking); in the latter case the lifespan of
> > the file is relatively short and thus any overhead associated with it
> > on the local filesystem is free (it may never even hit the platter).
> 
> Sounds perfect.
> 
> It would be nice to have whoever creates a pack-*.keep file put
> something useful as the content of the file, so we'll know what to clean
> up after abnormal termination:
> 
> $ grep -l ^git-receive-pack $GIT_DIR/objects/pack/pack-*.keep

Yes, that's a very good idea.  When I do the git-receive-pack
implementation tonight I'll try to dump useful information to the
.keep file such that you can easily grep for the stale .keeps
and decide which ones should go.

--