From: Elrond <elrond+kernel.org@samba-tng.org>
Date: Thu, 18 May 2006 13:29:57 GMT
Subject: Re: [Patch] git-cvsimport: tiny fix
Message-ID: <20060518132956.GA19476@memak.tu-darmstadt.de>
In-Reply-To: <46a038f90605172153mac96f40id9a50d2f29c75915@mail.gmail.com>

On Thu, May 18, 2006 at 04:53:56PM +1200, Martin Langhoff wrote:
> On 5/18/06, Junio C Hamano <junkio@cox.net> wrote:
> >Could somebody who actually works with CVS import Ack this?
> >Pretty please?
> 
> Sounds sane. It would be interesting to hear about what repo (and
> server) this was seen against. Elrond, can you tell us more about
> this?

This is a private local repository.
git-cvsimport starts a local "cvs server" for this.

I tried to create a minimal repo, that would trigger the
same behaviour, but didn't succeed. My current guessing is,
that this happened in cooperation with "cvsps" caching.
(cvsps is a tool used by git-cvsimport).

I will move the cvsps cache for my problematic repo out of
the way and try a git-cvsimport from scratch to verify that
the above mentioned issue is related to cvsps caching.
But I still think, that handling all cvs pserver replies
should be done anyway in git-cvsimport (when it relies on
this mode of operation).


    Elrond