Date: Wed, 27 Oct 2004 17:57:55 +0100 From: Colin Percival <colin.percival@wadham.ox.ac.uk> To: drhodus@machdep.com Cc: freebsd-ports@freebsd.org Subject: Re: please test: Secure ports tree updating Message-ID: <417FD393.3070706@wadham.ox.ac.uk> In-Reply-To: <fe77c96b041026200121723137@mail.gmail.com> References: <417EAC7E.2040103@wadham.ox.ac.uk> <fe77c96b041026200121723137@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
David Rhodus wrote: > Is this something that can be used to replace cvsup in a general term > like cvsync ? In theory, maybe, but in practice, no. Portsnap distributes snapshots of the ports tree; it isn't designed for distributing the entire CVS repository. If you tried to use the portsnap build code for distributing the ports repository, you'd find it to be impractically slow, since there are some huge files (INDEX,v is 82.5MB) which would take far too long for my binary diff code to handle. When it comes to using portsnap for distributing snapshots of the src/, doc/, or www/ trees, there are other problems. First, these are branched, which means that lots of different snapshots would need to be created; also, while the ports tree is easily split into bite-sized pieces which tend to change independently of each other (ie, individual ports -- usually a single commit will touch several files, but only within a single port directory), the other repositories don't divide so easily into independent parts. I'm glossing over a few details here, but basically: I've given this some thought, and I think that while portsnap is useful for the ports tree, I don't think it will be very useful for anything else. Colin Percival
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?417FD393.3070706>