On 6/20/07, seth vidal <skvidal@xxxxxxxxxxxxxx> wrote:
On Wed, 2007-06-20 at 16:06 -0400, Michael DeHaan wrote: > seth vidal wrote: > > So I was wondering if it might be possible to reverse the model a bit. > > Why not make it so our buildsys and related pieces can easily pull from > > upstream's scm's. > > This seems to be the best way to take advantage of distributed SCM's to > me, and it allows for having one less resource to manage (i.e. Fedora > CVS). Right now, I find Fedora CVS a bit annoying as, well, it means > using yet-another-version control repository -- and lots of seperate > checkins when I want to push content to 3 seperate repos. > > However it does rely on the accessibility of those upstream repos. > Shouldn't be a major issue. If it's down, no updates. but things can go away 'forever' and we still want them around. It seems like no matter which way I turn this around in my head we end up having to have a complete copy of everything in fedora's pkg vcs to reliably do what we need to do. Not to mention the issues of firewalls and the buildsys talking to hosts in $not_okay_countries. In short, we have to have everything local otherwise we'll be exchanging one set of problems for larger, more intractable ones. (ie: legal ones and general confusion-of-location)
Yeah... I kept running into issues like: Upstream changes SCM (often) Upstream forks.. No SCM (Tarballs baby... just like our lord Volkerding wants) No sane SCM (Distributed SCCS) No open/usable SCM (Distributed SCCS using bittorrent) Legal fork issues (mp3 code, patented one-clikc method, etc :) CLA issues (there has got to be someway the CLA will get in the way somewhere :)). With a couple hundred packages this might be surmountable.. with tens of thousands of packages? Maybe we can get our new Google Overlords to solve this problem with Google Forge? -- Stephen J Smoogen. -- CSIRT/Linux System Administrator How far that little candle throws his beams! So shines a good deed in a naughty world. = Shakespeare. "The Merchant of Venice"
