You are not logged in.
I have posted about this on the mailing list but have not yet had any responses.
I am looking at setting up an ABS-like CVS repository but I need some help understanding how the Arch repos have been structured vs "normal" CVS. I can see that the Arch repos are set up so that there are four unique roots, one for each repository. As far as I undestand it our hosting solution allows us access to one CVS root only. Does that mean that we can set up one ABS type repo? What about CVS modules? How do they fit into the equation?
Having worked out what we can actually achieve, how do I then go about creating the supfile and configuring cvsup on the server? I don't necessarily need exact details, I can read docs but I have no idea where to start. I don't understand the basics of the ABS/CVS interaction and I cannot find any documentation on it's creation.
I would appreciate some help on this one.
Offline
There was a similar (maybe not) discussion on the arch-ports ML.: http://archlinux.org/mailman/listinfo/arch-ports
Maybe you could try posting there as someone who wants to host an abs/cvs server for an Arch port will face the same issues. Maybe you'll get an answer there. Sorry, I can't suggest anything else.
Offline
There's more too it than that I think phrak but I will take a look
Offline
If you're trying to set up your own system, you might prefer subversion. Xentac uses it, we use it for Arch64, and it would be better all round if the devs used it too. In fact, if you used it, maybe that would help leverage the devs to grow up. Ask syamajala or Xentac for details...
Dusty
Offline
Ahhh but I need the support of ABS - I know cactus hacked together and SVN alternative but it would require more work on my part to implement that soley to support my SVN repo...
Maybe we could campaign for a complete change?
Offline
good luck campaigning.
Why do you need abs? Is anybody going to use it? Can't you just give them svn access?
Dusty
Offline
If we have have ABS integration then people can have easy access to PKGBUILDs to rebuild pkgs and we may even use the system in such a way that some pkgs are only available if you build them yourself. This is especially true of the data and docs pkgs, which are huge and would be a real bandwidth sucker. Keep them up to date would be an annoying pain too. Basically it would just give us another option. If we can intergrate with ABS then we can easily include scripts to checkout abs and build certain pkgs on request.
As far as people using it - people may never even use the thing at all! It's not worth considering that
Offline
well... I don't really see the difference between an anonymous svn checkout and an abs checkout, especially considering the gains you get (as a repo contributor) by using svn over cvs.
Having said that, I also have no idea how to answer your original question either, so I'm pretty much useless here. :-D
Dusty
Offline
There's more too it than that I think phrak but I will take a look
I don't think so - you setup a cvs repo, do some tagging or something to that effect (there's a mid step I can't recall), and then make a supfile.
Offline
I think you are right I read some and it looks easy - I just have to now wait for my support ticket to be answered so i can actually find out where my CVS repo is physically stored. It also seems that I can have any number of repos in any struture, cvsup can account for it all
Offline
WIth SVN, if I have a module called a can i have have dirs called called b, c and d ( /a/{b,c,d} ) within it that can be checkout individually?
Offline