Alexander Gottwald wrote:
Harold L Hunt II wrote:
Another thing to keep in mind is how we want to do development. It has been suggested that we keep the HEAD branch in sync with XFree86.org and that we do our development on another branch. The question here is whether cvsup can preserve a local branch of the code and still be used to sync with XFree86.org. I doubt that this is the case, since cvsup is essentially mirroring the files, not branches/tags/etc. Does this mean that we must manually track XFree86.org and apply their patches after the initial import?
My suggestion is to import the current "stable" release into our CVS. With CVS we can later import the next release and merge all patches we have already commited. Fixing severe bugs is still an issue and might be solved by regulary importing the snapshots of the "stable" branch and by monitoring the XFree-commit list (I still read every posting on this list and would just pay more attention to security fixes)
Mike Harris had a good point that we should grab XFree86's CVS tree with cvsup and use a perl script to change the root for all of the files. Then we have both the current version of all files *and* the history of all of those files.
He suggested using cvsps to generate patch sets. He also suggested doing our development on a branch, keeping HEAD more or less in sync with XFree86.org CVS HEAD, and merge HEAD to our branch whenever required (to get bug fixes, etc.).
I doubt that a complete mirror of the XFree86 CVS is a good solution since there is no way (at least I konw of none) to automaticly track changes in the XFree86 repository and commit them to ours too. So importing the whole repository is in my opinion a waste of space since we'd have to import all old revisions from the XFree repository too.
I think Mike had a good point that it would be wise to have the history of each file in the tree... what do you think?
Harold
