Hi Scott, > This is another talk I'd really rather someone else give, but I can > give a brief talk on our current status & problems & desires if it is > helpful. > > OLPC has forks of a number of Fedora packages, for a number of > reasons. We've been trying to keep better track of the what & why, at > http://wiki.laptop.org/go/Distro_version_migration_nastiness > (great page name, eh? ;-) > > There are a number of tools that could make manging this divergence > and pushing changes upstream easier. I've talked some recently with > J5 about work he's been doing on this front. To start with: some sort > of notification mechanism to let us know if a new upstream release has > occurred for a package we have forked would be invaluable, as would > the ability to easily generate lists of "forked packages" and see the > exact patches applied. Making it easier to apply the OLPC "packaging > patches" to new upstream versions would help, too. Tools to > automatically monitor dependencies, so that we can immediately detect > when perl sneaks back into our build, and act to get rid of it. > Better integration of Red Hat bugzilla and OLPC trac. J5 has also > mentioned "workflow tools" which would make it easier to guide new > OLPC developers through tasks like "fork a Fedora package for OLPC, > apply the OLPC patch, and notify the upstream developer". > > I can present some of these ideas, and solicit attendees to go off and > write tools to improve the situation. ;-) But if someone wants to > present a concrete proposal for "how I plan to make Fedora integration > better for 9.1" I'd love it.
I've been doing some work recently with this recently with Greg Dekoenigsberg. I posted some of my findings on the fedora-olpc list here https://www.redhat.com/archives/fedora-olpc-list/2008-October/msg00025.html. Alot of the current divergence isn't that bad and I was hoping to get most of the fixes into rawhide in time for Fedora 10 so that as OLPC is rebased most of the branches shouldn't be needed as those changes should already be upstream. The details in the message mentioned are mostly still up to date except that the hal battery polling is now fixed upstream and is in rawhide so its one less to fix. Peter _______________________________________________ Devel mailing list [email protected] http://lists.laptop.org/listinfo/devel
