Re: ffmpeg-1.2 update

W dniu 20.03.2013 23:53, Nicolas Chauvet pisze: > 2013/3/18 Nicolas Chauvet > <kwizart(a)gmail.com > <mailto:kwizart@gmail.com>> > > 2013/3/18 Julian Sikorski > <belegdol(a)gmail.com > <mailto:belegdol@gmail.com>> > > W dniu 18.03.2013 21:16, Nicolas Chauvet pisze: > > 2013/3/18 Julian Sikorski > > <belegdol(a)gmail.com > <mailto:belegdol@gmail.com> > > <mailto:belegdol@gmail.com > <mailto:belegdol@gmail.com>>> > > > > Dear all, > > > > ffmpeg-1.2 was released recently. I have run a round of > dep rebuilds and > > the only packages that failed to rebuild were bino, > lightspark and > > dvbcut. None of these failures point directly to ffmpeg > though. > > I'd opt for pushing this into devel and F-19 ASAP, and > then maybe even > > to F-18 - but that one should be carefully considered. > > rpmsodiff output is attached. > > > > devel is rawhide aka fedora-20 right now, Any update must > pass rawhide > > usability tests before anything else. > > As always I would like to have a sum-up of the ABI changes > involved if any. > > > > I'm not fundamentally against having ffmpeg-1.2 in > branched/F-19 at a > > later point. I just know some later ffmpeg update was locked > by vlc > > because they will build but the rework needed to make it run > will be > > done in the next vlc, not vlc bugfix. > > > > Nicolas (kwizart) > > > > Ps: I also don't have seen much mplayer update recently > > > > > > > > > As you can see from the rpmsodiff output, only one symbol was > removed. > > I should apologize, I've miss to see the rpmdiff output was attached in > the first message. > So the rpmsodiff looks more safe than I expected. > There is a need to know the reason from this symbol removal, but this > should be better to expect to F-19 soon, indeed. > > That been said, there are still pending FTBFS from the F-19 mass > rebuild, and some of theses are related to ffmpeg API. So unless there > is a newer version compatible with current ffmpeg, it would like them to > be moved to ffmpeg-compat (motion failed to catch it despite my attempt). > > Nicolas (kwizart) Hardware acceleration was overhauled so this might be the reason for symbol removal. I have now committed the updated spec so that maintainers can look into fixing their packages, I'll submit a build in several days.