On Sam, 2010-10-02 at 20:53 +0300, Adrian Bunk wrote: > On Fri, Oct 01, 2010 at 07:23:05PM +0200, Michel Dänzer wrote: > > On Fre, 2010-10-01 at 18:57 +0200, Julien Cristau wrote: > > > On Fri, Oct 1, 2010 at 18:29:54 +0200, Michel Dänzer wrote: > > > > > > > On Fre, 2010-10-01 at 18:27 +0200, Julien Cristau wrote: > > > > > #if ABI_VIDEODRV_VERSION < SET_ABI_VERSION(9, 0) > > > > > old code > > > > > #else > > > > > new code > > > > > #endif > > > > > > > > That would work perfectly, if the ABI was bumped by every change which > > > > breaks the ABI... > > So when there are 30 ABI-breaking commits between 1.9 and 1.10, > it should be required to bump the ABI 30 times?
Yeah, I don't see any problem with that. The ABI major versions are just numbers. > > > It'll work for this specific bump. > > > > Not when building against xserver commits between > > 5c6a2f93ebc16a78093782b442306de23ae94e78 and > > 1a9022d3289ff74043112808cc8aa315fd9c40e8 . Certainly better than merely > > checking the server version though. > >... > > Does it really have to be supported to build against random older > snapshots of unstable xserver? It would help e.g. for bisecting. But don't worry, I seem to be firmly in the minority camp on this — the trend has been to stop caring about such mundane things as backwards compatibility. Apparently it's okay to cause pain for testers for a little developer convenience. -- Earthling Michel Dänzer | http://www.vmware.com Libre software enthusiast | Debian, X and DRI developer _______________________________________________ [email protected]: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: http://lists.x.org/mailman/listinfo/xorg-devel
