On Tue, 2011-08-02 at 17:39 +0100, Adam D. Barratt wrote: > [simgear maintainer added to Cc] > > On Tue, 2 Aug 2011 18:25:28 +0200, Kurt Roeckx wrote: > > On Tue, Aug 02, 2011 at 01:19:59PM +0100, Adam D. Barratt wrote: > >> On Tue, 02 Aug 2011 08:49:07 +0100, Christopher Baines wrote: > >> >The maintainer for simgear has been updating parts of the library > >> for > >> >the openscenegraph transition. However this appears to make fgrun > >> >segfault (when built with simgear 2.0.0-2 at least). > >> nmu fgrun . ALL . -m "Rebuild against simgear-dev 2.0.0-4 to fix > >> segaults" > >> dw fgrun . ALL . -m 'simgear-dev (>= 2.0.0-4)' > > > > It's not clear to me why this is a good idea. Did simgear break > > it's ABI? > > The change in -4 only affects a private symbol, so shouldn't be an > issue. > > There are indeed a few suspicious looking changes in -3 though: > > - virtual void forceLoad(osgDB::DatabasePager* dbp); > + virtual void forceLoad(osgDB::DatabasePager* dbp, > + osg::NodePath& path); > > for example. My fault for only checking the latest changes :/ > > Adam The reason for the changes to simgear is the release teams transition of openscenegraph. The transition meant that if simgear could not build against the new version of openscenegraph (3), it and its dependants (flightgear, fgrun) would be removed from testing. Simgear's maintainer, Ove, used the upstream source and backported the relevant changes to allow simgear to build against the version of openscenegraph entering testing. From the wikipedia description of ABI, I think openscenegraph broke its ABI when it moved to the new major version (3), thus breaking simgear. Chris
Attachment:
signature.asc
Description: This is a digitally signed message part