On Sun, Jul 31, 2005 at 08:22:26AM -0600, Wesley J. Landaker wrote: > On Sunday 31 July 2005 00:19, Andreas Barth wrote: > > we currently have almost 800 RC bugs in etch due to small glitches that > > started to make code FTBFS with the new gcc version.
> > It is urgently necessary that maintainers start to fix their own > > packages, and that whoever has some time at their hands, NMU such > > packages. > A lot of that is waiting on the C++ ABI transition, and dependencies. And if > you NMU, please make sure you know what you're doing and read the bugs in > question, so that you are not hurting more than you help. FWIW, if the package actually *can* be built against the existing libraries using g++-4.0, then it doesn't actually hurt to do so as long as the package doesn't build libraries that need to be renamed. Yes, the package will have to be rebuilt again once the libs have transitioned to g++-4.0, if only because of the package name changes; but if these NMUs let us clear out the libjack transition ahead of the various other transitions, that would certainly be a good thing. > I don't want to rant, but since you are urging NMUs and apparently doing > them yourself. Your recent NMU of cheeesetracker compiled it against two > C++ ABIs; you could avoided this if you read the bug report and my > response. Also, always contact the maintainer first, and use a delayed > upload queue. You did neither of these when NMUing cheesetracker. Compiling against two C++ ABIs should not actually cause any damage *in cases where the linker command succeeds*, since libstdc++ uses fully versioned symbols. But of course one should always try to communicate with the maintainer before uploading an NMU, regardless. -- Steve Langasek postmodern programmer
signature.asc
Description: Digital signature