Bug#726165: mumble with protobuf 2.5.0-7 works
Chris Knadle wrote:
> On Friday, January 31, 2014 15:18:18 Robert Edmonds wrote:
> [...]
> > Chris Knadle's input in #737246 makes me believe that the changes in
> > 2.5.0-6 / -7 just aren't correct. I'm thinking we should probably go
> > back to the approach in 2.5.0-5 (though with a fallback atomic
> > implementation for architectures where the default gcc is < 4.7).
>
> Unfortunately the feedback I gave you about protobuf 2.5.0-6 / -7 turns out
> to have been wrong; my local cowbuilder had something weird going on. That
> mumble works when built against protobuf 2.5.0-7 got reported to me in #737223
> by Gonéri Le Bouder, with after some efforts was able to replicate with
> cowbuilder.
Hi, Chris:
After further investigation, reading upstream bug #351, and commits
r409, r410, r413, r414, and r415 [1], I'm not convinced that the changes
I made in protobuf 2.5.0-6 / -7 are complete, and in any case I'm now no
longer convinced that it's feasible to forward port the "once"
implementation from protobuf <= 2.4.1 to later versions.
[0] https://code.google.com/p/protobuf/issues/detail?id=351
[1] https://code.google.com/p/protobuf/source/detail?r=409, ?r=410, etc.
I've uploaded protobuf 2.5.0-8 to experimental, which has the exact same
ABI/API as protobuf 2.5.0-5. Can you tell me if the current version of
mumble in the archive works with libprotobuf8 2.5.0-8, once it's
available at your mirror? (I suspect that it will, but just want to
make sure.)
I am pretty sure 2.5.0-8 will not work on ia64 or sparc, where the
default compiler is gcc-4.6, but it also seems that this problem is not
so serious now.
> Should I file a release.debian.org bug to binNMU mumble?
I think this is a problem in the protobuf transition, so #726165 is the
right bug for this discussion :-)
That is, with protobuf 2.5.0-8 there should be no additional binNMUs
required. If that's the case, I'll upload -8 to unstable as -9,
provided it is acceptable to break the architectures with the old
gcc-4.6 compiler.
--
Robert Edmonds
edmonds@debian.org
Reply to: