Bug#844608: [Pkg-protobuf-devel] Bug#844608: Looking for sponsor for protobuf NMU
Hi,
On Fri, Nov 25, 2016 at 4:05 PM, lumin <cdluminate@gmail.com> wrote:
> It is now present in unstable. However the buildd status seems bad.
Yes, seen that. :(
> ppc64el due to segfault during python2 test.
Do you know any reason why it may happen?
> kfreebsd-* had never compiled protobuf since 3.x.x release.
That's bad. I'd like to support kFreeBSD for Stretch, but don't think
I'll have time to really look into it. :(
> armel,armhf,hppa,i386,mips,mipsel,powerpc,x32 failed during python3 test due to the same reason:
>
> Traceback (most recent call last):
> File "/«PKGBUILDDIR»/python3/google/protobuf/internal/reflection_test.py", line 639, in testIntegerTypes
> TestGetAndDeserialize('optional_uint32', 1 << 31, long)
> NameError: name 'long' is not defined
>
> There is no "long" type in python3, so this is an upstream
> py2 -> py3 porting bug. See:
>
> https://github.com/google/protobuf/issues/1504
Found this, seems to be solvable. But next time I'll do test builds
on several architectures before upload.
> I looked into protobuf's code at master branch and it seems to be solved:
>
> https://github.com/google/protobuf/blob/master/python/google/protobuf/internal/reflection_test.py#L658-L659
> https://github.com/google/protobuf/blob/master/python/google/protobuf/internal/reflection_test.py#L642-L645
This should be the way, seems to be a natural try this and if doesn't
work then go with the safe default.
> Should we consider to cherry-pick the corresponding upstream commit to
> Debian's protobuf 3.0.0 package or, to import protobuf 3.1.0
> (maybe impossible due to transition freeze) ?
As I understand, small transitions are still possible, that doesn't
affect too many packages. But to be honest, the 3.0 transition was big
enough and still not finished (see the kFreeBSD build problems). I do
_not_ want to risk having more problems with 3.1 so close to the full
freeze.
Regards,
Laszlo/GCS
PS: please don't Cc me, I'm subscribed to the list.
Reply to: