Bug#739261: libhdf5-openmpi-dev: Version in stable (wheezy) does not work with gfortran from stable

[Emilio Pozuelo Monfort]

On 12/06/14 21:51, Gilles Filippini wrote:
> fixed 739261 hdf5/1.8.12+docs-1.1
> thanks
> 
> Emilio Pozuelo Monfort a écrit , Le 12/06/2014 20:11:
>> On 04/06/14 01:17, Gilles Filippini wrote:
>>> Hi,
>>>
>>> Frank Loeffler a écrit , Le 03/06/2014 21:01:
>>>> Being hit by this myself now, I am a bit surprised by the reaction "can
>>>> wait a little longer", for an issue that clearly breaks the Fortran
>>>> interface and seems to be easily fixable.
>>>>
>>>> But this aside - is there a plan to get this into _any_ of the future
>>>> point releases of stable?
>>>
>>> I have no plan but getting the binNMU #740561 processed.
>>> And it all depends on the good will of the release team.
>>
>> You've requested a binnmu for stable on ALL architectures. Before scheduling
>> that, I'd like to clarify some things:
>>
>> Is this bug affecting testing/unstable? If not, please mark it as fixed as
>> appropriate in #739261.
> 
> This bug doesn't affect testing nor unstable. Marking it fixed for the
> related version.
> 
>> Is this bug really affecting all architectures? From what I can see, gfortran in
>> wheezy is 4.6 everywhere except on amd64, i386, kfreebsd-amd64 and kfreebsd-i386:
>>
>>   gfortran |     4:4.6.3-8 |     stable | armel, armhf, ia64, mips, mipsel,
>> powerpc, s390, s390x, sparc
>>   gfortran |     4:4.7.2-1 |     stable | amd64, i386, kfreebsd-amd64, kfreebsd-i386
>>
>> And hdf5 1.8.8-9 was built against 4.6 everywhere, from what I can see on:
>>
>> https://buildd.debian.org/status/logs.php?pkg=hdf5&ver=1.8.8-9&suite=sid
>>
>> So do we need the binnmu everywhere, or only on those architectures where the
>> default gfortran was bumped to 4.7, i.e. on amd64, i386, kfreebsd-amd64 and
>> kfreebsd-i386?
> 
> My mistake: I took for granted that gfortran was upgraded to 4.7
> on all architectures.
> 
> nmu hdf5_1.8.8-9 . amd64 i386 kfreebsd-amd64 kfreebsd-i386 . stable . -m "Rebuild with current gfortran in wheezy (closes: #739261)"
> 
> Hoping to get it right this time :/

Scheduled. This still needs to be accepted by the SRMs for the next point
release. Note that the "closes: #739261" doesn't work for binnmus, so you'll
have to close that manually.

Regards,
Emilio