Bug#701710: libav: FTBFS on powerpcspe

[Reinhard Tartler]

On Wed, Feb 27, 2013 at 5:53 PM, Roland Stigge <stigge at antcom.de> wrote:
> Hi Reinhard,
>
> On 02/27/2013 05:47 PM, Reinhard Tartler wrote:
>> On Wed, Feb 27, 2013 at 5:41 PM, Roland Stigge <stigge at antcom.de> wrote:
>>> Hi Reinhard,
>>> On 02/27/2013 02:57 PM, Reinhard Tartler wrote:
>>>> I don't understand this. Why does gcc on powerpcspe enable altivec by
>>>> default? If powerpcspe ships a gcc that has broken default settings,
>>>> then this should be fixed first.
>>>
>>> I had a second look at this, and it turned out that it's actually a bug
>>> in upstream's configure.
>>
>> Please propose a patch to fix that instead of changing debian/rules then.
>
> In the last mail (that you cite) there is a new patch attached. This
> patch alone fixes the issue completely (an upstream bug), without
> changes to debian/* at all.

Oh, I'm sorry. No I've seen it. Sorry for the confusion, I will
forward it to upstream shortly.

>>> Will file a similar upstream fix to the mplayer2 package/bug.
>>
>> You did not answer my question/suggestion at all. Can you please
>> comment on my suggestion to fix the gcc package?
>
> I don't consider this a gcc bug. It's working fine in the powerpcspe
> scene for many years now. You should just refrain from explicitly
> activating "-maltivec" and everything is fine.

Ah, so you say that libav's configure need to check if the compiler
actually supports altivec before forcefully enabling it. That sounds
like a very valid feature enhancement.

Again, sorry for the confusion and thanks for your patch!


-- 
regards,
    Reinhard