[FFmpeg-devel] [PATCH] Check for signal.h, conditionally compile parts of ffmpeg.c

[Måns Rullgård]

?smail D?nmez <ismail at namtrac.org> writes:

> Hi,
>
> 2009/7/29 M?ns Rullg?rd <mans at mansr.com>:
>> Martin Storsj? <martin at martin.st> writes:
>>
>>> On Tue, 28 Jul 2009, Ramiro Polla wrote:
>>>
>>>> On Fri, Jul 17, 2009 at 4:48 AM, Martin Storsj?<martin at martin.st> wrote:
>>>> >
>>>> > Here's another update to the patch series; I noticed that not all
>>>> > HAVE_SIGNAL_H checks in ffmpeg.c were needed.
>>>>
>>>> Is it necessary to check for HAVE_SIGNAL_H in the #ifdef HAVE_TERMIOS_H part?
>>>
>>> No, not for my use case. If termios.h existing implies signal.h existing
>>> too, I guess the check is redundant.
>>
>> signal.h is a mandatory part of standard C, and if that's not enough
>> to ensure it exists everywhere, well...
>
> If all operating systems were POSIX compliant we wouldn't need OS
> specific hacks like those in os_support.c . The whole WinCE thing puts
> a bad taste in my mouth too but if we can work around it easily lets
> do it.

I don't mind (much) if an OS isn't POSIX compliant, but that's not
what this is about.  The signal.h header and signal() function are
standard ISO C, not a POSIX addition.  There is no excuse for not
having a complete standard C library.

That said, I'm not against workarounds where these are simple.
Indeed, we have a fair number already, mostly for missing math.h
stuff.  However, we should not add workarounds for standard things
without first notifying, where feasible, those responsible for the
system in question.

-- 
M?ns Rullg?rd
mans at mansr.com