[FFmpeg-devel] [PATCH] remove MSVC cruft

Måns Rullgård mans
Tue Feb 12 01:06:43 CET 2008


Diego Biurrun <diego at biurrun.de> writes:

> On Sat, Feb 09, 2008 at 08:10:05PM +0100, Diego Biurrun wrote:
>> On Sat, Feb 09, 2008 at 02:51:53PM +0100, Michael Niedermayer wrote:
>> > On Sat, Feb 09, 2008 at 02:09:04PM +0100, Diego Biurrun wrote:
>> > > On Sat, Feb 09, 2008 at 02:06:01PM +0100, Diego Biurrun wrote:
>> > > > As noted by Reimar, the following two lines in libavutil/mem.h are
>> > > > probably MSVC cruft:
>> > > > 
>> > > >   #define DECLARE_ALIGNED(n,t,v)      __declspec(align(n)) t v
>> > > >   #define DECLARE_ASM_CONST(n,t,v)    __declspec(align(n)) static const
>> > > > 
>> > > > I suggest the attached patch removing them.
>> > > 
>> > > *sigh*
>> > 
>> > I am against it.
>> > 
>> > it rather should be
>> > #elif HAVE_DECLSPEC
>> > #define DECLARE_ALIGNED(n,t,v)      __declspec(align(n)) t v
>> > #define DECLARE_ASM_CONST(n,t,v)    __declspec(align(n)) static const t v
>> > #else
>> > #warning no align and asm directives, this might fail
>> > #define DECLARE_ALIGNED(n,t,v)      t v
>> > #define DECLARE_ASM_CONST(n,t,v)    static const t v
>> > #endif
>> > 
>> > There are people who maintain a hacked up version of ffmpeg which does
>> > compile under msvc. Theres no need to make their work harder by removing
>> > clean and seperated code. Its only the messy parts which we should reject.
>> 
>> These two lines do not help them.  Also, this code is not clean and
>> separated, it is the fallback when __GNUC__ is not defined.  So anybody
>> trying compiler X will run into problems.  I don't see anybody coming
>> up with a proper configure check.  Until then these lines do more harm
>> than good and should IMO be removed.
>
> May I insist here?  The fallback case should definitely not be MSVC,
> i.e. an unsupported and non-standards-compliant compiler.

I agree.  Besides, are those macros used by any code that doesn't
already require gcc-style inline asm?

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




More information about the ffmpeg-devel mailing list