[Ffmpeg-devel] VP3 decoding issue

[Michael Niedermayer]

Hi

On Thu, Dec 01, 2005 at 03:00:37PM +0100, Alexander Strasser wrote:
> Hi,
> 
> Reimar D?ffinger wrote:
> > On Mon, Oct 31, 2005 at 07:56:51PM +0100, Alexander Strasser wrote:
> > >   Does the attached patch fix the crash?
> > > It would also fix the timing macros for some more platforms
> > > (mingw and cygwin) but i don't know if it breaks some others
> > > too.
> > > 
> > >   Alex (beastd)
> > 
> > > Index: libavutil/common.h
> > > ===================================================================
> > > RCS file: /cvsroot/ffmpeg/ffmpeg/libavutil/common.h,v
> > > retrieving revision 1.153
> > > diff -u -r1.153 common.h
> > > --- libavutil/common.h	19 Sep 2005 23:26:47 -0000	1.153
> > > +++ libavutil/common.h	31 Oct 2005 17:31:22 -0000
> > > @@ -512,7 +512,7 @@
> > >    }else\
> > >        tskip_count++;\
> > >    if(256*256*256*64%(tcount+tskip_count)==0){\
> > > -      av_log(NULL, AV_LOG_DEBUG, "%Ld dezicycles in %s, %d runs, %d skips\n", tsum*10/tcount, id, tcount, tskip_count);\
> > > +      av_log(NULL, AV_LOG_DEBUG, "%"PRIu64" dezicycles in %s, %d runs, %d skips\n", tsum*10/tcount, id, tcount, tskip_count);\
> > 
> > I tested this kind of solution quite extensively for MPlayer, it should
> > work on all systems that have inttypes.h - if this requirement is okay
> > for ffmpeg, I think this should be applied.
> 
>   Can I go ahead and commit this?
>   Has anyone tested it on more platforms?

lets see, if you ask people to test they have to manually apply this and try
if OTOH you just commit it its less work for you, less work for the others
so waiting with such things which noone knows if they will break on some obscure
system is IMHO stupid, its also a bugfix, the current code is just wrong so we
have wrong code which does break on some systems but noone dares to commit the
correct code just because it might break on some obscure system, what weird 
logic ...

[...]
-- 
Michael