[XviD-devel] [BUG] dev-api-4 -- Probably a vop padding mistake

[Christoph Lampert]

On Sat, 29 Mar 2003, Edouard Gomez wrote:

> Hello,
> 
> after spending some couples of hours to find a stupid namespace coflict
> with libavcodec (@gruel: yep the same you fixed in CVS_HEAD) i could
> finally produce my first AVI with dev-api-4.
> 
> But  mplayer+libavcodec were  reporting some  (small) slices  problems i
> used to have in early dev-api-3 when vops were not padded the right way.
> 
> <quote mode="repeated from time to time">
> slice end not reached but screenspace end (0 left 000000)
> concealing errors
> </quote>
> 
> The AVI (2.9MB) is available here:
> http://ed.gomez.free.fr/vrac/test.avi
> 
> It has been coded with vop/vol/motion flags set to zero.
> 
> If a padding expert could spend a few minutes on that problem. Thanks.

Simple question: Does dev-api-4 call  BitstreamPadAlways() at the same 
places where CVS_HEAD does? Does it call "...Always()" at all? 

I guess the rule I "implemented" in CVS_HEAD (dev-api-3) was:

Always call BitstreamPad() if padding is needed, execpt for at the
end of VOP (end of encode_something()) there PadAlways() is needed. 
And there is some other position, which I forgot, where PadAlways() is 
called as well.

Yes, very helpful, sorry.

gruel