[FFmpeg-devel] [PATCH] Chinese AVS encoder preparation

Måns Rullgård mans
Sat Jun 30 13:16:43 CEST 2007


Stefan Gehrer <stefan.gehrer at gmx.de> writes:

> Hi comrades,
>
> as somebody is already working on integrating an encoder for
> an unbeloved format and of disputable necessity (yes I am
> thinking of wmv3), I thought I might just add another one :)

I'm somehow less averse to AVS than WMV, so I say go for it if you
think it's fun.

> What I would like to do before discussing the encoder would
> be some cosmetic steps to make the encoding patch as simple
> as possible, in particular:
>
> - move structure definitions from cavs.c to cavs.h
> - move commonly used inline-functions from cavs.c to cavs.h
> - some variable renamings to better seperate decoder specifics
> - split off cavsdec.c from cavs.c
>
> A first step towards this is attached. If noone objects to
> more of these changes I would commit them without further ado,
> as they only touch what I maintain and are of cosmetic nature.
> Then I will come back to you with a proposed encoder patch.
> How does that sound?
>
> Stefan
> Index: libavcodec/cavsdata.h
> ===================================================================
> --- libavcodec/cavsdata.h	(revision 9451)
> +++ libavcodec/cavsdata.h	(working copy)
> @@ -106,7 +106,7 @@
>
>  /** marks block as unavailable, i.e. out of picture
>      or not yet decoded */
> -static const vector_t un_mv    = {0,0,1,NOT_AVAIL};
> +const vector_t ff_un_mv    = {0,0,1,NOT_AVAIL};

That's a very generic name that could easily conflict with something
from another module if made extern.  Perhaps you should consider
renaming it to something like ff_cavs_un_mv instead.

The rest of the shuffle looks OK to me.  I suppose Michael would want
you to svn cp some of the files though.

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




More information about the ffmpeg-devel mailing list