[FFmpeg-soc] [soc]: r385 - in dirac: dirac.c ffmpeg.diff

Aurelien Jacobs aurel at gnuage.org
Thu Jul 19 11:28:58 CEST 2007


On Thu, 19 Jul 2007 10:57:29 +0200
Marco Gerards <mgerards at xs4all.nl> wrote:

> Michael Niedermayer <michaelni at gmx.at> writes:
> 
> 
> Hi Michael,
> 
> > On Thu, Jul 12, 2007 at 12:34:28PM +0200, marco wrote:
> >> Author: marco
> >> Date: Thu Jul 12 12:34:28 2007
> >> New Revision: 385
> >> 
> >> Log:
> >> First attempt to get coefficients on the screen.
> >> 
> >
> > [...]
> >> @@ -544,10 +548,10 @@ enum arith_context_indices {
> >>  #define ARITH_CONTEXT_COUNT (ARITH_CONTEXT_DC_SIGN + 1)
> >>  
> >>  /* XXX: Check the spec again on this.  */
> >> -typedef int arith_context_t[1];
> >> -static arith_context_t arith_contexts[ARITH_CONTEXT_COUNT];
> >> +//typedef int arith_context_t[1];
> >> +static int arith_contexts[ARITH_CONTEXT_COUNT];
> >>  
> >> -static void arith_init (GetBitContext *gb, int length) {
> >> +static void arith_init (AVCodecContext *avctx, GetBitContext *gb, int length) {
> >>      int i;
> >>  
> >>      align_get_bits(gb);
> >
> > all the arithmetic coding stuff should be moved into its own file(s)
> > eventually ...
> 
> Yesterday I did so for my golomb code.  I plan to do this with the
> arithmetic decoding code, but I have some doubts on this.  I noticed
> that arithmetic decoding is at least used once in FFmpeg, in cabac.c.
> 
> Is the cabac.[ch] code generic enough to be used by me, or would you
> rather have the current code I wrote in a separate file?
> (dirac_arith.[ch] or so?)

As long as your code is only used by dirac, it should be in a dirac
specific file (dirac_arith.[ch] sounds like a good choice).

Aurel



More information about the FFmpeg-soc mailing list