[FFmpeg-devel] [RFC] move H.263 parser to its own file

[Aurelien Jacobs]

On Wed, 9 May 2007 01:52:59 +0200
Diego Biurrun <diego at biurrun.de> wrote:

> On Tue, May 08, 2007 at 11:29:55PM +0200, Aurelien Jacobs wrote:
> > On Tue, 8 May 2007 10:57:50 +0200
> > Diego Biurrun <diego at biurrun.de> wrote:
> > 
> > > On Mon, May 07, 2007 at 12:36:49PM +0200, Michael Niedermayer wrote:
> > > > 
> > > > On Mon, May 07, 2007 at 10:01:33AM +0200, Diego Biurrun wrote:
> > > > > Attached patch does $subject.  It misses declaring a dependency on the
> > > > > H.263 parser for the H.263 decoder and a change to the Makefile, but I
> > > > > need to run out now.  Let me know if this is the way to proceed or not.
> > > > 
> > > > yes the way looks ok
> > > > 
> > > > [...]
> > > > > +
> > > > > +int h263_find_frame_end(ParseContext *pc, const uint8_t *buf, int buf_size);
> > > > 
> > > > ff_ prefix
> > 
> > It seems to me that moving ff_h263_find_frame_end() to h263_parser.c wasn't a
> > very good idea. This function is used as well by h263dec.c as by h263_parser.c.
> > h263dec.c is compiled unconditionnaly whereas h263_parser.c is compiled
> > conditionnaly.
> > So IMO, ff_h263_find_frame_end() should be moved back to h263dec.c.
> 
> What about fixing dependencies and conditionally compiling h263dec.c as
> well?

That would be nice. The problem is that almost every video codec has a
dependency (direct or indirect) over some code in h263dec.c.
Splitting it properly and fixing dependencies seems quite hard.
If you want to try to do it, that's great. Else I think that moving
ff_h263_find_frame_end() back to h263dec.c is the right thing to do.

Aurel