[FFmpeg-cvslog] r20950 - in trunk/libavcodec: h263.h mpegvideo_enc.c

[Michael Niedermayer]

On Fri, Jan 08, 2010 at 01:52:01AM +0100, Diego Biurrun wrote:
> On Thu, Jan 07, 2010 at 04:19:19AM +0100, Michael Niedermayer wrote:
> > On Tue, Jan 05, 2010 at 11:10:55PM +0100, Diego Biurrun wrote:
> > > On Mon, Jan 04, 2010 at 05:11:10PM +0100, Michael Niedermayer wrote:
> > > > On Mon, Jan 04, 2010 at 04:42:37PM +0100, Diego Biurrun wrote:
> > > > > On Fri, Jan 01, 2010 at 09:37:57PM +0100, Michael Niedermayer wrote:
> > > > > > On Mon, Dec 28, 2009 at 06:36:03PM +0100, diego wrote:
> > > > > > > 
> > > > > > > Log:
> > > > > > > Replace CONFIG_ANY_H263_ENCODER definition by CONFIG_H263_ENCODER.
> > > > > > > CONFIG_ANY_H263_ENCODER is just a logical OR of H.263 family definitions,
> > > > > > > for which the build system enables CONFIG_H263_ENCODER anyway.
> > > > > > 
> > > > > > This is not ideal
> > > > > > the common code that any h263 encoder uses is less than what h263
> > > > > > encoding as such uses. For example there are things like h263 loop filtering
> > > > > > b frames (with motion estimation and all) OBMC, ... in h263 but not in h263
> > > > > > like codecs like msmpeg4
> > > > > 
> > > > > I see no direct relation to the content of my commit.
> > > > 
> > > > the relation is that you removed CONFIG_ANY_H263_ENCODER and that is the
> > > > define which should be used for the core h263 encoder code.
> > > > As it is now you must define h263_encoder and thus must pull all the h263
> > > > encoder code in for any codec that uses a subset of the h263 encoder code.
> > > 
> > > Umm, I can see that this would be an improvement if somebody were to
> > > actually implement it.  However, it is not a description of the state
> > > of things before my change.  I made no semantic changes.
> > 
> > You removed code that is a prerequestie to spliting the core h263 routines
> > from the rest.
> > Please put the code related to any_h263 back
> 
> If you need it for your current work, fine.  But it was not used before
> and the proper place to define it is configure.

i dont need it, iam not trying to make things more modularly compileable
but maybe ill split things further and this should make it alot easier
if someone wants to make things more modularely compileable.

that "any" thing could come in handy there somewhere


-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Avoid a single point of failure, be that a person or equipment.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-cvslog/attachments/20100108/c141991e/attachment.pgp>