[FFmpeg-devel] [PATCH] Realmedia / RTSP (RDT)
Diego Biurrun
diego
Thu Jan 3 00:00:28 CET 2008
On Wed, Jan 02, 2008 at 11:50:50PM +0100, Luca Abeni wrote:
>
> Diego Biurrun wrote:
> [...]
>>> I do not know how the dependencies between muxers and demuxers are
>>> organised, but... This looks very funny.
>>> As far as I know, the dynamic payload handlers are used for receiving RTP
>>> traffic, not for sending it... So, I guess this should be
>>> #ifdef CONFIG_RTSP_DEMUXER
>>> av_register_rtp_dynamic_payload_handlers();
>>> #endif
>> Test and commit the fix then :)
>
> Well, as I said I do not know how dependencies are organised, so my patch
> can be wrong... I prefer to post it here for some review instead of
> directly committing it.
> Anyway, I had a look at these dependencies, and I now understand why the
> rtsp demuxer currently depends on the RTP muxer: this dependency works in
> practice, but looks strange from the logical point of view.
> The attached rtp_dependencies.diff patch seems to work well, and I think
> that the dependencies now look more logical. What's your opinion on it?
Looks correct, if it works, apply.
> After applying the patch, there still are some strange dependencies (for
> example, the RTP muxer depend on the TS demuxer, rtp_h264.c is compiled
> when the RTP muxer is enabled, etc...), but they can be removed with a
> second patch (for example, by adding some "#ifdef"s as in the attached
> rtp_mux_demux.diff patch).
Also looks correct, if it works, apply.
> --- libavformat/Makefile.orig 2008-01-02 23:29:02.000000000 +0100
> +++ libavformat/Makefile 2008-01-02 23:41:51.000000000 +0100
> @@ -122,8 +122,7 @@
> OBJS-$(CONFIG_ROQ_DEMUXER) += idroq.o
> OBJS-$(CONFIG_ROQ_MUXER) += raw.o
> OBJS-$(CONFIG_RTP_MUXER) += rtp.o rtp_h264.o rtp_mpv.o rtp_aac.o
> -OBJS-$(CONFIG_RTSP_DEMUXER) += rtsp.o
> -OBJS-$(CONFIG_SDP_DEMUXER) += rtsp.o
> +OBJS-$(CONFIG_SDP_DEMUXER) += rtsp.o rtp.o rtp_h264.o rtp_mpv.o rtp_aac.o
Umm, yes, your patch makes the RTSP demuxer depend on the SDP demuxer,
but still I think we should keep that line here.
Diego
More information about the ffmpeg-devel
mailing list