[FFmpeg-devel] [RFC] Separating the RTSP muxer/demuxer and SDP demuxer

Martin Storsjö martin
Thu Oct 21 14:27:47 CEST 2010


On Tue, 19 Oct 2010, Martin Storsj? wrote:

> I haven't moved forward with the splitting of the RTSP code yet, since I 
> wasn't really totally convinced that it was a good move to split all the 
> code into a lot of small pieces when it actually still is quite tightly 
> coupled.
> 
> The new proposed patchset, attached, achieves most of the benefits of the 
> previous suggestion without actually moving things around as much. 
> Everything that currently is in rtsp.c still is in rtsp.c, but things are 
> grouped into logical pieces surrounded by one #if block each.
> 
> From the top, we have:
> - Common code used by all code paths
> - SDP parsing, used by RTSP and SDP demuxers (grouped into a CONFIG_RTPDEC 
> item, as in the previous patch set)
> - RTSP protocol code
> - Code for receiving RTP packets
> - RTSP demuxer specific functions + entry points
> - SDP demuxer specific functions + entry points
> 
> Choosing this approach doesn't exclude future splitting into separate 
> files, of course, this is just a smaller initial version that is easier to 
> accept.
> 
> With this in place, --disable-everything + --enable-demuxer=rtsp, 
> --enable-demuxer=sdp or --enable-muxer=rtsp only enable and compile those 
> parts that actually are needed.

Since most seemed to agree that some sort of cleanup/split is needed, I 
went ahead and applied these patches. They should be quite 
uncontroversial.

Now the code is quite neatly set up in rtsp.c for splitting into smaller 
.c files, which can be done incrementally. That way, it is hopefully much 
easier to oversee, compared to the earlier patch I sent, where everything 
was split in one single commit.

// Martin



More information about the ffmpeg-devel mailing list