Hi Josh, On Tue, 22 Jun 2010, Josh Allmann wrote:
004 breaks receiving video (mpeg4 AND h264) if streamed in conjunction with AAC, but video seems to work OK in isolation. Which is weird because 004 should only touch the AAC codepath.
Been staring at this bug all day, so I'm just gonna turn in what I have, get some sleep, and hopefully dream up a fix.
Whoa, this is good stuff. Can't wait to commit it :-) I found the missing thing that messes up streams with both audio and video - in the code branch in rtpdec.c for formats without a parse_packet function, the generic code sets pkt->stream_index = st->index, which you need to do in your dynamic payload handler. Except for that, patch #5 had some issues where you had forgotten to rename infos to data - I guess you've missed to commit something here? In the same patch, you've also got a case where you do if (ptr) av_free(ptr), which is't necessary, av_free() can handle that itself. Also, you don't need any new_context/free_context for mp4v_es, since it doesn't use that context. In order to do that, you need to make sure that you won't accidentally use the PayloadContext in parse_sdp_line unless the codec is AAC. In free_context, you should move the opening brace of the for loop. :-) I'm not sure about the copyright owner of this file, that you've set to "the FFmpeg developers" - is it possible to dig through the version history to see who wrote this initially? Apart from these details - I really like this, good work so far! // Martin