Hi again, after the discussion the last few days, I started doing some coding and first of all have a first very simple, quick and dirty demo client: http://www.j-ogg.de/rtp/index.html During the coding, I've run into a few problems: - I was planning to replace the identification header with appropriate entries in the SDP descriptor. However, the frame sizes are defined here (why not in the setup header) and if we'll allow different codebooks (and then likely different frame sizes), there must be a way to get these together with the new setup header. - The flags in the audio packets describing the frame length (short or long) for the previous and next packets are problematic when handling packet loss. - Streaming CMML as a complete XML document is difficult enough for unicast transmission (packet loss handling) and impossible for multicast purposes. Would it be a feasible solution to drop the document header <?xml...><cmml> and footer </cmml> and demand RTP packets to contain separate entities, which may be decoded as standalone XML snippets (e.g. just one clip or title element in one RTP packet)? <p>Tor --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
On Sun, May 30, 2004 at 02:34:03AM +0200, Tor-Einar Jarnbjo wrote:> > - Streaming CMML as a complete XML document is difficult enough for > unicast transmission (packet loss handling) and impossible for multicast > purposes. Would it be a feasible solution to drop the document header > <?xml...><cmml> and footer </cmml> and demand RTP packets to contain > separate entities, which may be decoded as standalone XML snippets (e.g. > just one clip or title element in one RTP packet)? >yes, that's the right thing to do -- that's also how cmml is encapsulated in an ogg stream for the same reason :) Conrad. --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
On Sun, May 30, 2004 at 02:34:03AM +0200, Tor-Einar Jarnbjo wrote:> - I was planning to replace the identification header with appropriate > entries in the SDP descriptor. However, the frame sizes are defined here > (why not in the setup header) and if we'll allow different codebooks > (and then likely different frame sizes), there must be a way to get > these together with the new setup header.I'd guess it's done this way because the blocksizes are useful for measuring stream length, which doesn't require access to the codebooks.> - The flags in the audio packets describing the frame length (short or > long) for the previous and next packets are problematic when handling > packet loss.Can you elaborate of this? Surely the packet sequence number informs you when there's a hole in the data? In that case you need to skip the overlap with the previous packet anyway.> - Streaming CMML as a complete XML document is difficult enough for > unicast transmission (packet loss handling) and impossible for multicast > purposes. Would it be a feasible solution to drop the document header > <?xml...><cmml> and footer </cmml> and demand RTP packets to contain > separate entities, which may be decoded as standalone XML snippets (e.g. > just one clip or title element in one RTP packet)?What about putting the whole CMML document in a single packet? Way cool on the implementation effort! -r --- >8 ---- List archives: http://www.xiph.org/archives/ Ogg project homepage: http://www.xiph.org/ogg/ To unsubscribe from this list, send a message to 'vorbis-dev-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.