[xiph-rtp] GStreamer Vorbis/Theora RTP implementation

[Wim Taymans]

Good day!

Here is some feedback about the RTP drafts I gathered while implementing
GStreamer plugins for vorbis and theora RTP payloading and depayloading.
The following drafts were used:

 - Vorbis:
http://svn.xiph.org/trunk/vorbis/doc/draft-ietf-avt-rtp-vorbis-01.txt
 - Theora:
http://svn.xiph.org/trunk/theora/doc/draft-ietf-avt-rtp-theora-00.txt

The plugins interoperate fine. They were not yet tested against an
alternative implementation.

These are some observations:

*) Payload Length (Vorbis, Theora Figure 3, ...) Endianness is not
explicitly
   defined, I just assumed big endian...

*) Section 3: Configuration Headers. This was a little confusing, when
talking
   about the delivery methods (inline, in-band, out-of-band) the inline
method is
   not clearly described. For example:

    "For non chained streams, the RECOMMENDED Configuration
     delivery method is inline the Packed Configuration (Section 3.1.1) in
     the SDP as explained in the IANA considerations (Section 6.1)"

   but then section 3.1.1 and Figure 6 talks about the _in-band_ delivery method of the
   packed configuration. I think it was meant to point to Packed Headers (Section 3.2.1).

*) Vorbis 8.1 (and Theora 7.1) Stream Radio: 

    "When the streamed content changes the ....  has to be
     sent inline in the SDP updated.  Since the [inline] method is
     unreliable, an out of band fallback is provided."

   [inline] should be changed to in-band?
 
*) It's unclear what protocol/syntax one should/can/must support for the out-of-band
   method. Is it possible to fetch a specific Ident? If yes, how? Vorbis Section 6.1.1
   gives an SDP example:

    "The first configuration packet is inlined in the sdp, other configurations
     could be fetched at any time from the first provided uri [using] or all
     the known configuration could be downloaded using the second uri."

   Is this the rule? Is the first uri the currently streaming Ident? Is the second one
   the aggregated Idents?


Hope this helps, keep up the work!

Wim.
 

-- 
Wim Taymans <wim at fluendo.com>