Re: [AVT] draft-kerr-avt-vorbis-rtp-04

[Phil Kerr <[e-mail]>]

Hi Colin,

Colin Perkins wrote:

> Hi,
>
> On 4 Jan 2005, at 15:54, Phil Kerr wrote:
>
>> An -04 update of the Vorbis-RTP draft was submitted a few days ago.
>>
>> http://www.ietf.org/internet-drafts/draft-kerr-avt-vorbis-rtp-04.txt
>>
>> Major changes include:
>>
>> Removal of RTCP-based transport for configuration headers, replaced 
>> with in-band transport.
>> Inclusion of a CRC32 header field used to tie the stream to the 
>> associated decoding codebook.
>
>
> This is an interesting idea. How is the CRC32 calculated? What is the 
> chance of collision, and how would that be resolved?

The CRC32 is calculated from the codebook header itself.  The chance of 
collision should be quite low, MD5 would be a lot safer in this regard 
but it does add quite a lot of overhead to each packet.  There may need 
to be an additional identifier, possibly a host id which ties the CRC32 
value to the host, which is sent via SDP to reduce the chance of 
collision further.

>
>
>> Simplification of SDP parameters.
>
>
> These certainly look reasonable. However, we try to avoid mandating 
> SDP (since there may be other session description formats). Can you 
> move the description of the parameters into the MIME registration, and 
> make this section describe how they can be mapped onto SDP (for those 
> applications which use SDP). Section 5 of RFC 3557 is a good example 
> of what is needed.

Will do.

>
>> Plus a number of textual changes.
>>
>> Comments/feedback welcomed.
>
>
> A few issues to consider:
>
>  - Why is the payload length only an 8 bit field? It seems that this 
> might
>    force fragmentation when it's not otherwise needed. Could one use a 16
>    bit field instead?

The 8 bit field is a historical feature from the original Vorbis I-D.  
Changing it to 16 bits was raised on the Xiph-RTP list just after the 
latest version was submitted and if it doesn't cause any other issues, 
such as causing an inordinate number of packets to exceed path MTU, I 
think it can be altered

>
>  - Providing a URI for the configuration header is appropriate, but the
>    draft needs to define the syntax of that URI (e.g. by referencing an
>    appropriate specification, perhaps RFC 2396). It also needs to define
>    the legal MIME types for the data referenced, so the format is known.
>    Is there an existing MIME type for this?
>
>  - Need to discussion offer/answer considerations when using SDP. See
>    section 5 of RFC 3952 for an example.

Will do.

Cheers

Phil

>
> Cheers,
> Colin
>
>
>


_______________________________________________
Audio/Video Transport Working Group
[e-mail]
https://www1.ietf.org/mailman/listinfo/avt