RE: [rtcweb] e= lines (Re: Summary of Application Developers' opinions ..)

1. RFC 4566 doesn't apply (at least not yet) through the chain of references as the WEBRTC specification references "SDP" as RFC3264, which itself references RFC2327. Now RFC4566 happens to obsolete 2327 and 3266, but 3264 doesn't make note of that fact (and hasn't itself been obsoleted by something that does) nor does the WEBRTC specification. I don't believe it to be appropriate for the W3C specification to rely on a complex unraveling of the obsoletions and updates over in IETF... it should instead say exactly which specifications and which versions of those specifications the API is expected to understand.

2. Yes, I think that each line that can appear in the SDP needs to be addressed *in the W3C specification*... at the very least with a blanked statement "if it wasn't mentioned here, it shall be ignored", but we don't even have that (because the W3C specification doesn't go into detail about any of the lines that can appear in the text-form API it has chosen).

3. I sent a long list of things that could be modified about the SDP and for which the results were underspecified. You have picked the easiest one from that list. The working group will need to address all of them, especially the most difficult, before there is a specification that a browser vendor can independently implement. Perhaps you'd like to take a stab at one of the others?

Matthew Kaufman


________________________________________
From: Harald Alvestrand [harald@alvestrand.no]
Sent: Wednesday, July 17, 2013 10:42 PM
To: Matthew Kaufman (SKYPE)
Subject: Re: [rtcweb] e= lines (Re: Summary of Application Developers' opinions ..)

On 07/18/2013 07:10 AM, Matthew Kaufman (SKYPE) wrote:
> But there is no spec for webrtc saying what the browser should or should not do with an e= line in each of the places where SDP is a parameter. When there is, we can move on to the next thing that isn't specified.

The SDP specification actually answers this, by omission (RFC 4566
section 5.6).

There is no spec anywhere that says that anything should be done with an
e= line.
So the browser should conform to that spec, and do nothing with it.

Of course, if you choose to represent everything that *is* actually
defined in an SDP specification as undefined for the browser use case,
you will have a lot of undefineds to cover.

But I don't think that's a reasonable way to approach the problem.

Received on Thursday, 18 July 2013 15:06:25 UTC