RE: Microsoft API Proposal

There will ultimately be a specification produced by the Working Group. Whether that is done by making piecewise edits to an existing document or wholesale substitution of all or part of an existing document, it is still "successive refinement".

Yes, a core feature of the proposal is to replace the amalgamation of functionality on "PeerConnection" with an architecturally cleaner set of components, and another core feature is to replace both SDP as a description and the SDP Offer/Answer mechanism.

These are still key things that we are proposing... in addition to many other changes, some major and some minor that have varying levels of existing support among other members of the WG, including yourself.

There is nothing about our submission that "insists" that, for instance, SDP be removed if we are to expose an API for determining when congestion is occurring. We believe that both of these changes should happen, but they are independent. (And of course this is the case... just because our feedback about the current specification is presented in the form of a specification does not mean that our feedback must be "take it or leave it".)

But in any case, our proposal can be read as a long list of proposed changes... some major, some minor, and each of those differences should be addressed and discussed. I suggest that we open the appropriate issues in the issue tracker.

Matthew Kaufman

-----Original Message-----
From: Harald Alvestrand [mailto:harald@alvestrand.no] 
Sent: Tuesday, August 28, 2012 12:25 AM
To: public-webrtc@w3.org
Subject: Re: Microsoft API Proposal

Matthew, thanks for this reformulation!

It seems to me that a lot of the points you make below are already on the table in different contexts, many of which have languished because they were not being pushed while people were off implementing the stuff required for interoperation. (at the moment, we have people reporting interworking between Chrome and Asterisk + some other "legacy" devices).

When I read the Microsoft proposal first, it seemed to me that a core feature of the proposal was 3 "no"s:

- No PeerConnection object (replaced with a cloud of transport-related
entities)
- No SDP description (replaced with a cloud of APIs on those transport-related entities)
- No mandatory-to-implement codecs (not replaced with anything)

It so happens that I (as a technical contributor) disagree with all 3 of those proposal components; if you're happy to work on a successive refinement of current proposals without necessarily insisting that those
3 removals be done, I'm sure we'll find easier ways forward.

             Harald

Received on Tuesday, 28 August 2012 13:41:40 UTC