- From: Martin Thomson <martin.thomson@gmail.com>
- Date: Thu, 30 Aug 2012 09:22:34 -0700
- To: Harald Alvestrand <harald@alvestrand.no>
- Cc: public-webrtc@w3.org
On 29 August 2012 23:00, Harald Alvestrand <harald@alvestrand.no> wrote: > Use of offer/answer as defined in RFC 3264 was not part of the poll. Thanks for the clarification. I could have inferred otherwise, and I know that others have. > Reasonable people can differ on whether any given refactoring is an > improvement or not. > > One standard approach to making refactoring possible is also to hide > internal interfaces as long as there's not a clear use case for exposing > them. The MS approach exposes more lower layer interfaces, thus locking in a > greater number of features to the presently proposed implementation > strategy. While it is true that features exposed in this fashion are more predictable, that is, in this instance, a benefit. Having reliable browser behaviour is necessary from a user perspective. And SDP results in significant uncertainty. We've discussed the use case aspects. There is more to it than that. The PeerConnection API does not support some of our existing use cases particularly well. Take stats as an example. Stats require that the tangle be unravelled into a parallel structure just to support reporting on individual flows or streams. That's a feature that is much easier with a more cleanly structured API. --Martin
Received on Thursday, 30 August 2012 16:23:08 UTC