- From: Suhas Nandakumar (snandaku) <snandaku@cisco.com>
- Date: Wed, 3 Jul 2013 07:10:18 +0000
- To: tim panton <thp@westhawk.co.uk>
- CC: Iņaki Baz Castillo <ibc@aliax.net>, Ted Hardie <ted.ietf@gmail.com>, Christer Holmberg <christer.holmberg@ericsson.com>, cowwoc <cowwoc@bbs.darktech.org>, Robin Raymond <robin@hookflash.com>, "Roman Shpount" <roman@telurix.com>, Adam Bergkvist <adam.bergkvist@ericsson.com>, "piranna@gmail.com" <piranna@gmail.com>, "public-webrtc@w3.org" <public-webrtc@w3.org>, Eric Rescorla <ekr@rtfm.com>
Sent from my iPhone On Jul 3, 2013, at 12:02 AM, "tim panton" <thp@westhawk.co.uk> wrote: > > On 3 Jul 2013, at 00:51, "Suhas Nandakumar (snandaku)" <snandaku@cisco.com> wrote: > >> >> ________________________________________ >> From: Iņaki Baz Castillo [ibc@aliax.net] >> Sent: Tuesday, July 02, 2013 4:37 PM >> To: Ted Hardie >> Cc: Christer Holmberg; cowwoc; Robin Raymond; Roman Shpount; Adam Bergkvist; piranna@gmail.com; public-webrtc@w3.org; Eric Rescorla >> Subject: Re: VS: Teleco Integrators vs Web Developers vs Browser Implementers >> >> 2013/7/3 Ted Hardie <ted.ietf@gmail.com>: >>> The question we're trying to answer is: >>> >>> How do peers agree on the set of data sources to be exchanged, send them to >>> each other via a browser, and understand how to associate the received flows >>> with application behaviour? >> >> Hi Ted, I replied to this in my mails yesterday: >> >> Instead of passing SDP between JS and WebRC stack, use a real JS >> Object based API. >> Instead of sending an opaque SDP blob in-the-wire send the media info >> in any custom format designed by the website developer. Anyone is fine >> if the receiver JS can de-serialize/decode/parse it, construct the >> requires JS Object(s) with it, and pass it to the WebRTC stack via a >> JS API. >> >> just curious, how does the one define the agreed upon structure to understand the semantics of the data stored in such a data structure passed from one end-point to the another ... > > At the standards level you don't . > > The W3C standard describes what the browser API looks like to the javascript developer. > > It defines what the developer has to do to set up a peer-connection with a set of desired properties. > The way the javascript programmer determines what those desired properties are is up to her. > Likewise the way the desired properties are communicated from one end to the other(s) can be > independent of the API. > > Those properties will be derived 5 places : > 1) from the local user's wishes, > 2) from the capabilities of the local hardware, > 3) from the service's policies, > 4) from the remote hardware capabilities, > 5) from the remote users wishes > > Our current API surface (SDP) is good at expressing 2) and 4) but weak at expressing the other sources, especially 3) > > Tim. > Hello Tim I would be interested in learning about minimally extending existing API or constraints to solve the use cases that aren't covered by 1 and 3. Also 5 is can be covered within existing API and data model semantics My 2 cents Cheers Suhas
Received on Wednesday, 3 July 2013 07:11:13 UTC