- From: Stefan Hakansson LK <stefan.lk.hakansson@ericsson.com>
- Date: Fri, 29 Jun 2012 07:37:55 +0200
- To: Cullen Jennings <fluffy@iii.ca>
- CC: Dominique Hazael-Massieux <dom@w3.org>, Bjoern Hoehrmann <derhoermi@gmx.net>, "public-webrtc@w3.org" <public-webrtc@w3.org>
On 06/28/2012 06:38 PM, Cullen Jennings wrote: > > Objects that declare all their members in the constructors tend to > optimize better. They also have a much clearer intent of what we are > trying to represent here - logically speaking this is not a > dictionary though clearly a dictionary could be used as one way to > implement it. They also make it easier to add methods later - this > might be slightly controversial but given a bunch of people want it, > seems like we should at least keep that future door open for now. If > we were using any language other than JS which blurs the distinction > between objects and dictionaries, we would not even be having this > discussions and would have just declared a class or extensible > structure for this. > > Is there any concrete argument for dictionary over object for the > thing that holds the SDP? > > I normally don't care too much about the things like this that only > change the syntax and not the semantics of what is possible. But one > way or another, I'd like to wrap up this thread and move forward. Me too. I think it has been argued that dictionaries are simpler, fulfills all we need for a first phase at least and are future proof (http://lists.w3.org/Archives/Public/public-webrtc/2012Jun/0204.html) so perhaps that is what we should do. But most important is to decide and then move forward. > > > > > On Jun 27, 2012, at 2:09 , Stefan Hakansson LK wrote: > >> On 06/27/2012 08:15 AM, Dominique Hazael-Massieux wrote: >>> Le mardi 26 juin 2012 à 20:20 +0200, Bjoern Hoehrmann a écrit : >>>> Well, PeerConnection.localDescription and .remoteDescription >>>> would have to be changed aswell since they are >>>> SessionDescription attributes, and attributes cannot be >>>> dictionaries in WebIDL. >>> >>> Good point, I had missed that; I guess they would have to be >>> declared as "object" with their actual structure defined in the >>> prose. >> >> So would not that be the same as what the >> SessionDescriptionCallback is provided with, e.g. an object à la >> {sdp:"m=lotsofmumbo", type:"offer"}? >> >>> >>> Dom >>> >> >> >> >
Received on Friday, 29 June 2012 05:38:22 UTC