Re: [rtcweb] New Version Notification for draft-jesup-rtcweb-data-01.txt

[Harald Alvestrand <[e-mail]>]

On 11/14/2011 07:20 PM, Wolfgang Beck wrote:
> On Mon, Nov 14, 2011 at 6:42 PM, Justin Uberti<[e-mail]>  wrote:
>> On Mon, Nov 14, 2011 at 4:35 AM, Wolfgang Beck
>> <[e-mail]>  wrote:
>>> If not, how is the content of the data stream -- the 'codec'
>>> -- negotiated between the different JS clients? ROAP will not work as
>>> it is the JS clients which writes to the data stream. The ROAP
>>> component of the browser doesn't know what's in there.
>> The format of the data stream will be standardized.
> Now it gets interesting: who is responsible for the rtcweb-data content?
>
> if it is the browser:
> - every new application will have to be standardized.
> - for every new application, the browser has to be updated.
> ***whether we do server-to-server or not***
>
> It basically means every rtcweb-data application needs to get the
> approval of Firefox, Google, and Microsoft.
>
> If it is the JS client:
> - the JS client has to handle Offer/Answer. The browser doesn't know
> enough about the content to do this.
I believe the discussion above bears very little relationship to what 
has been proposed.
As I read the proposal:

- Data channel does not touch the servers.
- Of course two JS apps can only have sensible communication if they 
agree on the data format. Being the same application is one way to 
agree; it's not the only one.
- The data format (content of data) is not negotiated in offer/answer. 
Just the fact that there is a data channel is negotiated.
- Standardizing a data format may be useful. But it is not required. 
Bilateral agreements work too.