On 10/12/12 9:21 AM, Waqas Hussain wrote:
> On Fri, Oct 12, 2012 at 11:23 AM, Michael Weibel
> <[email protected]> wrote:
>> Hi,
>>
>>>
>>> How would we handle fallback? I suppose that's a good topic to discuss
>>> in the meeting, but it would also be good to know what our options are
>>> (e.g., how do you discover the WebSocket endpoint in the first place
>>> and how do you discover that you can fall back to BOSH?). XEP-0156
>>> might be relevant here.
>>
>> we're currently doing this manually, i.e. we're giving our javascript client 
>> both endpoints and this one looks if WebSockets are available in the browser.
>> When WebSockets are available it connects to the WebSockets endpoint. If he 
>> can't connect successfully after some time it will fallback to BOSH.
>>
>> This works not too bad although it could be improved more.
>>
>> What I'm wondering is: Could it be possible that you first connect to BOSH 
>> and afterwards check if Websockets is available and somehow upgrade the 
>> current connection?
>> This would speed up the whole process I guess while being still quite stable 
>> and reliable.
>>
> 
> Or if you care about user experience more than server load, just
> initiate both connections, and pick the first one to succeed :)

Right, that's the "happy eyeballs" approach.

https://datatracker.ietf.org/doc/rfc6555/

Peter

-- 
Peter Saint-Andre
https://stpeter.im/


Reply via email to