Re:WCF WebSockets
by davidpadbury on Thursday January 13, 2011 @09:44AM
(#2099897)
The WebSocket protocol is still very much in flux. This drop is only to give an idea of the server side API would look like (I think anyways). I imagine as the protocol calms down they'll implement whatever's decided on.
[ Parent
| on HN ]
I'm not sure if you'd really call it 'in flux'.
We've been using websocket for over a year now on Mibbit. I think around 20% of our users currently do websocket.
Fair enough to wait and see though, hopefully they'll do it the same way everyone else has.
Re:WCF WebSockets
by davidpadbury on Thursday January 13, 2011 @10:22AM
(#2100033)
By "in flux" I was intending to vaguely refer to the problems with the protocol which has led Firefox and Opera to ship with WebSockets disabled. I'm guessing we're probably going to see some changes to it in the future. Hopefully sooner rather than later.
I honestly don't know a thing about the version of the protocol that they've implemented. It looks like one that they (Microsoft) have submitted - http://tools.ietf.org/html/draft-montenegro-hybi-upgrade-hel....
That's cool you're seeing so many clients with WebSockets enabled.
Ah true, I'd forgotten about the firefox and opera hoo-haw. Hopefully they sort that out ASAP It's a great shame really.
Re:WCF WebSockets
by davidpadbury on Thursday January 13, 2011 @10:39AM
(#2100097)
Yeah - A "hoo-haw" is good. The quicker issues are brought to attention and resolved the better.
FWIW I think it was an absolute edge case bordering on silliness.
if there is a crappy silly caching proxy, an attacker can use websocket to get it to cache things it shouldn't.
I think the fact Chrome hasn't done anything about it, (especially since they can auto update and turn off support tomorrow if they wish), is telling.