Marek
> In the send i think it boils down to exposing callback both on server and client side on the first /info request-response handlers. That's a neat idea. The problem...
Also, we should test what happens when one sends a message if the session is in 'closed' state. The message should be discarded, but the 'send' shouldn't raise 404.
You're free to port it to new cowboy and merge some issues. I'm happy to link to your fork.
There are at least three issues with current implementation: - We do hibernate cowboy websocket process, but we don't hibernate sockjs_session process - We don't hibernate cowboy xhr processes -...
weren't there any major API changes between cowboy 0.8.3 and 0.9? I thought there were.
Have you tried running sockjs-protocol tests against the server?
Interesting! Well, the idea behind websocket-multiplex is that you can have more than one emulated 'websocket' connection over a single real/sockjs 'websocket' connection. But emulated 'websocket' still behaves like a...
"name" of the channel must be unique, this is a design decision of websocket-multiplex, and as you describe it now I see that it's probably wrong. But, please note that...
btoa in visual basic for ie: https://github.com/jseidelin/binaryajax/blob/master/binaryajax.js
Bad news: `This release drops R14 compatibility`. I can't do that.