Re: [hybi] Call for interest: multiplexing dedicated for WebSocket

[Bruce Atherton <[e-mail]>]

I agree. During the planning stages of the protocol there was a great 
deal of interest expressed in stand-alone WebSocket servers that were 
independent of any HTTP server. That has not been the real world 
deployment behaviour, at least in my experience. So long as HTTP/2.0 
doesn't introduce so much added complexity that it becomes too costly to 
include a server for it when you are deploying WebSockets, I don't see 
much benefit to a WebSockets-only mux definition any more, much though I 
appreciate the effort that Takeshi and John have put into it.

On 19/05/2013 11:37 AM, Simone Bordet wrote:
> My impression is that I have not seen much request for WebSocket multiplexing.
> Given the complexities related to defining and implementing ws mux
> correctly, and considering that HTTP/2.0 comes with mux built-in,
> building on the SPDY experience, my preference would be to
> deprioritize ws mux in favour of HTTP/2.0 integration.
> It is true that there may be systems that may rely on a pure-ws
> communication (and therefore do not need a full HTTP/2.0
> implementation) and as such would benefit from ws mux, but my guess is
> that those system will be a minority or can get by without ws mux. I
> may be biased on this view though, so comments welcome.
>
>