Re: [hybi] Framing take IV

[Greg Wilkins <[e-mail]>]

Jamie,

Interesting idea.  I think we had assumed that negotiating extension opcodes
would be too complex, but what you have suggested looks like the start of a
simple solution.   Could you care to expand on it a bit more with exactly
how the client/server would agree on a final set of extended op-codes?

So it looks now like we have  4 possible extension approaches:

hixieasque - extension via defined opcodes
jamiesque - extension via negotiated opcodes
robertoesque - extension via per frame meta-data
martineque - extension via a dedicated mime frame


cheers



On 4 August 2010 12:27, Jamie Lokier <[e-mail]> wrote:

>
> 4. Extension codes identified at handshake negotiation time, or a
> later negotiation time in control messages.
>
> For example: "WebSocket-Protocol: x-multiplex;control-byte=12,
> keepalive-aggregator;control-byte=13"
>
> Doing it at handshake time has the advantage that it's easily
> future-proof, very little needs to be defined up front, and simple
> implementations can ignore them entirely.
>
> -- Jamie
>