Re: [hybi] Protocol simplicity and the "amateur programmer" standard

[Micheil Smith <[e-mail]>]

On:

> I think one browser shipping doesn't create an ecosystem that depends on WebSockets not undergoing incompatible changes (although I note that I have seen multiple people ask how to make their servers compatible with both -75 and -76 handshakes in order to work with both Safari 5 and Chrome). I think that many browsers shipping compatible implementations is highly likely to create such an ecosystem. Therefore I would say that the timescale for substantial changes to websockets is soemthing like the release timescale of the next generation of browsers minus the stabilization time and the time needed to implement changes. It seems to me that that timescale could be rather short.

The current way I do it in my project, Node WebSocket Server ( http://github.com/miksago/node-websocket-server ) 
is to switch between draft -75 and -76 based on the existence of the Sec-* headers and Key3. I also have an option 
where by a developer, if they so choose, can force the server to only use one version of the WebSocket protocol, 
and all other requests will be rejected (this is still not implemented quite right, but it's not really a super high priority 
on the project just yet).

Yours,
Micheil Smith
-- 
BrandedCode.com