Re: [hybi] Comments about draft-13

[Greg Wilkins <[e-mail]>]

John

if we interpret multiple protocols in the same way that we interpret
multiple extensions, then it does make sense.   Ie there is an
ordering of protocols, so if you had "oauth","chat"  then the oauth
protocol would be given the messages first and it could consume them
and generate replies before potentially passing messages to the chat
protocol.

Of course not all protocols will be additive in this way, but i think
it is still a needless restriction to have only 1.

But then it's not a big issue and unless everybody says +1 now, it's
not worth holding up the spec for.

cheers




On 8 September 2011 10:41, John Tamplin <[e-mail]> wrote:
> On Wed, Sep 7, 2011 at 8:37 PM, Greg Wilkins <[e-mail]> wrote:
>>
>> The only thing I think is missing is the ability to speak multiple
>> protocols.
>>
>> The server should be able to say "foo" and "bar" as well as "foo" or "bar"
>>
>> I don't see any reason that protocols have to be exclusive.  For
>> example I might want to use an oauth protocol with a chat protocol.
>
> I don't think that makes any sense -- a subprotocol defines the
> interpretation of the data that rides in the WS messages.  How can it be
> both oauth data and chat data?  It might be an additional layer that can
> carry both, but that is the subprotocol that should be negotiated.
> --
> John A. Tamplin
> Software Engineer (GWT), Google
>