Re: [hybi] Multiplexing - closing a logical channel

[Zhong Yu <[e-mail]>]

Got it, thanks.

Another question,
http://tools.ietf.org/html/draft-ietf-hybi-thewebsocketprotocol-17#section-5.5.1

5.5.1. Close
   The application MUST NOT send any more data frames after sending a
   close frame.

doe it imply that more control frames may be sent after a close frame?

Zhong Yu

On Fri, Mar 15, 2013 at 3:50 AM, Takeshi Yoshino <[e-mail]> wrote:
> Hi
>
> On Fri, Mar 15, 2013 at 10:34 AM, Zhong Yu <[e-mail]> wrote:
>>
>> A vanilla WebSocket connection can be half-closed, i.e.
>>
>>     X sends a close frame to Y,
>>     then continues to read frames from Y.
>>
>>     Y receives a close frame from X,
>>     but continues to send data frames to X.
>>
>> However the multiplexing draft seems to require that a logical channel
>> cannot do that - a logical channel is either active or inactive, as
>> soon as it becomes inactive, by sending or receiving DropChannel, the
>> channel cannot be used further, on either direction.
>
>
> Sending a DropChannel corresponds to TCP closure ("Close the WebSocket
> Connection" operation in RFC 6455).
>
> WebSocket endpoints can make their connections half-closed even on mux.
>
> See
> http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-09#section-19.
>
>>
>> If my understanding is correct, then I wonder what's the reason for
>> this behavior? Why don't we allow close frames exchanged on a logical
>> channel so that the closing process of a logical channel is consistent
>> with that of a vanilla WebSocket connection?
>
>
> Thanks
> Takeshi