Re: [hybi] I-D Action:draft-ietf-hybi-thewebsocketprotocol-05.txt

[Gabriel Montenegro <[e-mail]>]

Just responding to one point:



“No, see 4.4.  Only control frames can be sent in the middle of a fragmented message (I believe the "open question" line is out of date -- I think the consensus was that control frames can be injected in the middle of a fragmented message, otherwise the latency of responding to ping, for example, would be very long if behind a large message), and control frames cannot be fragmented. “


Agree, I think this can be deleted from 4.4:

   o  _Note: There is an open question as to whether control frames be
      interjected in the middle of a fragmented message.  If so, it must
      be decided whether they be fragmented (which would require keeping
      a stack of "in-progress" messages)._

I also believe we agreed that the answer to the first questions is “yes”. Regardless, it is logical. For example, you might want to send a Ping (the delay between fragments may be large as the data is produced) or more obviously, a Close.

And the answer to the second has already been settled in the draft for at least a couple of releases as “control messages cannot be fragmented”. Deleting the above means adding a sentence in 4.5 (control frames) the the effect that “Control frames can be interjected in the middle of a fragmented message.”