Re: [quicwg/base-drafts] Stream closed state and acknowledgments (#743)

[Ryan Hamilton <[e-mail]>]

If the question is specifically about when the stream is closed, doesn't
the state diagram make it clear that it's when the FIN is sent, not when
the FIN is ACK'd?

                            +--------+
                            |        |
                            |  idle  |
                            |        |
                            +--------+
                                 |
                        send/recv STREAM/RST
                             recv MSD/SB
                                 |
                                 v
                 recv FIN/  +--------+    send FIN/
                 recv RST   |        |    send RST
                  ,---------|  open  |-----------.
                 /          |        |            \
                v           +--------+             v
         +----------+                          +----------+
         |   half   |                          |   half   |
         |  closed  |                          |  closed  |
         | (remote) |                          |  (local) |
         +----------+                          +----------+
             |                                        |
             |   send FIN/  +--------+    recv FIN/   |
              \  send RST   |        |    recv RST   /
               `----------->| closed |<-------------'
                            |        |
                            +--------+


On Tue, Aug 22, 2017 at 5:29 PM, Martin Thomson <[e-mail]>
wrote:

> It depends on the model you assume. That assumes a handoff between the
> stream and some transmit/retransmit machinery.
>
> The other model, one that I prefer, is one where the transmission
> machinery pulls from available streams. It uses priority to determine what
> to send, preferring data that has already been sent. In that case, data on
> streams could remain attributed to the stream until it is acknowledged. The
> advantage here is potentially better prioritization and the cost is a
> potential increase in the number of bytes that are outstanding without
> acknowledgment.
>
> Neither model is more modular than the other, the line just appears in a
> different place.
>
> —
> You are receiving this because you are subscribed to this thread.
> Reply to this email directly, view it on GitHub
> <https://github.com/quicwg/base-drafts/issues/743#issuecomment-324187648>,
> or mute the thread
> <https://github.com/notifications/unsubscribe-auth/ASp6ysIBUxSjFgyl_B1XpJvt7HABvFQlks5sa3JygaJpZM4O-AeX>
> .
>


-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/743#issuecomment-326443532