Re: Questions about stream state transitions

Thanks Martin,

You're right. HEADER+END_STREAM goes like two step transitions and
PUSH_PRMISE can be sent in hald closed(remote).

The table was fixed and updated.

The transition flow of PUSH_PROMISE confused me much. I hope no more fix is needed.

(2013/07/19 1:37), Martin Thomson wrote:
> On 18 July 2013 01:32, Shigeki Ohtsu <ohtsu@iij.ad.jp> wrote:
>>> Also, I think that anytime DATA can be successfully received,
>>> WINDOW_UPDATE can be sent (half closed (local) seems to violate
>>> that).
>>
>>
>> PRIORITY can also be received in reserved(local/remote) accrding to the
>> above model.
>
> No this is also incorrect in the table.  If you can receive DATA, then
> the protocol requires that you be able to send WINDOW_UPDATE.
>
> Another bug I noticed is that it should be possible to send
> PUSH_PROMISE in half closed (remote) and to receive it in half closed
> (local).
>

Received on Friday, 19 July 2013 10:50:56 UTC