Re: Size of window variables

6.9.1 has it:

A sender MUST NOT allow a flow control window to exceed 2^31 - 1 bytes. 
If a sender receives a WINDOW_UPDATE that causes a flow control window 
to exceed this maximum it MUST terminate either the stream or the 
connection, as appropriate.

On 07/03/2014 11:17 AM, Poul-Henning Kamp wrote:
> In message <CABkgnnW3neoQdXbtuRgc+eb6afOvRkcDdr9yMnp_mkex+QhmRQ@mail.gmail.com>, Martin Thomson writes:
>
>> You could permit the window to become negative.  Implementations are
>> required to track a negative window after settings changes..
> One nit related to windows:
>
> Window updates are 31 bits, but we don't put a bound on the actual
> window size.
>
> At the very least we should write that there is no guarantee that
> a 32bit signed integer will be enough to hold the window value.
>
> But simpler and clearer would be to mandate that "window sizes are
> 64 signed values" and be done with it.
>
>

Received on Thursday, 3 July 2014 18:25:48 UTC