Re: [lwip-devel] [bug #50837] TCP: zero window probe doesn't timeout

[address@hidden]

Joel Cunningham wrote:
> Follow-up Comment #38, bug #50837 (project lwip):
>
> I agree that treating the send window as effectively zero when the next unsent
> segment is larger, is becoming more messy to maintain.

Slightly off-topic so continuing on the list:
In the case of "0 < wnd < pcb->unsent->len" (next segment does not fit 
into the window), until now we discussed 2 options:
- handle this case the same as a true zero window
- split segments and send what fits into the window

What about just sending the full segment in this case? The receiver 
might or might not ACK the full segment, but
shouldn't that just work?


Simon