> That's intrinsic to tcp_write. tcp_write only enqueues. It needs
> tcp_output
> to actually send anything (possibly called via tcp_output_nagle as you
> notice below...)

Yeah, so I broke it adding the tcp_output_nagle.  I am emulating netconns
but with NO_SYS=1 and used that code as a reference.  I'll take out the
call.

> You can call tcp_output directly yourself, or you can call
> tcp_output_nagle
> to use the Nagle algorithm, which is usually recommended for a variety of
> reasons. http://en.wikipedia.org/wiki/Nagle's_algorithm

Thank you.  That was a good read.  As is Nagle's follow-up about delayed
ACKs and Nagle's algorithm not playing well together and his solution: 

http://developers.slashdot.org/comments.pl?sid=174457&threshold=1&commentsor
t=0&mode=thread&cid=14515105

Does lwIP use or support this delayed ACK "improvement"?

Thanks for clearing this up for me.

Bill



_______________________________________________
lwip-users mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/lwip-users

Reply via email to