Arnt Gulbrandsen wrote:
IMO, a TCP keepalive API needs contain only three functions. Two to answer the questions "are any acks overdue, and if so, by how long?" and "what are the current RTT and bandwidth estimates?" and one to provoke the peer into sending an ack.
I would say that RFC1122 already defines keepalives (unwillingly, but it does). Thus what's needed is an RFC saying "if you want to implement keepalives in a sane manner, do it like this".
Martin _______________________________________________ Ietf mailing list Ietf@xxxxxxxx https://www.ietf.org/mailman/listinfo/ietf