Re: [multipathtcp] Consensus call on potential MPTCP proxy work

[Joe Touch <[e-mail]>]

On 4/21/2017 2:00 AM, Yoshifumi Nishida wrote:
> In my view, TFO relies on the info from past connections and we
> somehow believe it's valid and not staled. But, it's still external
> info because it is from the outside of the current connection.
> This case relies on other type of external info and when we believe
> the info, we will send data in SYN. I personally don't see big
> differences.

Again, TFO semantics were vetted by TCPM over a fairly long set of
discussions.

If you want TFO behavior, use TFO.

If you want to argue that this is a different way to achieve the same
results, that's a very long road of discussions to begin.

That discussion is IMO outside the scope of MPTCP (there's no zero-delay
open in its charter).

Joe