[nwcrg] What I SHOULD have said about https://datatracker.ietf.org/doc/html/draft-kuhn-coding-congestion-transport

[Spencer Dawkins at IETF <[e-mail]>]

During Nicholas's presentation this afternoon, I pointed out that it might
be good to consider whether "repair equals loss" was the right model for
sender congestion control responses because we we had considered ECN as
equal to loss for many years, but (FINALLY) redefined the ECT(1) bit to
allow other responses. It might be useful to point the discussion of this
draft in the same direction - assume the use of scalable congestion control
and ECT(1) signaling.

I was talking to Gorry Fairhurst afterwards, and Gorry told me what I
SHOULD have said, was that the sender should perform whatever its response
to congestion would be - if it's an old-style sender, treat recovery as
loss, but if it supports scalable congestion control (L4S is one example),
it doesn't have to treat recovery as loss - it just needs to do something
appropriate.

Again, I hope that's useful, and I appreciate this work a lot, because
people do need it.

Thanks,

Spencer

p.s. cc: Gorry, in case I got that wrong.