Re: [quicwg/base-drafts] Prioritize Handshake probe packet over Short packet (#3583)

[Martin Thomson <[e-mail]>]

Leaving aside that recommendation for the moment (which I agree is close to mandatory), the experience is pretty poor if there is only one key.  You just moved to a 4x exponent on the PTO timer rather than a 2x timer.  I guess that's why we have the recommendation.

(For reference, neqo tells the frame sending apparatus that it lost some data when a PTO timer fires.  That causes those older frames to be sent again rather than trying to find fresh data to send.  That's on the principle that older data is more important.  But that isn't relevant to this discussion.  The main comment to look at is [this one](https://github.com/mozilla/neqo/pull/448#issuecomment-590001888), which shows how the interleaving can be a little janky.)

I realize that we haven't reconciled this with the requirement not to arm the application data PTO when there are handshake/initial packets outstanding.  That might limit this to the one case of interleaving Initial and Handshake.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/quicwg/base-drafts/issues/3583#issuecomment-618799323