Re: [quicwg/base-drafts] are flow control frames really idempotent? (#1612)

[Subodh Iyengar <[e-mail]>]

@kazuho I don't think we should make it a MUST on the receiver, just a MAY and maybe a MUST on the sender. It is already a MUST, i.e. you MUST not reduce the flow control window.

@dtikhonov of course other bugs are possible, however this is a pretty major one if the sender actually tried to reduce the flow control. Without enforcement the MUST of "sender MUST not reduce the window" is almost impossible to enforce. Ignoring it will just kick the can down the road with such issues only getting discovered way after deployment vs at dev time. 

While I think it's fine to allow senders to copy over the packets that they have sent before, I don't think that should be a goal of QUIC. That is senders that are doing this should be aware of what they are signing for and take appropriate precautions. There are efficient ways to implement copying packets but updating max data. Preventing the efficient enforcement of a clear MUST would be very sad.

-- 
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/1612#issuecomment-409251712