Re: [quicwg/base-drafts] Required state for retaining unacked RETIRE_CONNECTION_ID frames is unbound (#3509)

[Kazuho Oku <[e-mail]>]

@MikeBishop I think your summarization of the problem is accurate.

Regarding an attacker selectively acking packets carrying STREAM frames, yes, the attack exists, but it the sender can mitigate that problem in certain ways. We already state that in [section 21.5](https://quicwg.org/base-drafts/draft-ietf-quic-transport.html#name-stream-fragmentation-and-re).

This issue is the only case that a protocol cannot be implemented properly, with a bounded amount of state.

Regarding how we should fix the problem, _if_ we are going make a protocol change, my preference goes to applying the same design principle that we have in other places (i.e. introduce MAX_CONNECTIONS_IDS frame). I do not like designing a mechanism specific for this purpose, as there is a chance that we might make mistakes.

I can also live with what @martinthomson proposed in https://github.com/quicwg/base-drafts/issues/3509#issuecomment-597425775 (i.e. encourage leaky behavior).

-- 
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/3509#issuecomment-602974153