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

[Marten Seemann <[e-mail]>]

Thank you for the explanation, @kazuho. I think I now understood the problem.

> Mathematically, this might be unbounded, but it's more likely that in practice the limit is 2*active_connection_id_limit as it would be unlikely that unacknowledged RETIRE_CONNECTION_ID frames would remain so once more NEW_CONNECTION_ID frames start arriving. More so as the issuer of those NEW_CONNECTION_ID frames has to have received the RETIRE_CONNECTION_ID frames.

This would require endpoints to make assumptions about how the peer handles the `active_connection_id_limit` and how it issues new CIDs. This can get complicated really quickly and leave the connection in an unrecoverable state (or at the least a state with less CIDs than it should have). Introducing a new frame, as @kazuho suggested, seems a lot cleaner to me.

-- 
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-597441915