[quicwg/base-drafts] Coordinating spin (#2763)

[Martin Thomson <[e-mail]>]

When multiple connections share the same network path or multiple connection IDs are used for the same path, using the spin bit is complicated by the difficulty in correlating forward and reverse flows.  There is currently text that uses RFC 6919 language in the draft recommending coordination of the spin to avoid this problem.

However, as [some have observed](https://github.com/quicwg/base-drafts/pull/2751#pullrequestreview-244603953), this coordination seems difficult and therefore unlikely.  It also creates a strong signal that indicates that all the connections on the path are terminated at the same endpoint on both ends.  Maybe we don't want to be generating that signal.

Also, because this coordination can't be guaranteed, the on-path observer is forced to build code for correlating/segregating measurements by connection ID in the case where there is no coordination.

Maybe it's best just to remove this recommendation entirely.

-- 
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/2763