Proposed change to RTN20: restart in-progress connection attempts if an online event occurs in the connecting state

[paddybyers]

See https://sdk.ably.com/builds/ably/specification/main/features/#RTN20

The aim of RTN20 is to ensure that a client reconnects as soon as possible after the host comes online.

However, it has been reported that reconnections can take a long time if the client is connecting at the time of that online event - a connection attempt was initiated whilst offline, and is blocked in some state, but doesn't succeed as a result of the host coming online - it can stay blocked until a timeout occurs, and then the client will wait the usual retry interval (30s if it was suspended) before retrying.

A way around this would be for RTN20 to require that an in-progress connection attempt is cancelled and restarted if the online event happens whilst in the connecting state.

Internal discussion: https://ably-real-time.slack.com/archives/C01EHJ8J5MG/p1676973802248769?thread_ts=1676913499.160499&cid=C01EHJ8J5MG

[sync-by-unito]

➤ Automation for Jira commented:

The link to the corresponding Jira issue is https://ably.atlassian.net/browse/SDK-3434

[SimonWoolf]

sgtm

[JCB-K]

Hi! What's the priority on this proposal? It would make a big difference in performance for an upcoming key feature of ours.