improve(relayer): Catch identifiable re-org events by blockHash #2447
Conversation
In order to make a subsequent refactoring diff smaller.
There's no need to filter based on the complete set of unique attributes - in practice it's more desirable to cast the net wide and discard anything matching the re-orged blockHash.
pxrl
requested review from
bmzig,
dohaki,
james-a-morris,
mrice32 and
nicholaspai
as code owners
| } | ||
| let removedEventIdx: number; | ||
| do { | ||
| removedEventIdx = pendingEvents.findIndex(({ blockHash }) => blockHash === event.blockHash); |
There was a problem hiding this comment.
Does this require the event which was reorged to be reincluded in this block? What if the event was removed by a reorg and not mined again yet?
There was a problem hiding this comment.
Yeah; these dropped events are currently reported per-event over the event subscription. Per current implementation we require one notification per event, per blockHash. The change implemented here is to permit a single removed event notification to drop all stored events for a given blockHash, so we basically cast the net a little wider, which is beneficial in case one of the removed event notifications doesn't materialise (for whatever reason).
There's a follow-on change required which will actually track block hashes and reconcile when a re-org happens by finding a common ancestor and invalidating everything inbetween. That's still pending.
There's no need to filter based on the complete set of unique attributes -
in practice it's more desirable to cast the net wide and discard anything
matching the re-orged blockHash.
nb. This change only discards events by blockHash before they've been
read into SpokePoolClient state by
SpokePoolClient.update(). At thatpoint the blockHash is discarded so it's no longer possible to identify
the event by its blockHash; this needs a separate solution.