"Tom Lane" <tgl@sss.pgh.pa.us> writes:
> The simplest fix seems to be to invent an additional flag variable
> "signalAwaited" which is set/cleared by ProcWaitForSignal and checked by
> LockWaitCancel. This would make cancelling out of a ProcWaitForSignal call
> exactly analogous to cancelling out of a heavyweight-lock acquisition.
Is that the flag that is an assertion that no cleanup is needed? Or is that
something else?
-- Gregory Stark EnterpriseDB http://www.enterprisedb.com Ask me about EnterpriseDB's 24x7 Postgres support!