Andres Freund <andres@xxxxxxxxxxxxxxx> writes: > On 2014-07-01 15:20:37 -0400, Tom Lane wrote: >> It seems like there are three, not mutually exclusive, ways we might >> address this: > But I think more importantly it's probably quite possible to hit a > similar problem without ON COMMIT DROP relations. Say DISCARD TEMP > inside a transaction (with several subxacts) or so? So we probaly really > should fix the bad scaling. Well, my thought was that these approaches would address somewhat different sets of use-cases, and we might well want to do more than one. Even if StandbyReleaseLocks were zero-cost, not emitting the WAL in the first place is surely considerably cheaper yet. regards, tom lane
