On Wed, May 12, 2021 at 08:58:45AM +0200, Pavel Stehule wrote:
>
> I don't like the idea of implicit force enabling any feature flag, but it
> is better than current design. But it doesn't look like a robust solution.
>
> Does it mean that if somebody disables computed_query_id, then
> pg_stat_statements will not work?
It depends, but if you mean "setting up pg_stat_statements, intentionally
disabling in-core queryid calculation and not configuring an alternative
source" then yes pg_stat_statements will not work. But I don't see any
difference from "someone reduce wal_level and complain that replication does
not work" or "someone disable fsync and complain that data got corrupted". We
provide a sensible default configuration, you can mess it up if you don't know
what you're doing.
> Why is there the strong dependency between computed_query_id and
> pg_stat_statements? Can this dependency be just optional?
Once again no, as it otherwise would mean that postgres unilaterally decides
that pg_stat_statements' approach to compute a query identifier is the one and
only ultimate truth and nothing else could be useful for anyone.
ok. Understand.
If I understand well, then computed_query_id does not make sense for pg_stat_statemenst, because this extension always requires it.
Cannot be better to use queryid inside pg_stat_statements every time without dependency on computed_query_id? And computed_query_id can be used only for EXPLAIN and for pg_stat_activity.
pg_stat_statements cannot work without a queryid, so is useless to speak about configuration. If you use pg_stat_statements, then the queryid will be computed every time, but the visibility will be only for pg_stat_statements.
Or a different strategy. I understand so computed_query_id should be active. But I dislike the empty result of pg_stat_statements when computed_query_id is off. Is it possible to raise an exception instead of showing an empty result?
The most correct fix from my perspective is just check in function pg_stat_statements if query id is computed or not. If not, and there is no data, then raise an exception with the hint "enable compute_query_id". When there is data, then show a warning with the mentioned hint and show data.
What do you think about it?
Pavel