Re: compute_query_id and pg_stat_statements

On Wed, May 12, 2021 at 05:30:26PM +0800, Julien Rouhaud wrote:
> On Wed, May 12, 2021 at 10:57:25AM +0200, Pavel Stehule wrote:
> >
> > My second proposal can work for your example too. pg_stat_statements have
> > to require any active queryid computing. And when it is not available, then
> > the exception should be raised.
> >
> > The custom queryid can return null, and still the queryid will be computed.
> > Maybe the warning can be enough. Just, if somebody use pg_stat_statements
> > function, then enforce the check if queryid is computed (compute_query_id
> > is true || some hook is not null), and if not then raise a warning.
>
> Ah I'm sorry I misunderstood your proposal. Yes, definitely adding a warning
> or an error when executing pg_stat_statements() SRF would help, that's a great
> idea!
>
> I'll wait a bit in case someone has any objection, and if not send an updated
> patch!

Hearing no complaint, PFA a v2 implementing such a warning. Here's an
extract from the updated regression tests:

-- Check that pg_stat_statements() will complain if the configuration appears
-- to be broken.
SET compute_query_id = off;
SELECT pg_stat_statements_reset();
pg_stat_statements_reset
--------------------------

(1 row)

SELECT count(*) FROM pg_stat_statements;
WARNING: Query identifier calculation seems to be disabled
HINT: If you don't want to use a third-party module to compute query identifiers, you may want to enable compute_query_id
count
-------
0
(1 row)

I'm of course open to suggestions for some better wording.