Re: Feature improvement: can we add queryId for pg_catalog.pg_stat_activity view?

On Wed, Apr 7, 2021 at 11:27:04PM -0400, Álvaro Herrera wrote:
> On 2021-Apr-07, Bruce Momjian wrote:
>
> > On Thu, Apr 8, 2021 at 10:38:08AM +0800, Julien Rouhaud wrote:
>
> > > Thanks! And I agree with using query_id in the new field names while keeping
> > > queryid for pg_stat_statements to avoid unnecessary query breakage.
> >
> > I think we need more feedback from the group. Do people agree with the
> > idea above? The question is what to call:
> >
> > GUC compute_queryid
> > pg_stat_activity.queryid
> > pg_stat_statements.queryid
> >
> > using "queryid" or "query_id", and do they have to match?
>
> Seems a matter of personal preference. Mine is to have the underscore
> everywhere in backend code (where this is new), and let it without the
> underscore in pg_stat_statements to avoid breaking existing code. Seems
> to match what Julien is saying.

OK, let's get some details. First, pg_stat_statements.queryid already
exists (no underscore), and I don't think anyone wants to change that.

pg_stat_activity.queryid is new, but I can imagine cases where you would
join pg_stat_activity to pg_stat_statements to get an estimate of how
long the query will take --- having one using an underscore and another
one not seems odd. Also, looking at the existing pg_stat_activity
columns, those don't use underscores before the "id" unless there is a
modifier before the "id", e.g. "pid", "xid":

SELECT attname
FROM pg_namespace JOIN pg_class ON (pg_namespace.oid = relnamespace)
JOIN pg_attribute ON (pg_class.oid = pg_attribute.attrelid)
WHERE nspname = 'pg_catalog' AND
relname = 'pg_stat_activity' AND
attname ~ 'id$';
attname
-------------
backend_xid
datid
leader_pid
pid
queryid
usesysid

We don't have a modifier before queryid.

If people like query_id, and I do too, I am thinking we just keep
query_id as the GUC (compute_query_id), and just accept that the GUC and
SQL levels will not match. This is exactly what we have now. I brought
it up to be sure this is what we want,

--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com

If only the physical world exists, free will is an illusion.