Re: pg_stat_advisor extension

On Tue, Feb 6, 2024 at 12:06 AM Ilia Evdokimov
<ilya(dot)evdokimov(at)tantorlabs(dot)com> wrote:
>
> Hi hackers,
>
> I'm reaching out again regarding the patch with new extension 'pg_stat_advisor' aimed at enhancing query plan efficiency through the suggestion of creating statistics.
>
> I understand the community is busy, but I would greatly value any feedback or thoughts on this extension.
>

+ /* Define custom GUC variables. */
+ DefineCustomRealVariable("pg_stat_advisor.suggest_statistics_threshold",
+ "Set the threshold for actual/estimated rows",
+ "Zero disables suggestion of creating statistics",
+ &pg_stat_advisor_suggest_statistics_threshold,
+ 0.0,
+ 0.0,
+ INT_MAX,
+ PGC_SUSET,
+ 0,
+ NULL,
+ NULL,
+ NULL);

INT_MAX
should be 1.0?

+ if (!FindExtendedStatisticsOnVars(&rte->relid, colmap))
+ {
+ ereport(NOTICE, (errmsg("pg_stat_advisor suggestion: CREATE
STATISTICS %s %s FROM %s",
+ stat_name, create_stat_stmt, rel_name),
+ errhidestmt(true)));
+ }
now CREATE STATISTICS, the statistics name is optional.
but here you explicitly mention the statistics kind would be great.

+ elog(DEBUG1, "Estimated=%f, actual=%f, error=%f: plan=%s",
+ plan->plan_rows,
+ planstate->instrument->ntuples,
+ planstate->instrument->ntuples / plan->plan_rows,
+ nodeToString(plan));

` error=%f` seems not that right.
Also since the documentation is limited, more comments explaining
SuggestMultiColumnStatisticsForNode would be great.
overall the comments are very little, it should be more (that's my opinion).