Re: POC, WIP: OR-clause support for indexes

On Wed, Aug 21, 2024 at 4:08 PM Andrei Lepikhov
<a(dot)lepikhov(at)postgrespro(dot)ru> wrote:
> On 21/8/2024 02:17, Alexander Korotkov wrote:
> > Also, I convert the check you've introduced in the previous message to
> > op_in_opfamily(), and introduced collation check similar to
> > match_opclause_to_indexcol().
> Hi,
> I passed through the patches with fresh sight. Conceptually, this
> approach looks much better than the previous series.
> Just for the record: previously, we attempted to resolve two issues in
> one - improve the execution plan and save cycles during the
> optimisation. As I see it, it is almost impossible in this feature. So,
> I should come to terms with carrying long OR lists through the planning
> and the additional overhead this feature generates.
> I also see that the optimiser has obtained additional planning
> strategies with these patches and hasn't lost any.

Thank you for your feedback.

> Couple of findings:
>
> First:
> /* Only operator clauses scan match */
> Should it be:
> /* Only operator clauses can match */
> ?

Corrected, thanks.

> The second one:
> When creating IndexClause, we assign the original and derived clauses to
> the new, containing transformed array. But logically, we should set the
> clause with a list of ORs as the original. Why did you do so?

I actually didn't notice that. Corrected to set the OR clause as the
original. That change turned recheck to use original OR clauses,
probably better this way. Also, that change spotted misuse of
RestrictInfo.clause and RestrictInfo.orclause in the second patch.
Corrected this too.

------
Regards,
Alexander Korotkov
Supabase