Re: BUG #19435: Error: "No relation entry for relid 2" Triggered by Complex Join with Self-Referencing Tables

On Thu, Mar 19, 2026 at 3:03 PM Alexander Korotkov <aekorotkov@gmail.com> wrote:
> On Thu, Mar 19, 2026 at 9:18 AM Andrei Lepikhov <lepihov@gmail.com> wrote:
> > On 18/3/26 18:55, Kirill Reshke wrote:
> > > On Wed, 18 Mar 2026 at 18:31, Andrei Lepikhov <lepihov@gmail.com> wrote:
> > >>
> > >> On 18/3/26 13:21, Alexander Korotkov wrote:
> > >>> On Wed, Mar 18, 2026 at 2:18 PM Andrei Lepikhov <lepihov@gmail.com> wrote:
> > >>>> What is the reason for the second change? Tender, can you show us how to
> > >>>> reproduce the issue so we can support your update to
> > >>>> restrict_infos_logically_equal? If we include it, we should add a test.
> > >>>
> > >>> I think Tender already shown this in [1].  The same qual is present
> > >>> twice in the plan.
> > >> Got it. I suggest making this a separate commit to keep the history
> > >> clear. Let me share a draft with a test case for the bug fix first.
> > > Hi!
> > > Is `cool_bool` a typo of `col_bool` in regression test ?
> >
> > This is actually a happy coincidence, not a typo. We made this mistake
> > during initial development because I didn't realise that a boolean
> > operation on a bool variable is never wrapped in an expression
> > structure. It's a 'cool' example that shows a rare edge case.
> >
> >  > And the patch seems to have forgotten to add "Reported by".
> >
> > Yeah, let the committer manage the award part.
>
> I've revised the patch.  Renamed cool_bool to cool_col, added
> "Reported by", and revised authors list according to my opinion.

OK. I've pushed this.  Let's go back to
restrict_infos_logically_equal().  I'm still not convinced that we
need to check if required_relids is singleton.  Why we can ignore
outer_relids for singleton, but can't do if, for instance, two
relations involved?

------
Regards,
Alexander Korotkov
Supabase