Richard Guo writes:
> On Sat, Mar 25, 2023 at 11:41 PM Tom Lane wrote:
>> Richard Guo writes:
>>> Should we instead mark the constant-TRUE clause with required_relids
>>> plus the OJ relid?
>> I do not think it matters.
> Yeah, I agree that it makes no difference currently. One day if we want
On Sat, Mar 25, 2023 at 11:41 PM Tom Lane wrote:
> Richard Guo writes:
> > Should we instead mark the constant-TRUE clause with required_relids
> > plus the OJ relid?
>
> I do not think it matters.
Yeah, I agree that it makes no difference currently. One day if we want
to replace the is_pushe
Richard Guo writes:
> Should we instead mark the constant-TRUE clause with required_relids
> plus the OJ relid?
I do not think it matters.
> Even if the join does become clauseless, it will end up being an
> unqualified nestloop. I think the join ordering algorithm will force
> this join to be
I happened to notice a constant-TRUE clause with is_pushed_down being
true while its required_relids not including the OJ being formed, which
seems abnormal to me. It turns out that this clause comes from
reconsider_outer_join_clauses(), as a dummy replacement if we've
generated a derived clause.
