Re: [PATCH] REPLICA IDENTITY USING INDEX accepts column with invalid NOT NULL

On 2026-Jun-12, Álvaro Herrera wrote:

> On 2026-Jun-11, Masahiko Sawada wrote:
>
> > It would be better to factor the into a common function that checks
> > whether the column has a valid NOT NULL constraint, to remove the code
> > duplication.
>
> I agree, we should have a function like
> bool
> notNullConstraintValidated(Oid relid, AttrNumber attnum)
> in pg_constraint.c, and use it in the various places where we only care
> about the constraint validation status.

I take that back. I think this change would warrant merging this new
code with the existing verifyNotNullPKCompatible(), which wants to do
pretty much the same; but that's probably a larger refactoring that
should be done in pg20 only. So maybe we can just do the simple thing
(Ante's patch, mostly, I think) for this backpatchable bug fix, then
change things more fully later on.

--
Álvaro Herrera Breisgau, Deutschland — https://www.EnterpriseDB.com/