Re: not null constraints, again

On Wed, Sep 11, 2024 at 9:11 AM jian he <jian.universality@gmail.com> wrote:
>
> On Wed, Sep 11, 2024 at 2:18 AM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
> >
> > Hello, here's a v2 of this patch.  I have fixed --I think-- all the
> > issues you and Tender Wang reported (unless I declined a fix in some
> > previous email).
> >

after applying your changes.

in ATExecAddConstraint, ATAddCheckNNConstraint.
ATAddCheckNNConstraint(wqueue, tab, rel,
            newConstraint, recurse, false, is_readd,
            lockmode);
if passed to ATAddCheckNNConstraint rel is a partitioned table.
ATAddCheckNNConstraint itself can recurse to create not-null pg_constraint
for itself and it's partitions (children table).
This is fine as long as we only call ATExecAddConstraint once.

but ATExecAddConstraint itself will recurse, it will call
the partitioned table and each of the partitions.

The first time ATExecAddConstraint with a partitioned table with the
following calling chain
ATAddCheckNNConstraint-> AddRelationNewConstraints -> AdjustNotNullInheritance1
works fine.

the second time ATExecAddConstraint with the partitions
ATAddCheckNNConstraint-> AddRelationNewConstraints -> AdjustNotNullInheritance1
AdjustNotNullInheritance1 will make the partitions
pg_constraint->coninhcount bigger than 1.


for example:
drop table if exists idxpart, idxpart0, idxpart1 cascade;
create table idxpart (a int) partition by range (a);
create table idxpart0 (a int primary key);
alter table idxpart attach partition idxpart0 for values from (0) to (1000);
alter table idxpart add primary key (a);

After the above query
pg_constraint->coninhcount of idxpart0_a_not_null  becomes 2.
but it should be 1
?