On Wed, Apr 10, 2024 at 1:29 AM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote: > > On 2024-Mar-29, Tender Wang wrote: > > > I think aboved case can explain what's meaning about comments in > > dropconstraint_internal. > > But here, in RemoveConstraintById() , we only care about primary key case, > > so NOT NULL is better to removed from comments. > > Actually, I think it's better if all the resets of attnotnull occur in > RemoveConstraintById, for both types of constraints; we would keep that > block in dropconstraint_internal only to raise errors in the cases where > the constraint is protecting a replica identity or a generated column. > Something like the attached, perhaps, may need more polish. >
DROP TABLE if exists notnull_tbl2; CREATE TABLE notnull_tbl2 (c0 int generated by default as IDENTITY, c1 int); ALTER TABLE notnull_tbl2 ADD CONSTRAINT Q PRIMARY KEY(c0, c1); ALTER TABLE notnull_tbl2 DROP CONSTRAINT notnull_tbl2_c0_not_null; ALTER TABLE notnull_tbl2 DROP c1; \d notnull_tbl2
last "\d notnull_tbl2" command, master output is: Table "public.notnull_tbl2" Column | Type | Collation | Nullable | Default --------+---------+-----------+----------+---------------------------------- c0 | integer | | not null | generated by default as identity
last "\d notnull_tbl2" command, applying 0001-Correctly-reset-attnotnull-when-constraints-dropped-.patch output: Table "public.notnull_tbl2" Column | Type | Collation | Nullable | Default --------+---------+-----------+----------+---------------------------------- c0 | integer | | | generated by default as identity
Hmm,
ALTER TABLE notnull_tbl2 DROP c1; will not call dropconstraint_internal().
When dropping PK constraint indirectly, c0's attnotnull was set to false in RemoveConstraintById().