jian he <jian.universal...@gmail.com> 于2024年3月29日周五 14:56写道：
> hi.
> about v4, i think, i understand the changes you made.
> RemoveConstraintById(Oid conId)
> will drop a single constraint record.
> if the constraint is primary key, then primary key associated
> attnotnull should set to false.
> but sometimes it shouldn't.
>
Yeah, indeed.
>
>
> for example:
> drop table if exists t2;
> CREATE TABLE t2(c0 int, c1 int);
> ALTER TABLE t2 ADD CONSTRAINT t2_pk PRIMARY KEY(c0, c1);
> ALTER TABLE t2 ALTER COLUMN c0 ADD GENERATED ALWAYS AS IDENTITY;
> ALTER TABLE t2 DROP c1;
>
> + * If this was a NOT NULL or the primary key, the constrained columns must
> + * have had pg_attribute.attnotnull set. See if we need to reset it, and
> + * do so.
> + */
> + if (unconstrained_cols != NIL)
>
> unconstrained_cols is not NIL, which means we have dropped a primary key.
> I am confused by "If this was a NOT NULL or the primary key".
>
NOT NULL means the definition of column having not-null constranit. For
example:
create table t1(a int not null);
the pg_attribute.attnotnull is set to true.
primary key case like below:
create table t2(a int primary key);
the pg_attribute.attnotnull is set to true.
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.
>
> +
> + /*
> + * Since the above deletion has been made visible, we can now
> + * search for any remaining constraints on this column (or these
> + * columns, in the case we're dropping a multicol primary key.)
> + * Then, verify whether any further NOT NULL exists, and reset
> + * attnotnull if none.
> + */
> + contup = findNotNullConstraintAttnum(RelationGetRelid(rel), attnum);
> + if (HeapTupleIsValid(contup))
> + {
> + heap_freetuple(contup);
> + heap_freetuple(atttup);
> + continue;
> + }
>
> I am a little bit confused by the above comment.
> I think the comments should say,
> if contup is valid, that means, we already have one "not null"
> constraint associate with the attnum
> in that condition, we must not set attnotnull, otherwise the
> consistency between attnotnull and "not null"
> table constraint will be broken.
>
> other than that, the change in RemoveConstraintById looks sane.
>
Above comments want to say that after pk constranit dropped, if there are
tuples in
pg_constraint, that means the definition of column has not-null constraint.
So we can't
set pg_attribute.attnotnull to false.
For example:
create table t1(a int not null);
alter table t1 add constraint t1_pk primary key(a);
alter table t1 drop constraint t1_pk;
--
Tender Wang
OpenPie: https://en.openpie.com/
