Thread: pg_dump does not dump domain not-null constraint's comments

hi.

----------------------------------------------------
begin;
create schema test;
set search_path to test;
create domain d1 as int;
alter domain d1 add constraint nn not null;
alter domain d1 add constraint cc check (value is not null);
comment on constraint nn on domain d1 is 'not null constraint on domain d1';
comment on constraint cc on domain d1 is 'check constraint on domain d1';
commit;

------the above is the test script------------------
in PG17 and master, pg_dump (--schema=test --no-owner) will only produce
COMMENT ON CONSTRAINT cc ON DOMAIN test.d1 IS 'check constraint on domain d1';
but didn't produce
COMMENT ON CONSTRAINT nn ON DOMAIN test.d1 IS 'not null constraint on
domain d1';
we should make pg_dump to produce it too?

The attached patch tries to make it produce comments on not-null
constraints on domains.
I aslo renamed struct TypeInfo fields, nDomChecks will be renamed to
nDomConstrs;
domChecks will be renamed to domConstrs.

TypeInfo->domConstrs will also include not-null constraint
information, changing from
domChecks to domConstrs makes sense, IMHO.

On 2025-May-07, jian he wrote:

> in PG17 and master, pg_dump (--schema=test --no-owner)
> [...]
> didn't produce
> COMMENT ON CONSTRAINT nn ON DOMAIN test.d1 IS 'not null constraint on
> domain d1';
> we should make pg_dump to produce it too?

Yes, this is clearly a pg17 bug whose fix should be backpatched.

> The attached patch tries to make it produce comments on not-null
> constraints on domains.

Thanks, I'll have a look.

> I aslo renamed struct TypeInfo fields, nDomChecks will be renamed to
> nDomConstrs; domChecks will be renamed to domConstrs.
> TypeInfo->domConstrs will also include not-null constraint
> information, changing from domChecks to domConstrs makes sense, IMHO.

Hmm, for a backpatch I would leave the field names alone since they are
publicly visible; we can rename separately in pg19 once it opens.  Can
you resubmit splitting the renaming out to a 0002 patch?

-- 
Álvaro Herrera               48°01'N 7°57'E  —  https://www.EnterpriseDB.com/
"In fact, the basic problem with Perl 5's subroutines is that they're not
crufty enough, so the cruft leaks out into user-defined code instead, by
the Conservation of Cruft Principle."  (Larry Wall, Apocalypse 6)

Ooh, you know what?  I ran your the CREATE DOMAIN statement in your new
test and pg_dump'ed that, and as far as I can tell the _name_ of the
not-null constraint is not dumped either.  So it's not just the comment
that we're losing.  That's a separate bug, and probably needs to be
fixed first, although I'm not sure if that one is going to be
back-patchable.  Which means, I withdraw my earlier assessment that the
comment fix is back-patchable as a whole.

-- 
Álvaro Herrera         PostgreSQL Developer  —  https://www.EnterpriseDB.com/