PG Bug reporting form <noreply(at)postgresql(dot)org> writes:
> When we rename a constraint trigger it seems like the name is not updated in
> pg_constraint.
Hmm, yeah ... renametrig() is intent on renaming the entry in pg_trigger,
but I see no code there that pays any attention to pg_constraint.
Conversely, ALTER TABLE RENAME CONSTRAINT updates the pg_constraint
entry but not the associated trigger. It's not real clear to me
whether this is intentional. There's no a-priori reason why the
two objects have to have the same name; but I doubt we have pg_dump
support for recreating a situation where they are different.
[ experiments ... ] Interestingly, pg_dump seems to dump the name
from pg_trigger not the one from pg_constraint.
regards, tom lane