Philip Warner <pjw(at)rhyme(dot)com(dot)au> writes:
> At 20:02 31/03/01 -0500, Tom Lane wrote:
>> Perhaps we should just hack that code
>> to not suppress inherited attrs when they have default values and/or
>> NOT NULL that's not in the parent.
> That's what I meant; can we easily do the 'not in the parent' part, since
> we may have to go up a long hierarchy to find the parent?
pg_dump must already contain code to traverse the inheritance hierarchy
(I haven't looked to see where). Couldn't we just extend it to believe
that it's found a match only if the default value and NOT NULL state
match, as well as the column name?
regards, tom lane