Gregory Stark <stark@enterprisedb.com> writes:
> Well if there's a convenient later place to add the check then sure. Will it
> mean pg_dump will have to put DEFAULT NULL everywhere though? Or can it detect
> that it's an inherited table where the default doesn't match?
The latter --- I already committed that fix.
> Perhaps it should be even later and we should store the NULL default in the
> catalog but filter it out when we build the relcache?
No, I don't think we want to be making useless pg_attrdef entries.
I do want to put the test as late as possible though, maybe even
StoreAttrDefault?
regards, tom lane