Greg Stark <gsstark@mit.edu> writes:
> Tom Lane <tgl@sss.pgh.pa.us> writes:
>> The implementation I had in mind was to add columns similar to attinhcount
>> and attislocal to pg_constraint.
> Hm that would be simpler. That still leaves NOT NULL as a bit of a headache.
Yeah, I think we would want to start storing NOT NULL constraints
explicitly in pg_constraint so that we could track them. This would
allow fixing some other things too, like the fact that we fail to
remember names for NOT NULL constraints. attnotnull might still be
useful as an optimization, or maybe it should just go away.
regards, tom lane