On Sun, 6 Aug 2000 vectro@pipeline.com wrote:
> > Seriously, for foreign key constraints, you can remove them by removing
> > the rows in pg_trigger that are associated with them. If they're unnamed,
> > you'll have to use the data in tgargs to determine which is the correct
> > one.
>
> Perhaps we should come up with a better naming convention than <unnamed>
> for unnamed constraints? Like is done with implied indices and sequences.
Definately. It's on my hopefully to get done reasonably soon list since
Jan's really busy with TOAST. I'm actually trying to get it done mostly
correctly to the SQL spec which is why I haven't hit it yet. I want to
make sure that constraint names (or at least fk constraint names) never
conflict with any other constraint name in the database (schema
technically - but until we have schemas...).