Tom,
> Either one. They both have the potential to reference more than one
> column, so if the committee had meant errors to try to identify the
> referenced columns, they'd have put something other than COLUMN_NAME
> into the standard. They didn't.
I'm less concerned about the standard here and more concerned about what
helps our users. Having column names for an FK error is *extremely*
useful for troubleshooting, particularly if the database has been
upgraded from the 7.4 days and has non-useful FK names like "$3".
I agree that column names for CHECK constraints is a bit of a tar baby,
since check constraints can be on complex permutations of columns.
--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com