"Gregory Williamson" <Gregory.Williamson@xxxxxxxxxxxxxxxx> writes: > Harald, >> (Note the different column types.) > Precisely -- there are a number of casts that have been removed from > this release. This isn't a casting problem, I don't think. 8.3 is definitely stricter about implicit casts with FKs, eg try adding this to the test case: regression=# create table t3 (f1 int references t1); ERROR: foreign key constraint "t3_f1_fkey" cannot be implemented DETAIL: Key columns "f1" and "id" are of incompatible types: integer and character. But it's allowing the constraint and then failing at runtime, which seems like an outright bug, since the values *should* compare equal AFAICS. I wonder if this is related to bug #3938. regards, tom lane ---------------------------(end of broadcast)--------------------------- TIP 6: explain analyze is your friend
