On 5/24/13 2:07 PM, Erwin Brandstetter wrote:
> My tests on Postgres 9.1 and 9.2 seem to confirm that the check for
> non-deferrable constraints happens *after every row*, not after every
> command.
That only applies to unique constraints, not to other kinds of
constraints. That's why the unique constraint defer behavior is
described separately, but it's different from the normal constraint
defer behavior.