On Thu, 2003-06-19 at 09:40, Tom Lane wrote:
> Rod Taylor <rbt@rbt.ca> writes:
> > Anyway, I suppose you have indirectly confirmed that user triggers, etc.
> > should NOT fire on for the data update. I didn't see anything in the
> > spec that said one way or the other.
>
> Actually, I didn't mean to take a position one way or the other. You
> could certainly argue that they should fire ...
Do we want them to? If we don't mind them being executed, it is far
easier to:
- alter table structure
- Add all new constraints (without confirming their correctness at that
time)
- update table contents via an SPI call to UPDATE WHERE <column> IS NULL
The where clause would avoid issues with inherited data being
overwritten when the child tables are updated.
--
Rod Taylor <rbt@rbt.ca>
PGP Key: http://www.rbt.ca/rbtpub.asc
