> On Friday 19 September 2003 16:04, Tom Lane wrote:
>> Richard Huxton <dev@archonet.com> writes:
>> > On Friday 19 September 2003 09:00, btober@seaworthysys.com wrote:
>> >> Is this supposed to work that way? I would expect field3 and
>> field4 to have their respective column defaults assigned on the
>> second INSERT (row B), just like on the first INSERT (row A).
>> >
>> > Hmm - well, you're explicitly telling it to insert VALUES (...,
>> new.field3, ...) so if new.field3 is null then it *should* do that.
>>
>> Exactly. The defaults attached to the underlying table determine
>> what gets added to an INSERT into the underlying table. In this
>> case, since the rule's INSERT specifies all the fields, there is no
>> scope for those defaults to apply.
>>
>> What Berend actually wants is to attach column defaults to the
>> *view*, so that they apply to an INSERT mentioning the view. You can
>> do this in recent PG releases (7.3 for sure, not sure about 7.2)
>> using ALTER TABLE ... ADD DEFAULT.
>
> Hmm - didn't know that. Useful.
>
> Am I right in thinking that *in theory* we should be able to "inherit"
> the defaults to the view automatically?
I think that what you describe here as inheriting the defaults is the
behavior I mistakenly expected as normal. But being able to assigne
defaults to view columns is pretty darn cool.
~Berend Tober