> So TIMESTAMP 'now' does get coerced to a timestamp constant on sight,
> which is what I would expect. I find it rather surprising that the
> unknown-type literal isn't getting coerced during CREATE TABLE too.
> After looking at the code, I see that this is a deliberate hack to
> make the world safe for DEFAULT 'now' --- see catalog/heap.c around line
> 1630 in current sources. However, I think this is an ugly backwards-
> compatibility hack, rather than something the manual should recommend
> as preferred practice. So I think the docs are okay as is.
Fair enough. It just had me quite confused. Maybe it would be worth
putting in a qualifier that says that DEFAULT 'now' is ok?
Another thing, why is there a funtional form of 'now' called now(), but not
a functional form of 'today' called today()? (7.1.3)
Chris