On 11/15/06, Oystein Grovlen - Sun Norway <[EMAIL PROTECTED]> wrote:
>I thought we had established that SIN was non-deterministic. Are you
> saying that SIN is deterministic in some contexts, but not in others?
If we are talking about the current implementation of Derby, then all user
defined functions and built-ins are non-deterministic. As explained in
my previous comments, Derby does not have a way to declare user functions
deterministically, so they are all treated in non-deterministic manner implicitly.
What I was trying to convey was most of the built-in functions in Derby are *supposed*
to be deterministic (aggregates and SIN, COS, TAN, SUBSTR, etc.) since given the
same argument(s), they return the same result. And so far, I have been using
function determinisim to determine what is allowed in the group by _expression_ in
general and not limited to Derby. This is what I would like to describe in the
group by _expression_ documentation when Derby allows functions to be declared
determinisically in the future.
But for now, from a documentation perspective, the Derby ref manual should mention
what is allowed in the group by _expression_ in
10.2 due to implementation restrictions.
Regards,
Yip Ng
