> > Postgres assumes that a NULL input will give a NULL output,
> But why? That is not true in all case, i mean so like: "FALSE &&
> dont'know" is always FALSE.
Your example shows a flaw in the Postgres premise on this topic,
perhaps.
> > and never calls your routine at all.
> But! I see the output of elogs in function.
> The 6.4.x calls my
> functions always (with one or more NULL parameters).
It's been discussed before, and as you and others note it seems the
behavior has changed so that functions are called even with NULL
input. But the job wasn't finished since the results are ignored.
- Tom