> No, that's the point of the sentence about the range mechanisms not
> understanding special values of the underlying type. upper_inf tests
> for a range with no upper bound, period. It would need some
> type-specific special knowledge to do what you're hoping for, and it
> has not got that.
Got that, thanks Tom!
So I can simply use
select tstzrange('2013-05-01'::timestamp, 'infinity'::timestamp, '[]') @> 'infinity'::timestamptz as validity;
validity
----------
t
(1 row)
to test for the case where the upper bound is given (= not null) and infinite.
And when one formerly used two timestamptz fields (valid_from, valid_until) to simulate the range type before the range
typewas implemented and checked for
WHERE valid_until IS NULL
that's where we now use the "upper_inf()" function.
Regards
Ralph