> For same reasons, i.e. a need for precision, I find it hard to accept
> the idea of mixing positive and negative units in the same interval.
> The plus or minus sign should be outside of the interval.
The interval data type is really useful. I see no reason to restrict its
usefulness with an arbitrary constraint. Date arithmetic is treacherous
and INTERVAL is a lifesaver.
Forcing a global sign on the interval would break interval arithmetic.How would you compute '1 month'::interval - '1
week'::interval?
