Brent Verner <brent@rcfile.org> writes:
> This seems fair. Would this approach imply that CURRENT_TIME and
> CURRENT_TIMESTAMP should not apply default precision to their return
> values? Right now, "CURRENT_TIME" is equivalent to "CURRENT_TIME(0)"
> and "CURRENT_TIMESTAMP" eq to "CURRENT_TIMESTAMP(6)".
Yes, I had been thinking that myself, but hadn't got round to mentioning
it to the list yet. (Even if you do accept default precisions for time
& timestamp columns, I can see nothing in the spec that justifies
applying those default precisions to CURRENT_TIME/TIMESTAMP. AFAICS,
the precision of their results when they are given no argument is
just plain not specified.)
regards, tom lane