From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Richard Neill <rn214(at)cam(dot)ac(dot)uk> |
Cc: | pgsql-bugs(at)postgresql(dot)org, peter_e(at)gmx(dot)net |
Subject: | Re: BUG #4972: RFE: convert timestamps to fractional seconds |
Date: | 2009-08-10 15:15:42 |
Message-ID: | 2360.1249917342@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-bugs |
Richard Neill <rn214(at)cam(dot)ac(dot)uk> writes:
> (b) Nowhere on the page is there a full example for getting
> seconds+microseconds since the epoch
Yeah, we could change that example to include a fractional part in the
timestamp to make this clearer.
> What I think I meant was dividing a differential timestamp by an
> interval. In this case, both should be unambiguously expressed in
> seconds, and the result will be dimensionless.
What you're missing is that intervals are not single numbers, and
are not simply numbers of seconds.
> Do you agree that an explicit cast of a timestamp to a double should work?
Doesn't seem like a particularly good idea. You're free to add such
a cast to your own DBs, of course.
> Do you agree that abs() should be able to operate on an interval?
> select abs( interval '-1 week');
It's not as easy as that, because (once again) intervals aren't single
numbers. For example, what should become of
abs(interval '-1 month +1 day')
The negative of this would be '1 month -1 day'. It's not real clear to
me whether abs() should give that or '1 month 1 day', ie, make all the
fields positive independently.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Hiroshi Saito | 2009-08-10 16:19:40 | Re: BUG #4961: pg_standby.exe crashes with no args |
Previous Message | Tom Lane | 2009-08-10 15:10:02 | Re: Huge speed penalty using <>TRUE instead of =FALSE |