Re: Conversion errors for datetime fields

From: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
To: lfrittelli(at)tutopia(dot)com, pgsql-bugs(at)postgresql(dot)org
Cc: Thomas Lockhart <lockhart(at)alumni(dot)caltech(dot)edu>
Subject: Re: Conversion errors for datetime fields
Date: 2000-12-28 16:28:14
Message-ID: 5421.978020894@sss.pgh.pa.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

pgsql-bugs(at)postgresql(dot)org writes:
> select '0:00:59.99999999999999'::time as fourteen_dec,
> '0:00:59.999999999999999'::time as fifteen_dec;

> -- Output from psql
> -- fourteen_dec | fifteen_dec
> ----------------+-------------
> -- 00:00:59 | 00:00:00
> --(1 row)

What I'm getting with current sources is

regression=# select '0:00:59.99999999999999'::time as fourteen_dec;
fourteen_dec
--------------
00:00:59
(1 row)

which seems to be rounding in the wrong direction, and

regression=# select '0:00:59.999999999999999'::time as fifteen_dec;
ERROR: Bad time external representation '0:00:59.999999999999999'

That one seems reasonable, since this input is indistinguishable from

regression=# select '0:00:60'::time;
ERROR: Bad time external representation '0:00:60'

However, I also get

regression=# select now()::date+'0:00:59.999'::time;
?column?
---------------------------
2000-12-28 00:00:60.00-05
(1 row)

The cause is clear enough: the 59.999 seconds are being rounded off
to two digits for display. But it's unfortunate that this causes a
displayed output that will not be accepted as valid input. Perhaps
it would be a good idea to round off the seconds to display precision
*before* the value is broken down to hh/mm/ss.

regards, tom lane

In response to

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Larry Rosenman 2000-12-28 17:16:25 Re: Conversion errors for datetime fields
Previous Message pgsql-bugs 2000-12-28 15:55:52 Conversion errors for datetime fields