Re: Precision of data types and functions

I wrote:
> No, I think Bruce fixed this recently.  It's just a cosmetic mistake in
> the error message so we didn't back-patch it.

No, strike that, I remember the discussion now.  The pre-8.2 code is
correct on its own terms, which is that it's telling you what size
number you tried to put in:

regression=# select '123123123123123.2'::numeric(12,2);
ERROR:  numeric field overflow
DETAIL:  The absolute value is greater than or equal to 10^14 for field with precision 12, scale 2.
regression=# select '12123123123123123.2'::numeric(12,2);
ERROR:  numeric field overflow
DETAIL:  The absolute value is greater than or equal to 10^16 for field with precision 12, scale 2.
regression=#

Bruce and some other people thought this was confusing, so it's been
changed for 8.2.

            regards, tom lane