Re: Infinity vs Error for division by zero

Hi,

On 2019-03-01 12:46:55 -0500, Matt Pulver wrote:
> PostgreSQL FLOAT appears to support +/-Infinity and NaN per the IEEE 754
> standard, with expressions such as CAST('NaN' AS FLOAT) and CAST('Infinity'
> AS FLOAT) and even supports ordering columns of floats that contain NaN.
>
> However the query "SELECT 1.0/0.0;" produces an exception:
>
> ERROR: division by zero
>
>
> Question: If Infinity and NaN are supported, then why throw an exception
> here, instead of returning Infinity? Is it purely for historical reasons,
> or if it could all be done again, would an exception still be preferred?
>
> For purely integer arithmetic, I can see how an exception would make sense.
> However for FLOAT, I would expect/prefer Infinity to be returned.

It'd be good for performance reasons to not have to check for that, and
for under/overflow. But the historical behaviour has quite some weight,
and there's some language in the standard that can legitimate be
interpreted that both conditions need to be signalled, if I recall
correctly.

Greetings,

Andres Freund