At 11:35 7/08/00 -0400, Tom Lane wrote:
>
>Perhaps the old way of considering equality only to float accuracy
>is more useful, even though it opens us up to problems like overflow
>errors in "float4var = 1e100". Comments anyone?
>
The following frightened me a little:
pjw=# select float4(10.1);
At 12:11 7/08/00 -0400, Tom Lane wrote:
>Philip Warner <[EMAIL PROTECTED]> writes:
>> pjw=# select float8(float4(10.1));
>> float8
>> --
>> 10.103814697
>> (1 row)
>
>> I would have expected the latter to be at worst 10.10 +/-
>> .01.
>
>float4 is
Philip Warner <[EMAIL PROTECTED]> writes:
> pjw=# select float8(float4(10.1));
> float8
> --
> 10.103814697
> (1 row)
> I would have expected the latter to be at worst 10.10 +/-
> .01.
float4 is good to about 7 decimal digits (24 mantissa bits)
> Perhaps the old way of considering equality only to float accuracy
> is more useful, even though it opens us up to problems like overflow
> errors in "float4var = 1e100". Comments anyone?
I would not have anticipated this either. I agree that downconverting to
float4 is the right solution.
Po
