Huh, you know. Now that I'm not at work and therefore don't have my
numbers to check against, you're right.
Man, I must need more coffee.
Never mind me. :)
May be back tomorrow, though, when I have the numbers in front of me. I
know they didn't add up earlier...
Martin
Hassan Schroeder wrote:
> Martin wrote:
>
>> My recent test involved the following three values from the column:
>> 2.4950
>> 2.5950
>> 2.7700
>
>
>> When I use a SUM() on these I get: 7.860
>
>
> Sounds good to me...
>
>> If I switch the column over to a FLOAT, then the SUM() becomes
>> 7.8599998950958
>>
>> Using Excel to test the numbers, or hand-calculating, I get:
>> 7.8550.
>
>
> Time for a hand upgrade, I think :-) 5 + 5 = 5??? I don't even
> want to think about how Excel would come up with this...
>
>> Shouldn't the SUM() remain with the precision of the DECIMAL type and
>> not try to round to 2 decimal places?
>
>
> My own, possibly suspect, hand calculations show that SUM() is right;
> and it's common knowledge that floating point isn't the right thing to
> use for situations like this -- that's why there *is* a DECIMAL type.
>
> FWIW!