Re: rounding problems

Craig Ringer wrote:
> Justin wrote:
>
>> No floating point is being used every variable is declared as numeric
>> on the Postgresql side and in the C++ which is the UI side
>> everything is double.
>
> `double' in C++ refers to double precision floating point. `double' is
> subject to all the usual fun with rational decimals being irrational
> binary floats (and vice versa).
>
> One of the reasons I chose Java for my current work is that it has a
> built-in decimal type (like `numeric') called BigDecimal . This makes
> working with exact quantities a lot easier as there's no conversion
> and rounding occurring each time data goes to/from the database.
Not according to MS specific if i'm reading it correctly

*Microsoft Specific >*

The double type contains 64 bits: 1 for sign, 11 for the exponent, and
52 for the mantissa. Its range is +/--1.7E308 with at least 15 digits of
precision
>
> Are there any particular decimal/numeric libraries people here like to
> use with C++ ? Or do you just use double precision floats and a good
> deal of caution?
>
> I'd expect that using double would be OK so long as the scale of your
> numeric values never approaches the floating point precision limit of
> the double type. I'm far from sure about that, though, and it'd be
> handy to hear from people who're doing it. Personally I like to stick
> to numeric/decimal types.
>
> --
> Craig Ringer