On Sat, 17 Jun 2000, Christopher Browne wrote:
> _Can't_ we assume 64 bit integers?
No, you should assume that there is a struct definition. Period.
> I would strongly argue _against_ gmp: It means that money values are
> no longer of fixed size.
Although I strongly agree that gmp is of any real value in the implementation,
"variable size" is not a real reason because the class should provide the
mechanisms to handle ALL allowed operations. Conversion to/from storage
formats is a part of that.
> Floating point systems are indeed problematic to "prove right;"
> basic operators +, -,
> and * work properly, with some well-defined rounding rules.
> gmp is interesting, but I think it's no alternative to these.
Further, gmp does nothing to help "prove" the correctness of the monetary
calculations. And that is the harder part of "getting it right"
At least in theory, the implementation of this class should not be visible
to the outside world.
If I am in a CORBA world, I might import a different set of definitions from
those that I use somewhere else. The gnucash code should not care that I
choose BCD, or int64, or FP as long as I get the implementation correct.
--
Gnucash Developer's List
To unsubscribe send empty email to: [EMAIL PROTECTED]
