Il giorno , Adrien Destugues <pulkomandy@xxxxxxxxxxxxxxxxx> ha scritto:
You're doing things in fixed precision, so you should't use floats or
doubles.
Do all your math using cents, then *divide* by 100 in the end to get a
value in euros, dollars, or whatever. This way there will be no
rounding error.
I do all math in cents and then divide by 100 and cast to int, and I've rounding errors...
guaranteed :-)
It's easy enough in C++ to write some support classes for your own
software if you need to do this a lot. I see no need for doing it at
compiler level.
--
Adrien.
Well Adrien, YOU, have NOT to do nothing, GCC has already done it!_Decimal32 is a ***native*** type in C and decimal32 is already the class I want in C++ :-) simply we've to find the way to enable them in the compiler, yes I can find a way (in c++, in c it's very difficult in an elegant way, IMHO) to do an imitation of them, but why reinvent the wheel?
Thanks fano