Microsoft Calculator mistakes really [ Try it Now ]

[oskoskarsdottir]

-8,164846595551428716852118012­2928e+39=16,8056459=2-2 ????

[ashridah]

Right, because i expect perfect precision out of floating point arithmetic for an algorithm that's already known to be a fast approximation when you use the standard C libraries.

As far as i'm concerned, if you consider this a bug, you fail at algorithms in CS :P

If you want precision to arbitrary degrees, go use matlab or mathmatica or something

[JoeyTribiane]

@ashridah exactly---similarly, one can reference "mathlab"...as in, Carl Engelman, "Macsyma", "MATHLAB-68"---yes, I do have a PDP-10 and PDP-11---my company services them still, believe it or not, as well as other DECs, VAXs, etc. DECNET...YAY! TCP/IP over DECNET/LAT...GO CISCO!

[vinc17fr]

@ashridah The algorithms should be accurate enough to provide at least faithful rounding (internally). In particular, for any function provided by the calculator, if the mathematical result is exactly representable, it should be returned.

[vgfeit]

core dumped (once again)