visual studio math.h inconsistencies?

[hisdudeness]

Why is sin(pi) not exactly 0? I have written a text parsing calculator for Windows which interprets command lines similar to how a TI-85 can. When I print answers in standard 'f' notation, sin(M_PI), M_PI as defined by math.h, prints 0.00000000... However when the answer is printed using the scientific notation 'g' flag in printf/sprintf using 12-digit precision, it appears as 1.22460635382e-16

Using <cmath> so everything appears under std::

So if the definition of M_PI appears in the same header with std::sin(double), why the discrepancy? I guess I could write in a check for sin,cos,tan, etc calls
with M_PI and -M_PI, and force them to return the correct values... but that
doesn't really sound like it should be my job... call me lazy. Or am I just
misinterpreting something?

edit note: cos(pi) appears correct. -1. as does sin(0), cos(0), cos(pi), and tan(0). tan(pi) however is the same story as sin(pi) only negative.

[helios]

There's no discrepancy. The result has too many zeroes, so it can't be completely converted to the decimal system. What's printed are just the first six decimal places of 1.22460635382e-16.
By the way, trigonometrical operations are so slow, that unless high precision is necessary, it is common to just pre-calculate (before compile-time) the results and put them in a static vector.

[hisdudeness]

Thanks. Yeah I just checked it out in GCC and got the same thing. I just never have really messed with trig functions much and assumed they were programmed to recognize the major arguments.