Function fmod returns different values than C++ when mixing signs

Hello, I am writing some unit tests on nannou for a university course . I found that fmod has some slight differences compared to the C/C++ implementation. These specific examples are from the cppreference page on fmod.

C/C++ (tested with gcc 14.2.1 on fedora)

fmod(5.1, 3.0)   //  2.100000
fmod(-5.1, 3.0)  // -2.100000
fmod(5.1, -3.0)  //  2.100000
fmod(-5.1, -3.0) // -2.100000

nannou math::fmod

fmod(5.1, 3.0)   //  2.0999999999999996 ✅
fmod(-5.1, 3.0)  //  0.9000000000000004 ❌
fmod(5.1, -3.0)  // -0.9000000000000004 ❌
fmod(-5.1, -3.0) // -2.0999999999999996 ✅

As you can see, tests 2 and 3 are different. I believe this is due to the usage of floor (lower integer) rather than truncate (integer towards zero). The man page for fmod includes the following description:

These functions compute the floating-point remainder of dividing x by y. The return value is x - n * y, where n is the quotient of x / y, rounded toward zero to an integer.

So I believe using trunc() here is the correct move if we want to be close to the C implementation. Using trunc results in the expected results (see Rust Playground here)

If the project agrees, I'd like to submit a PR and unit tests for this change. Thanks!

nannou-org / nannou

Function fmod returns different values than C++ when mixing signs #983