Rust's F{32,64}::rem_euclid behaves differently from C/C++ remainder about faust

[sletz]

Thanks, do you have a fix for that ?
(I've just found that https://users.rust-lang.org/t/analog-of-c-std-remainder/59670/13)

[nuchi]

I don't have a fix, nor do I particularly need one. I'm not a rust user, nor am I using remainder for anything. I just happened to stumble on this and thought I should report it in case any rust users run across mysterious errors and search for it here :)

It looks like the jax backend has a similar issue (there's a "todo" comment next to the remainder line in jax_instructions.hh) so maybe a similar solution would work for rust (define a custom remainder in the architecture file).

But I expect it can wait until a contributor shows up who cares more about rust than I do.

[DBraun]

Thanks for the reminder about JAX. I missed the situation in which the second argument is negative... But I feel better about the new PR here #1001.

It's funny how remainder means something different in each library: jax.numpy.remainder is not std::remainder.

[sletz]

Done in 3fdc3d5
following this discussion: https://users.rust-lang.org/t/analog-of-c-std-remainder/59670