Proposal: add func for `float` in `@math` pkg with postfix `f`

moonbitlang / core

MoonBit's Core library

https://moonbitlang.com/

Apache License 2.0

615 stars 78 forks source link

Proposal: add func for `float` in `@math` pkg with postfix `f` #1069

Open qazxcdswe123 opened 2 weeks ago

qazxcdswe123 commented 2 weeks ago

Unfortunately unless we bring back Num trait in X and make it generic, it's not possible to overload the function with different argument and same name.

IMO adding f to every Float function looks good enough, like:

lnf
logf
sinf
cosf
...

Originally posted by @tonyfettes in https://github.com/moonbitlang/core/issues/1050#issuecomment-2398611224

@bobzhang

bobzhang commented 2 weeks ago

we are planning to allow some overriding so that ln ( self : Float | Double) could both work

Lampese commented 4 days ago

we are planning to allow some overriding so that ln ( self : Float | Double) could both work

I have two worries about this solution:

I know we can use self as Float/Double, and then how to infer type if let x = self and use the funcitons of self?
There is a problem that the algorithm of sin/cos/tan.... is based on the IEEE754 and double's bit. If we implement same algorithm for float, then this algorithm will be error. I think sin(self: Float) -> self.to_double().sin().to_float() is even a good solution.