Why is there no pow function for integers?

# announcements

[bjonnh]

Why is there no pow function for integers?

[bjonnh]

One explanation I see (on SO) is that it would quickly overflow an Integer anyway. But I wonder if that's the reason behind the design choice (of Java it seems)

[adamratzman]

My theory is that there’s no Int.NaN, but there is a Float.NaN and Double.NaN, and I don’t think the stdlib designers would consider a pow function “correct” without accounting for the cases where NaN should be returned

[Nir]

Why does NaN have anything to do with it?

[Nir]

I don't follow that theory at all

[Nir]

Exponentiation is better defined than division

[Nir]

And you can divide integers even with no way to account for dividing by zero

[adamratzman]

https://docs.oracle.com/javase/8/docs/api/java/lang/Math.html#pow-double-double-

[Nir]

An integer to the power of an integer is well defined for all values

[Nir]

You only get NaN for NaN inputs or infinite

[Nir]

Which integers don't support, so they can't be inputs

[adamratzman]

They can be infinite, but not nan

[Nir]

Integers can't be infinite

[Nir]

Unless this is some crazy feature of JVM integers that nobody else uses

[Nir]

Maybe a more compelling reason is Int.power(Int) returns a non Integer for any negative second argument

[adamratzman]

Oh wow, I mistakenly thought there was an Int.POSITIVE/NEGATIVE_INFINITY

[adamratzman]

disregard what I said then

[Nir]

So perhaps the result needs to be floating point regardless

[Nir]

But it would still be useful for convenience IMHO

[Nir]

Yeah floating point has infinite because it's part of IEEE and had native support

[Nir]

(NaN too)

[Nir]

With Int you'd need to emulate it all with branching and perf is bad

[Dominaezzz]

https://kotlinlang.slack.com/archives/C0922A726/p1502295729174417?thread_ts=1502289508.255680&cid=C0922A726

[Nir]

This isn't the same as the question on this thread, just FYI