How to handle floating-point array indices in Neo?

[AlienKevin]

Dear Mr. Crockford, I like your idea of combining floats and ints into one number type. One unified number type makes sense for most math operations and saves a lot of headache associated with handling two number types. However, when we are doing array indexing, how can we handle floating-point indices? I know that all array indices in JavaScript are strings so you can store values into floating-point indices:

let a = []
a[1.5] = 10
console.log(a[1.5]) // outputs: 10

To me, JS's behavior is both strange and confusing as many people including me regard array indices as whole numbers and expect JS to floor or round floating-point indices to whole numbers. What approach do you recommend to deal with floating-point indices? Is this a necessary evil for adopting a unified number type?

[douglascrockford]

I agree that JS is much too permissive here. Strings should not be used as array indexes. On reading, a[1.5] should produce null because there is no such element. On writing, it should fail.