NaN assertion is inconsistent with the standard and IEEE754

[zerkms]

The following test assertion (from https://github.com/chaijs/chai/blob/master/test%2Fassert.js#L332)

  it('isNaN', function() {
    assert.isNaN('hello');

    err(function (){
      assert.isNaN(4);
    }, "expected 4 to be NaN");
  });

must not pass, since the "hello" is definitely not a NaN value

Which is pretty much different from "it's not a number", but a special value defined by the standard and the IEEE754.

[zerkms]

Ok, seems like it's subjective, since the isNaN and Number.isNaN behave differently.

But at least the documentation should be changed.

Currently it states:

Asserts that the target is NaN.

technically it could state

Asserts that the target is not a number.

but then it also would be wrong since isNaN(false) // false.

So, any thoughts? ;-)

[keithamus]

Hey @zerkms thanks for the issue.

This is actually a duplicate of #681, which we found out is a dupe of #498 - and will be fixed in our 4.0.0 release, which is coming soon. I'll close this one, but thanks for filing this issue and please don't hesitate to open any new ones if you find anything new!