_.isNaN returns true for whatever created with new Number() instead of just NaN and new Number(NaN)

[debutt]

// All => true
_.isNaN(new Number(3))
_.isNaN(new Number(Infinity))
_.isNaN(new Number('donut'))

If _.isNaN is intended to return true for only NaN and new Number(NaN), I think in line 1304, obj !== +obj should be replaced with obj != +obj.

Correct me if I'm wrong : )

[michaelficarra]

I don't understand what you're trying to say.

[bjmiller]

I don't know if his proposed solution will work. But, if _.isNaN(new Number(3)) returns true, that's probably a bug.

[megawac]

Feel free to send a PR with the change to obj != +obj

[debutt]

If we change the code to obj != +obj, _.isNaN still returns true for things like new Number('dice'), but I guess that's effectively the same as NaN in operations : )

[debutt]

Since != won't make it through the build, I suggest we can use the good old isNaN with _.isNumber, thus

_.isNumber(obj) && isNaN(obj)

_.isNumber makes sure obj is a number, primitive or wrapped, and then isNaN makes sure only NaN and new Number(NaN) pass the test.