I originally asked about this on the Google Group, but GH seems to be more active, so I'm re-posting here (plus markdown!!)

I'm not sure if I've run into a bug, or if I'm misusing GopherJS, but I've run into a problem with the way my Go code receives errors passed by Javascript Callbacks. To demonstrate the problem, I've built this code:

My main.js:

window.foo = {
    nothing: function(cb) {
        return cb(null);
    }
}
require('main');

And main.go:

package main

import (
    "github.com/gopherjs/gopherjs/js"
    "honnef.co/go/js/console"
)

func main() {
    result := nothing()
    console.Log(result)
    if result == nil {
        console.Log("result is nil")
    }
    if result != nil {
        console.Log("result is not nil")
    }
}

func nothing() *js.Error {
    c := make(chan *js.Error)
    go func() {
        js.Global.Get("foo").Call("nothing",func(err *js.Error) {
            c <- err
        })
    }()
    return <-c
}

The output on my javascript console (in Chrome 43, FWIW) is:

null
result is not nil

If I do a s/*js.Error/*js.Object/ in main.go, I get the expected result:

null
result is nil

So it seems that somehow the Error struct around *js.Object is obfuscating the nullness of the javascript object? Am I doing this wrong?