Chakracore: Module object Object.prototype.toString value casing seems incorrect

[guybedford]

With the following inside of a <script type="module"> tag:

import * as m from './m.js';
Object.prototype.toString.call(m) === '[object Module]';

The assertion here is returning false as the result is [object module] instead of what should I believe be [object Module] since the Symbol.toStringTag value is Module and not module.

Fixing this would help to ensure consistency across browsers for checking if a given object is a module namespace object, which may be needed for interop scenarios in future.

[dilijev]

@boingoing re: modules

[dilijev]

Candidate line where the issue is:

lib/Runtime/Library/JavascriptLibrary.cpp:1122:        moduleTypeDisplayString = CreateStringFromCppLiteral(_u("module"));

[dilijev]

For this script:

import * as m from './m.js';
console.log(Object.prototype.toString.call(m));
console.log(m[Symbol.toStringTag]);
console.log(Object.prototype.toString.call(m) === '[object Module]');

Output:

[object module]
module
false

It appears that our current implementation for @@toStringTag for modules is also lowercase (I think it uses the same string, so that's not surprising -- I just wanted to check on this since it sounded like you were saying our @@toStringTag for modules had different casing).

[guybedford]

Yes it seems both @@toStringTag and the display string should be using capital letters, as in the spec at https://tc39.github.io/ecma262/#sec-@@tostringtag.

Thanks for looking into this and so quickly. It would really help to avoid a painful edge case in detection.

[ljharb]

(fwiw, by convention every Symbol.toStringTag value should be PascalCased, a la [object Object]) Great catch!

[guybedford]

Just checking if there's any movement here?

[boingoing]

Sorry for being late to the party on this one. Doug is right, this is just a simple oversight. Quick fix is over here https://github.com/Microsoft/ChakraCore/pull/2430