Unexpected semantics for util.is...

[deian]

The semantics for the util is... functions are somewhat unexpected. For example, the documentation for util.isRegExp says:

Returns true if the given "object" is a RegExp. false otherwise.

But I'm seeing that util.isRegExp(Object.create(RegExp.prototype)) === false.
Can we clarify the documentation to explain why this is not a RegExp or change the semantics of isRegExp please? (Same goes with the other is functions.) Thanks.

[Fishrock123]

The semantics almost certainly won't change. It is suggested you don't use these functions; I've been meaning to put deprecation notices into the docs.

How do you suggest we clarify that that won't work?

For posterity, the internals use Object.prototype.toString.call(input) === '[object RegExp]', since other methods can be bypassed, and this is the most reliable.

[Fishrock123]

See also #2201

[Ginden]

For posterity, the internals use Object.prototype.toString.call(input) === '[object RegExp]', since other methods can be bypassed, and this is the most reliable.

Symbol.toStringTag is implemented in Chrome and avalaible in io.js behind --es_staging flag.

Anyway - Object.create(RegExp.prototype) is not a RegExp.

> p = Object.create(RegExp.prototype);
 {}
> p.compile()
illegal access
> p.source
'(?:)'
> p.test('aaa')
TypeError: Method RegExp.prototype.test called on incompatible receiver [object Object]

ES6 allows to create classes extending RegExps, but it isn't implemented in any engine yet.

[deian]

@Fishrock123 thanks. I think the deprecation notice is a fine solution.

I'm not sure if it helps, but saying that the function returns true for objects created with the RegExp constructor would also be clear. (@Ginden a description what is not a RegExp is not necessarily useful.)

[Fishrock123]

Deprecated in the docs as of 93e2830, see #2447