828613 - typeof returns "function" for <object> element

Status: RESOLVED INVALID

(Core :: DOM: Core & HTML, defect)

Description

[DF]

User Agent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.11 (KHTML, like Gecko) Chrome/23.0.1271.91 Safari/537.11

Steps to reproduce:

var node = document.createElement("object");
alert(typeof(node));


Actual results:

A box containing the word "function"


Expected results:

A box containing the word "object"

Comment 1

[Paul Silaghi, QA [:pauly]]

I confirm on Mozilla/5.0 (Windows NT 6.1; WOW64; rv:21.0) Gecko/20130114 Firefox/21.0

Comment 2

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

This is correct:

HTML defines a legacycaller on HTMLObjectElement: <http://www.whatwg.org/html/#the-object-element>

WebIDL defines that a a platform object that implements an interface with at least one legacy caller must implement [[Call]]: http://dev.w3.org/2006/webapi/WebIDL/#call

ECMAScript defines that typeof returns "function" for "Object (native or host and does implement [[Call]])": http://ecma-international.org/ecma-262/5.1/#sec-11.4.3


(Paul, anything to do with HTML elements and DOM objects is most likely better suited for one of the DOM components than JS.)

Comment 3

[Paul Silaghi, QA [:pauly]]

(In reply to :Ms2ger from comment #2)
> This is correct:
Chrome shows "object". But I guess it's because of a different js engine ?

Comment 4

[Boris Zbarsky [:bzbarsky]]

In Chrome, the [[Call]] appears and disappears dynamically or something, which is not allowed per spec.  Please do file a bug on them, though!

Comment 5

[Paul Silaghi, QA [:pauly]]

(In reply to Boris Zbarsky (:bz) from comment #4)
> Please do file a bug on them, though!
Why shouldn't be this bug simply reopened?

Comment 6

[DF]

Every other browser i've tested with returns "object". While a strict reading of the standards does lead to the conclusion of "function", it requires a few leaps of logic which I don't believe were intended consequences of the standard.

Comment 7

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

(In reply to Paul Silaghi [QA] from comment #5)
> (In reply to Boris Zbarsky (:bz) from comment #4)
> > Please do file a bug on them, though!
> Why shouldn't be this bug simply reopened?

Because the bug should be filed on Google Chrome; this is not Google Chrome's bug tracker.

(In reply to DF from comment #6)
> Every other browser i've tested with returns "object". While a strict
> reading of the standards does lead to the conclusion of "function", it
> requires a few leaps of logic which I don't believe were intended
> consequences of the standard.

I can assure you these consequences were intended.

Comment 8

[Kohei Yoshino [:kohei]]

I just noticed this. Which bug made this change? The migration to WebIDL? It can be a site compatibility issue...

Comment 9

[Kohei Yoshino [:kohei]]

On Google Chrome 26, the current Stable channel, typeof HTMLElement === 'function'

Comment 10

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

(In reply to Kohei Yoshino from comment #8)
> I just noticed this. Which bug made this change? The migration to WebIDL? It
> can be a site compatibility issue...

It's been that way back to at least Firefox 2.

(In reply to Kohei Yoshino from comment #9)
> On Google Chrome 26, the current Stable channel, typeof HTMLElement ===
> 'function'

Note that this bug is about typeof document.createElement("object").

Comment 11

[Kohei Yoshino [:kohei]]

Then it won't be any problem. Thanks for the answer!