754772 - CSS rules are recognized as CSSNameSpaceRule

Status: VERIFIED FIXED

(Core :: DOM: CSS Object Model, defect)

Description

[Sebastian Zartner [:sebo]]

Testing a MozCSSKeyframesRule for "instanceof CSSNameSpaceRule" returns true.
See the attached test case for more info.

Sebastian

Comment 1

[Sebastian Zartner [:sebo]]

Attachment: Test case

Comment 2

[Sebastian Zartner [:sebo]]

Btw. MozCSSKeyframeRules are also recognized as CSSNameSpaceRules.

Sebastian

Comment 3

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I'm guessing this is because CSSNameSpaceRule doesn't have any of its own interfaces; it just implements the generic CSS rule interface.

Comment 4

[Sebastian Zartner [:sebo]]

Though a MozCSSKeyframeRule isn't an instance of a CSSNameSpaceRule.

I tried the following (via Firebug's Command Editor) to verify that:

function isInstanceOf(object, constructor)
{
    var proto = Object.getPrototypeOf(object);
    do {
        if (proto === constructor)
           return true;
        proto = Object.getPrototypeOf(proto);
    } while (proto !== null);
    return false;
}

isInstanceOf(document.styleSheets[1].cssRules[0], CSSNameSpaceRule.prototype);

Sebastian

Comment 5

[Boris Zbarsky [:bzbarsky]]

This is a bit weird.  CSSNameSpaceRule.prototype.__proto__ == CSSRule.prototype.  That part is fine.

Peter, is the issues that we end up QIing to nsIDOMCSSRule in some xpconnect hasInstance hook here?

In any case, new bindings, when they happen, will fix this.

Comment 6

[Sebastian Zartner [:sebo]]

Note that this problem also occurs for the new CSSPageRule.
A test case can be found at https://getfirebug.com/tests/manual/issues/6282/issue6282.html.

Executing document.styleSheets[2].cssRules[1] instanceof CSSNameSpaceRule will incorrectly return true.

Sebastian

Comment 7

[Sebastian Zartner [:sebo]]

Just stumbled over this again when trying to check an CSSCounterStyleRule.

As David and Boris indicate this actually happens for all types of CSS rules.

> Peter, is the issues that we end up QIing to nsIDOMCSSRule in some xpconnect hasInstance hook here?
@Boris, could it be that Peter didn't get this message? At least there's no Peter in the CC list.

Also you marked this issue being blocked by the big Paris Bindings issue. I guess the more specific issue this depends on is bug 851892, so I replaced the bug number. Please correct me if that's wrong.

Sebastian

Comment 8

[Sebastian Zartner [:sebo]]

ni'ing regarding comment 7.

Sebastian

Comment 9

[Boris Zbarsky [:bzbarsky]]

Looks like I forgot to cc Peter.  Anyway, the right fix here is the webidl conversion.

Comment 10

[Boris Zbarsky [:bzbarsky]]

Fixed in bug 851892.

Comment 11

[Sebastian Zartner [:sebo]]

I can confirm that it works in Nightly 53.0a1 2017-01-21. Thank you, Boris!

Sebastian