Extend HTMLCollection.prototype with Firefox?

Is it possible to extend the HTMLCollection prototype in Firefox (>=2.0)?
It looks like I can do it, but it doesn't work:

HTMLCollection.prototype.funk = function() { alert(this.length); }
window.onload= function() {
var x = document.getElementsByTagName('div');
alert(HTMLCollection.prototype.funk); // function() { ... }
alert(x instanceof HTMLCollection); // true
alert(x.item===HTMLCollection.prototype.item); // false
x.funk(); // error: x.funk is not a function
}

Is there any other way to do it that will work right now?

--
Matt Kruse
http://www.JavascriptToolbox.com
http://www.AjaxToolbox.com

re: Extend HTMLCollection.prototype with Firefox?

On Feb 2, 4:58 am, Matt Kruse wrote:The - getElementsByTagName - method is specified as returning a
NodeList, but may in reality return a NamedNodeMap. If you want to
draw conclusions about objects implementing HTMLCollection shouldn't
you use a source that is specified as being one, such as -
document.forms -?

Richard.

re: Extend HTMLCollection.prototype with Firefox?

Richard Cornford wrote:Doesn't the line:

alert(x instanceof HTMLCollection); // true

indicate that it is an HTMLCollection?


--
Rob

re: Extend HTMLCollection.prototype with Firefox?

Richard Cornford wrote:I don't know much in this area, but I thought NodeList was the interface,
and that HTMLCollection was the implementation. I tried to extend NodeList
also, but I got the same results. Also, since (x instanceof HTMLCollection)
== true, I thought I was on the right track. If it doesn't return an
HTMLCollection object, why would that check be true?
I just tried that as well with the same results.

Nevertheless, do you know which object, if any (except Object), I could
extend to add a method to the results returned by getElementById()?

--
Matt Kruse
http://www.JavascriptToolbox.com
http://www.AjaxToolbox.com

re: Extend HTMLCollection.prototype with Firefox?

Matt Kruse wrote:HTMLCollection is an interface, just line NodeList.
The - instanceof - operator is not necessarily usefully discriminating
in javascript.
Good, as starting with something that is supposed to be an -
HTMLCollection - eliminates possible sources of interference.
Remember that - NodeList - and - NamedNodeMap - are specified (in the
ECMAScript bindings) as mapping bracket notation (and therefore, due to
ECMA 262 provisions, dot notation) property accessors onto their - item -
methods, and - HTMLCollection - is specified as mapping them onto its -
item - and - namedItem - methods. So your - x.funk - may be equivalent
to - x.namedItem('func') -, and as - namedItem - may only return an
object implementing the - Node - interface or null it should not be too
surprising if you don't get access to your method.
Objects implementing the - Element - interface are returned by-
getElementById -.

Richard.

re: Extend HTMLCollection.prototype with Firefox?

RobG wrote:<snip><snip>It indicates that - HTMLCollection.prototype - is on the prototype chain
of - x -, assuming that as a host object - HTMLCollection - does not take
advantage of its option to provide a non-standard - [[HasInstance]] -
method. That does not even imply that the object - x - implements the -
HTMLCollection - interface.

Richard.

re: Extend HTMLCollection.prototype with Firefox?

Richard Cornford wrote:That makes sense, I hadn't even considered that.
Oops, what I meant was getElementsByTagName() of course. Do you know of any
way to add methods to the prototype of the Object returned by this call in
Firefox? (Other than Object.prototype).

--
Matt Kruse
http://www.JavascriptToolbox.com
http://www.AjaxToolbox.com

re: Extend HTMLCollection.prototype with Firefox?

Matt Kruse wrote:<snip>Then that would be an object implementing the - NodeList - interface.
I have never tried, but as - NodeList - maps property accessors to its -
item - method I don't expect you will successfully extend that object.

On the other hand, why? You can write a function that takes a -
NodeList - as an argument and do anything to/with that - NodeList - that
you could with a method.

Richard.