Prototype $$ method fails due to missing className on DOMNode

I've added a Prototype compatibility test and a patch for this on my GitHub fork. Here are the commits:

Prototype compat test: http://github.com/Lawjoskar/env-...

Patch: http://github.com/Lawjoskar/env-...

Ok I think the relavant code from prototype is

var nodes = $(element).getElementsByTagName('*');
    className = ' ' + className + ' ';

    for (var i = 0, child, cn; child = nodes[i]; i++) {
      if (child.className && (cn = ' ' + child.className + ' ') && (cn.include(className) ||
          (classNames && classNames.all(function(name) {
            return !name.toString().blank() && cn.include(' ' + name + ' ');
          }))))
        elements.push(Element.extend(child));
    }
    return elements;
  };

so it is iterating what should be a list of html elements so having className on the html Element interface should be enough. I need to double check but I think the actual problem may be in the getElementByTagName code. for the case where the name is '*'.

Yeah so the cleaner patch in this case is to getElementsByTagNameRecursive

Before we had:

if (elem.nodeType == DOMNode.ELEMENT_NODE || elem.nodeType == DOMNode.DOCUMENT_NODE) {
    
        if(elem.nodeName.toUpperCase() == tagname.toUpperCase() || (tagname == "*")){
            __appendChild__(nodeList, elem);               // add matching node to nodeList
        }

This put the Document Node in the list. ;(

So I changed it to this:

if (elem.nodeType == DOMNode.ELEMENT_NODE || elem.nodeType == DOMNode.DOCUMENT_NODE) {
    
        if(elem.nodeType !== DOMNode.DOCUMENT_NODE && 
            ((elem.nodeName.toUpperCase() == tagname.toUpperCase()) || 
                (tagname == "*")) ){
            __appendChild__(nodeList, elem);               // add matching node to nodeList
        }

so it'll only add elements to the list but still allow recursion on the docuement node.

You test passes then!

I'll close this ticket when I commit.