(03:22:06 PM) toshok: does mozilla "lazily" load plugins to associate them with <object> tags?
(03:22:40 PM) toshok: the stupid silverlight.js code creates a div, and sets its innerHTML to <object type="...."/>, then gets div.childNodes[0] and calls methods on it
(03:22:58 PM) doron: toshok: correct
(03:23:02 PM) toshok: everything about the object is coming up undefined, and i'm not seeing the plugin load
(03:23:03 PM) doron: in fact, innerHTML is lazy
(03:24:08 PM) toshok: doron: the object itself is returned, though - so the HTMLObjectElement is being created
(03:25:20 PM) doron: toshok: usually the way that is done is that plugins can call back into javascript once they are all done with their gunk
(03:26:14 PM) stuart: toshok: check the mwcontext
(03:26:21 PM) toshok: shaddup :)
(03:26:53 PM) toshok: this is the offending js:
(03:26:55 PM) toshok: http://pastebin.ca/919754
(03:27:17 PM) toshok: b.isVersionSupported is not a defined method, according to firefox
(03:28:30 PM) toshok: not sure what that document.body.innerHTML; line is there, except perhaps an attempt at either causing the plugin to load or delaying execution of the remainder of the JS in order to wait for the plugin to load (if plugins are loaded lazily)
(03:31:28 PM) toshok: in fact that bit of JS seems to act differently between ff2 and ff3
(03:31:42 PM) toshok: in ff2, for instance, the method call succeeds. in ff3, it doesn't.
(03:32:37 PM) toshok: is timing going to be an issue in ff3? or can js depend on plugins being instantiated synchronously when dom elements are created/inserted?