FF 18 problem?

[stevestorey]

I too can reproduce this issue, when adding > 38 nodes to a TreePanel as a result of an XHR request (that's not necessarily the minimal test case, merely what I've got to so far ...)

It's clearly a regression in IonMonkey, for which I've raised https://bugzilla.mozilla.org/show_bug.cgi?id=828319 along with a basic stacktrace of how I got to see the problem (without Firebug - as using their debugger disables Ion). There is a similar bugzilla report of https://bugzilla.mozilla.org/show_bug.cgi?id=818023 which suggests that it's possibly related to the use of apply() as well (which ExtJS clearly uses a lot).

I don't have a public test-case I provide yet - however I note that at least for me, the Tree examples hosted on cdn.sencha.io don't actually load any tree panel content at the moment, which might make it easier to show the problem if some of them are loading lots of nodes at the same time?

Steve

[Tyler___Durden]

Great, thank you for your efforts...

Cheers

[Holger]

Hi there,

after hours of testing I found a workaround for this problem I think. Seems pretty strange though:

if (Ext.firefoxVersion >= 18) {
    var noArgs = [];
    Ext.override(Ext.Base, {
        callParent : function(args) {


            var method, superMethod = (method = this.callParent.caller) &&
                    (method.$previous || ((method = method.$owner ?
                            method :
                            method.caller) && method.$owner.superclass[method.$name]));


            // Workarround for Firefox 18. I don't know why this works, but it does. Perhaps functions wich have
            // a try-catch block are handled differently
            try {
            } catch (e) {
            }


            return superMethod.apply(this, args || noArgs);
        }
    });
}

Greeting Holger

[Tyler___Durden]

Weird, works for me too, btw, WTH

[stevestorey]

I now have a reproducible test case that you can try here: http://adhoc.moo.com/ionmonkey-regression-test.html - remember to disable Firebug before trying it.

[evant]

Thanks for the test case, I'll check it out.

It seems like it might be an issue similar to what we saw in Chrome, it would happen almost randomly, due some changes to the JIT http://code.google.com/p/chromium/is...tail?id=157520

[interconcept]

Holger, your code really works also with our applications. Our problems were similar: some parts/or whole components were not loaded when firebug was disabled/not on (FF 18 only of course).

[jordandev]

I have started to get this problem as well. Seems intermittent and I cannot consistently recreate it with a test case.

[Vasiliy Faronov]

I can confirm that this happens with my app as well, and that Holger’s workaround helps. Thanks Holger.

[sebbio]

And I can confirm that this is happening with my application as well, and that the workaround Holger does not help.