Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Possible regression with XBL/replaceChild ordering

2 views
Skip to first unread message

Alex Vincent

unread,
Nov 30, 2009, 3:13:58 AM11/30/09
to
I had code which used to work in FF3.5:

var nodeparser = document.createElement("nodeparser");
expectedCell.replaceChild(nodeParser, cdata);
// nodeparser should now have a XBL binding on it, with a
defineNamespace method
nodeParser.defineNamespace("markup", MARKUP_NS);

In FF3.6b3, this is no longer working. Dump statements show
defineNamespace doesn't exist... and then the constructor for nodeparser
executes.

I have not yet tested FF3.6b4. I will gladly submit a reduced testcase
if desired, but I'd appreciate it if someone could run a quick
sanity-check first, using an existing global binding. Because I can't
verify it yet in 3.6b4, I don't want to file a bug.

Alex Vincent

unread,
Nov 30, 2009, 10:41:38 AM11/30/09
to
OK, I appear to be wrong. I ran the attached testcase (based on
test_481558.html) as a mochitest, and it failed in both FF3.5 and FF3.6.

I'm wondering why.

test_insertChild.html

Boris Zbarsky

unread,
Nov 30, 2009, 11:38:02 AM11/30/09
to
On 11/30/09 10:41 AM, Alex Vincent wrote:
> OK, I appear to be wrong. I ran the attached testcase (based on
> test_481558.html) as a mochitest, and it failed in both FF3.5 and FF3.6.
>
> I'm wondering why.

Because loading XBL over HTTP is not synchronous.

-Boris

0 new messages