384716 - xbl fields override arbitrary js properties when the binding is eventually attached

Status: RESOLVED INCOMPLETE

(Core :: XBL, defect)

Description

[Asaf Romano (gone)]

In places menu.xml, we do something like this:
  document.addBinding(popup, MENU_URI);
  popup._containerNodesMap = this._containerNodesMap;

If I switch the order here, popup._containerNodesMap is nulled out. This is even worse on non-mac because the binding is applied asynchronously (as a result of |popup.setAttribute("type", "places")|), see bug 384690.

Comment 1

[Olli Pettay [:smaug][[e-mail]]]

Any chance to get the regression range?

Comment 2

[Asaf Romano (gone)]

aha, ignore me.

So here's what really happens here,
1) The xbl binding has a <filed> named _foo, set to null.
2) when we set newElement._foo, newElement is not in the document yet, thus the element is not bounded; explicit addBinding calls are ignored too.
3) when the element is eventually bounded, it's _foo field override whatever newElement._foo was set too.

This is arguably the correct behavior I guess...

Comment 3

[Asaf Romano (gone)]

Actually, even if the field isn't around, explicit call to addBinding nulls out the js property; implicit attaching (via css) seems to work fine though.

Comment 4

[Boris Zbarsky [:bzbarsky]]

If it's a regression, what's the regression range?  In any case, where's a testcase?

Comment 5

[Asaf Romano (gone)]

Dunno, bug 337855 exposed these behaviors, I'm no longer sure at all that it is a regression (well, I'd hope comment 3 is).

Comment 6

[Boris Zbarsky [:bzbarsky]]

Interesting.  I'd still love a testcase and regression range for whatever issues you see here.

Comment 7

[Johnny Stenback (:jst)]

Not a blocker based on the info available. If we'd get a regression range, or testcases showing that this really is a regression, then please renominate.

Comment 8

[Boris Zbarsky [:bzbarsky]]

Bug 372769 might fix this bug as currently summarized.  Hard to tell without a testcase...  Care to attach one?

Comment 9

[Boris Zbarsky [:bzbarsky]]

Still waiting on testcase to resolve this.

Comment 10

[Boris Zbarsky [:bzbarsky]]

No response, no testcase... resolving incomplete.