Created attachment 314230
update XBL even when not notifying
Jst, Jonas, as far as I see this bug has actually always been there. Bug 395609 just made it visible in this case.
If XBL is bound to some element while content sink (without notifying) still
creates the DOM, XBL may not be up-to-date.
So, the idea in this patch is to update XBL if the parent of inserted/append
/removed node is somehow in XBL, even if aNotify is PR_FALSE.
The change to wizard.xml is needed to delay some initialization.
DOMContentLoaded is fired after initial reflow.
Passes mochitest, chrome and browser tests.
Initial testing shows that the special handling for XBL is called very rarely,
basically in FF UI only when opening 'ReportBroken' dialog.
If this is too scary (though shouldn't be that scary, and IMO is the right
thing to do), Jst's "delay" patch might be good enough.
And now it is really time to sleep :)
Created attachment 314230
update XBL even when not notifying
Jst, Jonas, as far as I see this bug has actually always been there.
Bug 395609 just made it visible in this case.
If XBL is bound to some element while content sink (without notifying) still
creates the DOM, XBL may not be up-to-date.
So, the idea in this patch is to update XBL if the parent of inserted/append
/removed node is somehow in XBL, even if aNotify is PR_FALSE.
The change to wizard.xml is needed to delay some initialization.
DOMContentLoaded is fired after initial reflow.
Passes mochitest, chrome and browser tests.
Initial testing shows that the special handling for XBL is called very rarely,
basically in FF UI only when opening 'ReportBroken' dialog.
If this is too scary (though shouldn't be that scary, and IMO is the right
thing to do), Jst's "delay" patch might be good enough.
And now it is really time to sleep :)