1261763 - walk the view tree when destroying display item data in nsSubDocumentFrame

Status: RESOLVED INVALID

(Core :: Layout, defect)

Description

[Timothy Nikkel (:tnikkel)]

Anything in the view tree has the potential to be painted, and therefore has stored display item data. The list of subdocuments stored on the document may not be in-sync with the view tree.

Comment 1

[Timothy Nikkel (:tnikkel)]

Attachment: patch

Comment 2

[Mats Palmgren (inactive)]

(In reply to Timothy Nikkel (:tnikkel) from comment #0)
> The list of subdocuments stored on the document may not be in-sync with the view tree.

This seems a bit far fetched to me.  Do you have any indication that we might
have views/frames that aren't reachable from any (sub-)document entry?
That seems like a rather severe bug if that's the case (and we should fix 
the root cause of it if so).

Comment 3

[Mats Palmgren (inactive)]

After reading bug 1261758, I'm guessing the views you're talking about are from previous
navigations?  Do we have these in BeginSwapDocShells?  If so, what state are those view /
frame trees (and their associated pres shell/contexts) ?  Where is the code that puts
the view tree in this state?

Comment 4

[Timothy Nikkel (:tnikkel)]

(In reply to Mats Palmgren (:mats) from comment #3)
> After reading bug 1261758, I'm guessing the views you're talking about are
> from previous
> navigations?

The case I am thinking about is _during_ a current navigation. Bot the old and new document have presentations that are linked into the view tree


(In reply to Mats Palmgren (:mats) from comment #2)
> (In reply to Timothy Nikkel (:tnikkel) from comment #0)
> > The list of subdocuments stored on the document may not be in-sync with the view tree.
> 
> This seems a bit far fetched to me.  Do you have any indication that we might
> have views/frames that aren't reachable from any (sub-)document entry?

I think you might be right, after investigating some testcases locally. I'll withdraw this patch.