Closed Bug 725210 Opened 12 years ago Closed 12 years ago

Add internal observer notification for last PB docshell going away


(Core :: DOM: Navigation, defect)

Not set





(Reporter: jdm, Assigned: jdm)




(2 files, 2 obsolete files)

      No description provided.
Attached patch <required> (obsolete) — Splinter Review
Attachment #595318 - Flags: review?(bzbarsky)
Attachment #595319 - Flags: review?(bzbarsky)
Attachment #595318 - Attachment is obsolete: true
Attachment #595318 - Flags: review?(bzbarsky)
Assignee: nobody → josh
Blocks: PBnGen, 722845
Blocks: 722859
Comment on attachment 595319 [details] [diff] [review]
Add observer notification for last PB docshell going away.

Seems ok, but do you need to stabilize the count around going back to session restored stuff, so that we don't end up bouncing off zero for the subframes?
Attachment #595319 - Flags: review?(bzbarsky) → review+

I added an assert to the function to be safe, but I think the counting should be ok.
Blocks: 722850
Blocks: 722981
Attachment #595319 - Attachment is obsolete: true
Whiteboard: [needs landing]
Blocks: 722857
Can this land, Josh?
Yes, I just haven't got around to it.
Blocks: 722861
Blocks: 722868
Whiteboard: [needs landing]
Target Milestone: --- → mozilla14
Attached patch Test caseSplinter Review
Attachment #611548 - Flags: review?(josh)
I don't think the whole window close + gc dance should be necessary; this is meant to fire after the last docshell leaves private mode. Why doesn't setting the privateWindow flag to false trigger it?
Because the value is decremented when the docshell gets destroyed, and that happens only when a CC is run after the window has been closed.

Do you have any better suggestions?
Comment on attachment 611548 [details] [diff] [review]
Test case

It makes me sad that this is necessary, but ok.
Attachment #611548 - Flags: review?(josh) → review+
Closed: 12 years ago
Resolution: --- → FIXED
Blocks: 722979
You need to log in before you can comment on or make changes to this bug.