Closed
Bug 54524
Opened 25 years ago
Closed 25 years ago
FindFrameWithContent doesn't deal with XBL <children> insertion points
Categories
(Core :: XBL, defect, P3)
Tracking
()
RESOLVED
FIXED
People
(Reporter: hyatt, Assigned: hyatt)
Details
(Keywords: regression, Whiteboard: [xbl1.0][need info])
Attachments
(1 file)
1.18 KB,
patch
|
Details | Diff | Splinter Review |
We're ending up doing all sorts of bad things (like mistakenly recreating
frames) because the function FindFrameWithContent isn't savvy enough about
XBL-interleaved anonymous content. It fails to find frames because it assumes
the content trees and frame trees match perfectly (an incorrect assumption).
This is why the XBL sticky note test is broken.
Assignee | ||
Comment 1•25 years ago
|
||
This is a regression in my XBL test suite. I'm guessing something related to
:hover got patched or fixed to be better recently and that exposed the bug in
FindFrameWithContent.
Will attach patch shortly. Nominating dogfood, since I'm trying to keep all XBL
testcases working, and having one broken interferes with my testing.
Keywords: dogfood,
regression
Whiteboard: [xbl1.0]
Assignee | ||
Comment 2•25 years ago
|
||
Assignee | ||
Comment 3•25 years ago
|
||
Looking for a review for the attached patch.
Status: NEW → ASSIGNED
Comment 4•25 years ago
|
||
The change looks good (although I have to admit I don't really understand what
the parentScope concept is all about...). r=attinasi
Assignee | ||
Comment 5•25 years ago
|
||
The basic idea is that the content tree can look like this...
A --> B
but the frame tree can look like this...
A --> (C) --> B
The check asks if the child frame (in this case C) was generated from A. If so,
you crawl into it. Basically I'm ensuring we crawl into anonymous frames that
have interposed themselves between the frames for parent and child content
nodes.
Comment 6•25 years ago
|
||
For dogfood rating: Please explain why this is a stopper bug, and for whom.
Thanks,
Jim
Whiteboard: [xbl1.0] → [xbl1.0][need info]
Assignee | ||
Comment 7•25 years ago
|
||
I have a set of 5 regression tests for XBL that are accessible from the debug
menu. Those tests should all be kept in working order. When one of them
breaks, it means XBL has regressed. That's why I have the tests in the first
place. :)
Assignee | ||
Comment 8•25 years ago
|
||
Fixed on trunk. Not critical for branch.
Status: ASSIGNED → RESOLVED
Closed: 25 years ago
Resolution: --- → FIXED
You need to log in
before you can comment on or make changes to this bug.
Description
•