Bug 50373 – [Qt] Sub-Frame content is not updated when scrolling in certain circumstances

• Status: RESOLVED
• Resolution: FIXED
• Component: Layout and Rendering

[Thomas Thrainer]

Created attachment 75355 [details]
Patch wich fixes the problem, but very inefficiently

Steps to reproduce:
 * Start QtTestBrowser
 * Enable QGraphicsWebView
 * Enable Accelerated Compositing (not sure if that's needed)
 * Enable Resizes to Contents Mode (not sure if that's needed)
 * Enable Tiled Backing Store (needed)
 * go to images.google.com
 * Search for asdf
 * Click on one of the result images, make sure that the browser window is quite small
 * Google will display an iFrame in the lower part of the screen
 * If you try to scroll the iFrame content, only the scroll bar is updated, but not the content of the frame

I attach a patch which fixes the bug, but which obviously is very inefficient.

[Thomas Thrainer]

The previous patch did not work in all cases, so I'll attach the new one. It's also more efficient.

[Thomas Thrainer]

Created attachment 75911 [details]
New patch which invalidates the backing store more often and more efficiently

[Thomas Thrainer]

BTW, I'm using QtWebkit 2.1 week 47 (plus a patch from week 48).

[Thomas Thrainer]

Created attachment 75916 [details]
New patch, this time avoiding to invalidate the backing store too often.

The previous patch was somewhat premature and did invalidate the backing store way too often.
Fixed in this version.

BTW: I do know that I instead should enable frame flattening in order to work around this bug, but as frame flattening has its own problems (#50391), I need this to work too.

[Benjamin Poulain]

Please follow this to report bugs on QtWebKit: http://trac.webkit.org/wiki/QtWebKitBugs
Otherwise bugs are not seen by us.


Please follow http://trac.webkit.org/wiki/QtWebKitContrib to contribute patches.
If you do not use the review flag, the reviewer will not see you patches.

[WebKit Review Bot]

Attachment 75916 [details] did not pass style-queue:

Failed to run "['Tools/Scripts/check-webkit-style', '--diff-files']" exit_code: 1

Total errors found: 0 in 0 files


If any of these errors are false positives, please file a bug against check-webkit-style.

[Andreas Kling]

Comment on attachment 75916 [details]
New patch, this time avoiding to invalidate the backing store too often.

Needs a ChangeLog entry.

Please see http://trac.webkit.org/wiki/QtWebKitContrib for more information on how to contribute patches to QtWebKit.

[Diego Gonzalez]

Still reproducible with r79447

[Alexis Menard (darktears)]

Created attachment 89025 [details]
Version with changelog and rebased.

[Kenneth Rohde Christiansen]

Doesn't it happen for div's with overflow as well?

[WebKit Commit Bot]

The commit-queue encountered the following flaky tests while processing attachment 89025 [details]:

animations/play-state-suspend.html bug 50959 (author: [e-mail])
The commit-queue is continuing to process your patch.

[WebKit Commit Bot]

Comment on attachment 89025 [details]
Version with changelog and rebased.

Clearing flags on attachment: 89025

Committed r83473: <http://trac.webkit.org/changeset/83473>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Antonio Gomes]

No mention to Thomas Thrainer work? =/ in your changelog

[Antonio Gomes]

(In reply to comment #10)
> Doesn't it happen for div's with overflow as well?

afaik, div scrolling goes through invalidateContentsAndWindow codepath.