Bug 84245 – Artifacts during scrolling with changing selection

• Component: Layout and Rendering
• Status: UNCONFIRMED

[Alexei Bykov]

To reproduce this bug you need to select text and scroll page in parallel. First you select text by moving the mouse cursor down. Then you need to move the mouse cursor up. This action causes deselecting of text and repainting of deselected text. If you open an attachement you see that webkit repaints deselected texts with some artifacts. I have managed to reproduced this bug on different computers under Windows 7. Versions of Webkit: Apple Windows port and Cairo Windows port, revisions: r106194 and r112503, application - WinLauncher. Probably the bug is reproduced on other revisions, I didn't test other revisions.

P.S. Sorry for my poor english. I hope that you manage to understand me.

[Alexei Bykov]

Created attachment 137698 [details]
artifacts

This is a screenshot of test application written by me. The same bug is reproduced in WinLauncher

[Alexey Proskuryakov]

Are you seeing this in Safari, too? Is it a regression from Safari 5.1.5?

Also, does this happen with all web sites, or just this example?

[Alexei Bykov]

(In reply to comment #2)
> Are you seeing this in Safari, too? Is it a regression from Safari 5.1.5?
> 

No, I don't see this in Safari (5.1.2 & 5.1.4). I see it only in WinLauncher and in my application.

> Also, does this happen with all web sites, or just this example?

Yes, it happens with all web sites.

[Alexey Proskuryakov]

One more question: did you try Safari nightly builds from <http://nightly.webkit.org>?

[Alexei Bykov]

(In reply to comment #4)
> One more question: did you try Safari nightly builds from <http://nightly.webkit.org>?

Yes, I tried to use binaries from nightly.webkit.org. I watched the same behaviour: nice rendering in Safari (Webkit.exe) and artifacts in WinLauncher.

[Alexei Bykov]

Created attachment 139193 [details]
patch

I found a solution. This solution is ugly. But I think that it may help to write a more correct code. The reason of the bug is incorrect calculating height of a dirty region during scrolling.

[Alexey Proskuryakov]

Interesting!

I'm not qualified to judge whether the approach here is right, hopefully Simon or Mitz will weigh in.

Either way, the patch has a number of stylistic issues making it impossible to land as is, lacks tests and ChangeLog. Would you be interested in working on it further to make it appropriate for WebKit? Please see <http://www.webkit.org/coding/contributing.html> for details.

[Alexei Bykov]

(In reply to comment #7)
> Interesting!
> 
> I'm not qualified to judge whether the approach here is right, hopefully Simon or Mitz will weigh in.
> 
> Either way, the patch has a number of stylistic issues making it impossible to land as is, lacks tests and ChangeLog. Would you be interested in working on it further to make it appropriate for WebKit? Please see <http://www.webkit.org/coding/contributing.html> for details.

>the patch has a number of stylistic issues making it impossible to land as is, lacks tests and ChangeLog
I understand this. I just wanted to show the cause of the bug and possible ways of fixing it.

>Would you be interested in working on it further to make it appropriate for WebKit?
Yes, I am intersted in working on it. In any case, I will have to explore the source code of Webkit in the near future, because we plan to use Webkit in our project. Now we have 2 problems with Webkit: bug which is described on this page and numerous memory leaks.