Bug 55514 - [chromium] invalidating a rotated border causes different pixels to be painted
Summary: [chromium] invalidating a rotated border causes different pixels to be painted
Status: RESOLVED WONTFIX
Alias: None
Product: WebKit
Classification: Unclassified
Component: WebCore Misc. (show other bugs)
Version: 528+ (Nightly build)
Hardware: PC OS X 10.5
: P2 Normal
Assignee: Nobody
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-03-01 15:01 PST by Tony Chang
Modified: 2013-04-11 13:21 PDT (History)
3 users (show)

See Also:


Attachments
test case (255 bytes, text/html)
2011-03-01 15:01 PST, Tony Chang
no flags Details
better test case (342 bytes, text/html)
2011-03-01 15:54 PST, Tony Chang
no flags Details
initial load (22.04 KB, image/png)
2011-03-01 15:55 PST, Tony Chang
no flags Details
after invalidate (21.79 KB, image/png)
2011-03-01 15:56 PST, Tony Chang
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Tony Chang 2011-03-01 15:01:27 PST
Repro steps:
1) Load the attached file in test_shell.exe.  You should see a box rotated 20deg.
2) Drag the window off screen and back.
3) If you look carefully, you'll see that the edge is now a tiny bit different (a bit darker?).

If you reload, the edges return to the lighter state.

I suspect this is a problem in skia since I couldn't repro on mac.  It's also hard to repro on Chromium because it saves a backing store of the image.  It's the extra invalidate that causes the border to change.
Comment 1 Tony Chang 2011-03-01 15:01:47 PST
Created attachment 84303 [details]
test case
Comment 2 James Robinson 2011-03-01 15:15:24 PST
Can you post before/after screenshots of what you are seeing?
Comment 3 Tony Chang 2011-03-01 15:54:57 PST
Created attachment 84314 [details]
better test case
Comment 4 Tony Chang 2011-03-01 15:55:31 PST
Created attachment 84315 [details]
initial load
Comment 5 Tony Chang 2011-03-01 15:56:04 PST
Created attachment 84316 [details]
after invalidate
Comment 6 Tony Chang 2011-03-01 15:56:42 PST
(In reply to comment #3)
> Created an attachment (id=84314) [details]
> better test case

This shows a much more noticeable difference.  The original test case only differed by a dozen pixels or so.