Bug 15157

Summary: REGRESSION: Image defined in background-position: top center gets unexpectedly truncated
Product: WebKit Reporter: Gérard Talbot (no longer involved) <browserbugs2>
Component: Layout and RenderingAssignee: Nobody <webkit-unassigned>
Status: RESOLVED FIXED    
Severity: Normal CC: mitz, mrowe
Priority: P1 Keywords: HasReduction, InRadar, Regression
Version: 523.x (Safari 3)   
Hardware: All   
OS: All   
URL: http://www.hixie.ch/tests/adhoc/css/background/12.html
Attachments:
Description Flags
Account for the case where the body's background doesn't propagate to the root mjs: review+

Gérard Talbot (no longer involved)
Reported 2007-09-07 13:55:59 PDT
Steps to reproduce: ------------------- 1. Load provided URL Actual results -------------- The 2 cats "smallcats" image inside the green-bordered box are truncated. We can see only the top half of the 2 cats image. We can see only the top half of the 2 cats image. Expected results ---------------- The 2 cats "smallcats" image inside the green-bordered box should be rendered entirely, without truncation. We should see the whole 2 cats "smallcats" image. Nota bene: - the top of the ears of the top-most 2 cats image in the blue-bordered box touching the blue top-border is a consequence of bug 10794. - There is no truncation with other browsers like Firefox 2.0.0.6, Opera 9.23 build 8808, Opera 9.50a1 build 9500, Seamonkey 2.0a1pre rv:1.9a8pre build 2007090402 and MSIE 7. I also get expected results with Swift 0.2 which is using Safari 2.0.4; this is also confirmed when using http://www.browsrcamp.com/ - I can upload a screenshot if required, requested - I'm using Safari 3.0.3 build 522.15.5 under XP Pro
Attachments
Account for the case where the body's background doesn't propagate to the root (53.04 KB, patch)
2007-09-08 05:39 PDT, mitz 		mjs: review+
DetailsFormatted DiffDiff
Show Obsolete (1) View All Add attachment proposed patch, testcase, etc.
Gérard Talbot (no longer involved)
Comment 1 2007-09-07 14:03:59 PDT
I searched for a duplicate and did not find any.
Gérard Talbot (no longer involved)
Comment 2 2007-09-07 14:35:04 PDT
Screenshot: http://browsershots.org/http://www.hixie.ch/tests/adhoc/css/background/12.html
Dave Hyatt
Comment 3 2007-09-07 15:55:04 PDT
I don't see this in Safari 3 on Tiger. This might be Windows-specific.
Mark Rowe (bdash)
Comment 4 2007-09-07 22:06:07 PDT
I reproduced this in TOT on 10.5. It appears to be an incremental painting issue. Resizing the browser window causes the full image to be displayed. After it has occurred once I needed to empty the browser cache to trigger it again.
Mark Rowe (bdash)
Comment 5 2007-09-07 22:11:56 PDT
<rdar://problem/5469095>
Dave Hyatt
Comment 6 2007-09-07 22:44:44 PDT
This a regression? I think it might be.
mitz
Comment 7 2007-09-08 05:09:23 PDT
(In reply to comment #6) > This a regression? I think it might be. Probably regressed with the patch that refined background image repaints. I simply ignored the fact that when both <html> and <body> specify backgrounds, the body's doesn't propagate to the root.
mitz
Comment 8 2007-09-08 05:39:27 PDT
Created attachment 16226 [details] Account for the case where the body's background doesn't propagate to the root
David Kilzer (:ddkilzer)
Comment 9 2007-09-08 09:10:06 PDT
Yes, this is a regression from Safari 2.0.4 (419.3) with original WebKit on Mac OS X 10.4.10 (8R218).
David Kilzer (:ddkilzer)
Comment 10 2007-09-08 09:11:52 PDT
(In reply to comment #9) > Yes, this is a regression from Safari 2.0.4 (419.3) with original WebKit on Mac > OS X 10.4.10 (8R218). Tested with a local debug build of WebKit r25410 with Safari 3 Public Beta v. 3.0.3 (522.12.1) on Mac OS X 10.4.10 (8R218).
Maciej Stachowiak
Comment 11 2007-09-10 19:47:21 PDT
Comment on attachment 16226 [details] Account for the case where the body's background doesn't propagate to the root r=me
Mark Rowe (bdash)
Comment 12 2007-09-10 22:27:44 PDT
Landed in r25487.
Eric Seidel (no email)
Comment 13 2007-10-02 11:43:48 PDT
The pixel tests seem to suggest that fast/repaint/body-background-image.html may have regressed on the newly merged feature branch. The apple is no longer "highlighted" by a lighter square in the "actual" results. The entire page is instead a dim grey. I don't really know how repaint tests work though. Please re-close this bug (after updating the pixel test results) if I have simply misunderstood the test. I don't see any expected results checked into either feature-branch or trunk however. Odd.
mitz
Comment 14 2007-10-02 13:03:02 PDT
(In reply to comment #13) > I don't see any expected results checked into either feature-branch or trunk > however. Odd. <http://trac.webkit.org/projects/webkit/browser/trunk/LayoutTests/platform/mac/fast/repaint/body-background-image-expected.png?rev=25487>
Mark Rowe (bdash)
Comment 15 2007-10-14 04:55:10 PDT
This is reopened, but the comments don't really seem sure about why it was reopened. Mitz, can you verify that your patch did in fact fix it and that it's passing on trunk at present?
Mark Rowe (bdash)
Comment 16 2007-10-14 07:06:36 PDT
Looks fine to me with TOT.
Note You need to log in before you can comment on or make changes to this bug.