Bug 90574 - REGRESSION (r98542?): writing-mode-dependent properties applied incorrectly, apparently due to bad caching
Summary: REGRESSION (r98542?): writing-mode-dependent properties applied incorrectly, ...
Status: NEW
Alias: None
Product: WebKit
Classification: Unclassified
Component: CSS (show other bugs)
Version: 528+ (Nightly build)
Hardware: Unspecified Unspecified
: P1 Major
Assignee: Nobody
URL:
Keywords: InRadar
Depends on:
Blocks:
 
Reported: 2012-07-04 12:47 PDT by mitz
Modified: 2012-07-09 10:28 PDT (History)
3 users (show)

See Also:


Attachments
Test case (324 bytes, text/html)
2012-07-04 12:47 PDT, mitz
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description mitz 2012-07-04 12:47:01 PDT
Created attachment 150833 [details]
Test case

To reproduce, open the test case, then open the Web Inspector in DOM tree mode, then close the Web Inspector, then reload the page (there may be other ways to reproduce the bug). After doing so, even after repeated reloading of the page, the second <div class="test"> gets both a width and a height of 100px, and gets borders on all four sides. Apparently findFromMatchedPropertiesCache() decides to copy from the style of the first <div class="test"> into the style of the second one, despite their writing mode difference.
Comment 1 Radar WebKit Bug Importer 2012-07-04 12:47:55 PDT
<rdar://problem/11806873>
Comment 2 mitz 2012-07-04 12:55:35 PDT
Looks like this was caused by <http://trac.webkit.org/r98542>.
Comment 3 Antti Koivisto 2012-07-09 05:56:40 PDT
The test case is confusing. What does the correct rendering look like?

Inspecting the test changes (corrects?) the rendering but I don't see anything persisting over reloads. The cache is per-document so it shouldn't be possible.
Comment 4 mitz 2012-07-09 10:28:03 PDT
(In reply to comment #3)
> The test case is confusing. What does the correct rendering look like?

Black upside-down L, no blue.

> Inspecting the test changes (corrects?) the rendering but I don't see anything persisting over reloads. The cache is per-document so it shouldn't be possible.

That’s a good start :-)