Bug 9652

Summary: REGRESSION: Wrong RTL table layout after changing an RTL table cell to display:none
Product: WebKit Reporter: mitz
Component: TablesAssignee: Nobody <webkit-unassigned>
Status: RESOLVED FIXED    
Severity: Normal Keywords: Regression
Priority: P1    
Version: 420+   
Hardware: Mac   
OS: OS X 10.4   
Attachments:
Description Flags
Test case (will assert)
none
Proposed patch darin: review+

mitz
Reported 2006-06-29 09:08:59 PDT
The attached testcase causes an assertion failure (invalid index when accessing the columnPos vector in layoutRows().
Attachments
Test case (will assert) (389 bytes, text/html)
2006-06-29 09:09 PDT, mitz 		no flags
Details
Proposed patch (34.07 KB, patch)
2006-06-29 09:36 PDT, mitz 		darin: review+
DetailsFormatted DiffDiff
View All Add attachment proposed patch, testcase, etc.
mitz
Comment 1 2006-06-29 09:09:47 PDT
Created attachment 9091 [details] Test case (will assert)
mitz
Comment 2 2006-06-29 09:24:51 PDT
I have a test case that shows that this is in fact a regression from shipping WebKit, and that it affects layout. I'm going to include that case in the patch. 
mitz
Comment 3 2006-06-29 09:36:15 PDT
Created attachment 9094 [details] Proposed patch Includes change log and layout test. Passes all layout tests.
Darin Adler
Comment 4 2006-06-29 09:39:49 PDT
Comment on attachment 9094 [details] Proposed patch r=me
Dave Hyatt
Comment 5 2006-06-29 12:08:25 PDT
nEffCols is definitely cleaner and more readable, but I don't think the old code was wrong either... unless I'm misunderstanding it, you should have gotten the same results, and it sounds like the Row* may not be properly up to date.
David Kilzer (:ddkilzer)
Comment 6 2006-06-29 21:34:04 PDT
Committed revision 15101.
Note You need to log in before you can comment on or make changes to this bug.