Bug 13251

Summary: editing/style/fontsize-1.html fails intermittently when run multiple times on Intel, generates incorrect DOM
Product: WebKit Reporter: mitz
Component: HTML EditingAssignee: Nobody <webkit-unassigned>
Status: RESOLVED FIXED    
Severity: Normal Keywords: LayoutTestFailure
Priority: P2    
Version: 523.x (Safari 3)   
Hardware: Mac   
OS: OS X 10.4   
Attachments:
Description Flags
Don't generate 'size' attribute for relative font sizes, and ASSERT on wildly illegal conversions darin: review+

Description mitz 2007-04-01 07:41:52 PDT
On a debug build of TOT running on an Intel machine, "run-webkit-tests editing/style/fontsize-1.html editing/style/fontsize-1.html editing/style/fontsize-1.html editing/style/fontsize-1.html editing/style/fontsize-1.html" fails one or more instances of the test sometimes. Even when the test passes, when opened in Safari one can see that the DOM tree includes FONT elements with size="1" (or in the failing case, one of them has size="7"; neither size is correct, and the actual size is determined by an inline style attribute on a child SPAN).

Patch forthcoming.
Comment 1 mitz 2007-04-01 08:07:34 PDT
Created attachment 13914 [details]
Don't generate 'size' attribute for relative font sizes, and ASSERT on wildly illegal conversions
Comment 2 Darin Adler 2007-04-02 07:59:38 PDT
Comment on attachment 13914 [details]
Don't generate 'size' attribute for relative font sizes, and ASSERT on wildly illegal conversions

I'd prefer not to have ad-hoc comparisons of the unit type with values. All the < PX, > PC, <= CSS_DIMENSION is quite unclear and should be done with helper inline functions instead.

But for this fix, r=me
Comment 3 Mark Rowe (bdash) 2007-04-03 06:13:25 PDT
Landed in r20678.