Bug 12630

Summary: <use> cursor not changing (to hand) on hover
Product: WebKit Reporter: jay <jay>
Component: SVGAssignee: Nobody <webkit-unassigned>
Status: RESOLVED FIXED    
Severity: Normal CC: zimmermann
Priority: P2    
Version: 420+   
Hardware: Mac   
OS: OS X 10.4   
URL: http://www.peepo.co.uk
Attachments:
Description Flags
look: No Hands
none
A much nicer test case
none
fix and manual test lars.knoll: review+

Description jay 2007-02-06 01:34:41 PST
cursor should cahnge to hand on hover over a link in default case.
Comment 1 jay 2007-02-06 01:41:34 PST
Created attachment 12962 [details]
look: No Hands
Comment 2 Eric Seidel (no email) 2007-02-06 05:00:29 PST
Created attachment 12967 [details]
A much nicer test case
Comment 3 Eric Seidel (no email) 2007-02-06 05:02:07 PST
A few things were wrong with the original test case:

1.  it was uploaded as image/svg-xml.  That's incorrect.  The final SVG 1.1 spec uses image/svg+xml.
2.  it used fill='eee', that worked OK in Safari (and would have in IE) but it isn't valid by the SVG spec.  fill='#eee' would have been correct.
Comment 4 jay 2007-02-06 05:28:27 PST
bugzilla autodetect for mozilla and safari uploads as image/svg-xml
Comment 5 Eric Seidel (no email) 2007-02-06 05:40:10 PST
The patch attached to bug 12580 is a step in the right direction.  However it's still not quite fixed.  This shouldn't be hard to fix, but I should probably speak with WildFox first.
Comment 6 Eric Seidel (no email) 2007-02-06 05:40:59 PST
That's odd.  When I upload from Safari, bugzilla correctly autodetects image/svg+xml
Comment 7 jay 2007-02-06 07:17:37 PST
likely Opera...
Comment 8 David Kilzer (:ddkilzer) 2007-02-06 08:14:41 PST
(In reply to comment #7)
> likely Opera...

Usually the operating system plays a role in file types as well.  In Windows, it's the registry.  In Linux, it's /etc/mime.types.

Comment 9 Eric Seidel (no email) 2007-06-01 03:57:55 PDT
Created attachment 14824 [details]
fix and manual test

I'm pretty certain this is the correct fix.  The only thing which could possibly be wrong is if Layer::hitTest has some caller which wouldn't want to know the URLElement ignoring shadow nodes.  Such a caller seems highly unlikely since URLElement is only used by link hover code AFAIK.
Comment 10 Eric Seidel (no email) 2007-06-01 04:31:14 PDT
Landed as r21937 on the feature-branch.  Yet another change which probably should move to TOT.