Bug 6157 – WebCore+SVG has some problems with rounded rectangles

• Component: SVG
• Status: RESOLVED FIXED

[David Astling]

Asymmetrical rounded rectangles, where rx != ry, do not seem to be supported in the latest WebKit 
build (Tue Dec 20 4:20 GMT 2005).  Also if either rx or ry attribute is omitted the rectangle is not 
rounded.  According to the SVG schema the missing attribute is assumed to be equal to the other.

For example , the larger green rectangle in the attached SVG file is missing the rx attribute, but is not 
rounded in WebKit.  Also the smaller blue rectangle should be asymmetrical, but is symmetrical in 
WebKit.

------------------------------------------------------

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<svg
   xmlns:svg="http://www.w3.org/2000/svg"
   xmlns="http://www.w3.org/2000/svg"
   version="1.0"
   width="500"
   height="500">
  <rect
     width="400"
     height="400"
     ry="50"
     x="50"
     y="50"
     style="fill:#00ff00;fill-opacity:1;stroke:none"
     id="rectWithNoRx" />
  <rect
     width="200"
     height="300"
     rx="35"
     ry="150"
     x="150"
     y="100"
     style="fill:#0000ff;fill-opacity:1;stroke:none"
     id="rectAsym" />
</svg>

[David Astling]

Created attachment 5170 [details]
attached SVG file of rounded rectangles

[David Astling]

Created attachment 5171 [details]
attached PNG file of rounded rectangles

Compare the correctly rendered rectangles in this PNG file with the rectangles
in the SVG file

[Joost de Valk (AlthA)]

Confirmed, behavior is different than in Firefox.

[Eric Seidel (no email)]

Yeah, the rounded rect code is... um... less than desireable.  We noticed this during SVG code review, and 
will fix.  Perhaps Julien would like to take a stab at this easy bug.  The "quesitonable code" is in 
KCanvasCreator (which is slated to die anyway -- the code to be moved into the device probably, or at 
least KCanvasCreator renamed to PathCreator).

[Rob Buis]

Created attachment 5223 [details]
Potential fix

I am a bit out of ksvg2 development this week, however I created
this simple patch, since there is clear typo in the original.
Hopefully someone can test and see whether it helps fixing the problem.
Cheers,

Rob.

[Eric Seidel (no email)]

Comment on attachment 5223 [details]
Potential fix

Looks good.  r=me.

[Geoffrey Garen]

Are the attached files supposed to land as layout tests?