Both of these attributes are defined as #IMPLIED in the DTD. Webkit+svg fails to define them correctly if missing. An example is here : http://www.croczilla.com/svg/samples/tiger/tiger.svg Just add width="800" or anything else in the <svg> element and the problem disappears. You can also check with drawTest. The rendering is done, but not on the white background.
Created attachment 4335 [details] reduced test case
When width or height are not specified in the <svg> element, they are implied as 100% (according to the spec) but the render fails, 100% is not mapped to the actual window size Same thing for an element like <rect> with width or height set with percentage. The element will not render. This bug is specific to OSX.
Created attachment 4446 [details] Temporary patch A simple patch to handle the case where only width or height is not specified in the <svg> element.
This is still a bug in TOT.
Looks OK, but might be because we're not clipping the <svg>: http://bugzilla.opendarwin.org/show_bug.cgi?id=6093
*** Bug 6599 has been marked as a duplicate of this bug. ***
Created attachment 5830 [details] Testcase for percentage sized rect
Comment on attachment 4446 [details] Temporary patch This code is no longer usable... Submitter: please set a review flag next time.
Increasing to P2, SVG without support for % for scaling .... well you get it :)
Changing subject, rounded corners don't work either, testcase forthcoming.
Created attachment 5833 [details] Testcase for percentage rounded rectangle
Created attachment 5854 [details] patch fixing issue
Comment on attachment 5854 [details] patch fixing issue I'm not sure this fix is really complete (What about % used for text or images, or <svg>?) I also think the comment could be cleaner. I can fix the comments on landing since they're trivial issues, and we'll use other bugs to track the missing support.
Comment on attachment 5854 [details] patch fixing issue Actually looking at this further, this does not fix the issue. I'm posting a patch which contains slightly modified comments, as well as a test case which demonstrates how this patch is incorrrect.
Created attachment 5857 [details] Patch with corrected comments, and test case demonstrating incorrect fix.
Created attachment 5858 [details] Simple test case
Created attachment 5959 [details] latest patch with code removal and corrected comments
Created attachment 5960 [details] svn-create-patch version of previous patch for WebCore
Created attachment 5961 [details] layouttests patch
Comment on attachment 5960 [details] svn-create-patch version of previous patch for WebCore Looks great! r=me.
Comment on attachment 5961 [details] layouttests patch I'll tweak this a bit when I land, but looks good. Thanks again. r=me.
Thanks again Alexander.