17784 2012-07-15 16:09:40 +0000 request: even/odd winding algorithm in addition to non-zero 2015-06-16 17:08:46 +0000 1 1 1 Unclassified HTML WG HTML5 spec unspecified PC Windows NT RESOLVED FIXED P1 normal --- 1 pomax dave.null cabanier mike plh public-html-admin public-html-wg-issue-tracking public-html-bugzilla oldest_to_newest 70018 0 pomax 2012-07-15 16:09:40 +0000 I'd like to request that the canvas 2d context fill style is extended with support for the even/odd winding algorithm, in addition to the non-zero winding algorithm, so that drawing shapes on the canvas that are based on paths obtained from graphics sources that encode shapes using the even/odd winding algorithm can be drawn without having to manually analyse the path for subpaths that need their direction flipped. Use cases are using image from .svg resource with the intention of using them as rasterised image data (not vector graphics) and shapes such as font outlines (for use as polygon graphics, not "letters") to the canvas. (As a bonus, this algorithm is already available for reuse in every current browser, since SVG requires algorithms both to be available, making implementing it for canvas 2d mostly a matter of hooking it up correctly) 70257 1 contributor 2012-07-18 07:13:34 +0000 This bug was cloned to create bug 17893 as part of operation convergence. 72520 2 cabanier 2012-08-21 22:41:22 +0000 This is indeed very useful and widely used. Moving to HTML.Next as it is a new feature request 81670 3 robin 2013-01-21 15:58:11 +0000 Mass move to "HTML WG" 81788 4 robin 2013-01-21 16:00:57 +0000 Mass move to "HTML WG" 121038 5 mike 2015-06-16 10:15:24 +0000 Making this a higher priority to actively seek more feedback on from implementers and webdevs. 121104 6 cabanier 2015-06-16 16:50:04 +0000 This was implemented by every browser 1-2 years ago. 121107 7 pomax 2015-06-16 17:08:46 +0000 indeed, this bug has since been resolved, but never got closed. http://jsbin.com/kemogesubu/edit?html,js,output demonstrates "evenodd" as fill rule working (I only tested it in firefox, chrome, and IE, but kind of assume that means evreyone supports it). You can change "evenodd" to "nonzero" to see it do a flood fill, or keep it "evenodd" to see it only fill the parts that uniquely cordon off a polygonal section.