This bug is filed due to the investigation of the disabled test, layout/reftests/svg/stroke-dasharray-02.svg. It appears that the only failure in the test is caused by the large scaling in [1]. To be clear, the phenomenon is that a <circle> with large scaling transform is not rendered at all, so the red circles (rendered by [2]) are not covered with lime, which cause the test failure. Here are a pair of simplified examples: ex1: <svg xmlns="http://www.w3.org/2000/svg"> <rect width="100%" height="100%" fill="lime"/> <circle cx="0" cy="0" r=".0008" fill="none" stroke="red" stroke-width=".0001" transform="translate(180, 300) scale(1e5, 1e5)"/> </svg> ex2: <svg xmlns="http://www.w3.org/2000/svg"> <rect width="100%" height="100%" fill="lime"/> <circle cx="200" cy="30" r="30" fill="none" stroke="brown" stroke-width="10" transform="translate(200, 300) scale(1e0, 1e0)"/> </svg> In stylo, ex1 can't be rendered at all, whereas ex2 can. In gecko, both examples can be rendered properly. [1] http://searchfox.org/mozilla-central/rev/2933592c4a01b634ab53315ce2d0e43fccb82181/layout/reftests/svg/stroke-dasharray-02.svg#39-40 [2] http://searchfox.org/mozilla-central/rev/2933592c4a01b634ab53315ce2d0e43fccb82181/layout/reftests/svg/stroke-dasharray-02.svg#37-38

Second thoughts, this might be related to precision issues of stroke-width, cause I tried to get the computed value of stroke-width for ex1 in stylo, and I got 0 instead of 0.0001.

Hmmm, this is definitely a parsing/computing issue of stroke-width property. Or, more like a issue for unitless length. In gecko, we keep the unitless length as it is, and compute/store it as a pure factor number in mStrokeWidth [1]. However, in stylo, we do some rounding stuff (round to absolute length) during parsing [2]. Maybe we should match stylo to gecko by adding the ability for length value to parse unitless number and compute/store as a factor value. Fix the bug summary then. [1] http://searchfox.org/mozilla-central/rev/2933592c4a01b634ab53315ce2d0e43fccb82181/layout/style/nsRuleNode.cpp#962 [2] http://searchfox.org/mozilla-central/rev/2933592c4a01b634ab53315ce2d0e43fccb82181/servo/components/style/values/specified/length.rs#622-629

There're 3 SVG properties that accept SVG unitless length, stroke-width, stroke-dasharray, and stroke-dashoffset [1]. In stylo, we compute and store SVG unitless length as a factor number for stroke-dasharray, but not for stroke-width and stroke-dashoffset. So, we shoud fix it. I wonder maybe we should add some more tests for them. Filed Bug 1367977. [1] http://searchfox.org/mozilla-central/search?q=symbol:M_8b66591b5a234643ab62b847a70746707a32d04b&redirect=false

The test is passed on my local, so ask for review while waiting for a positive try result.

try looks good: https://treeherder.mozilla.org/#/jobs?repo=try&revision=3da02825c514754a26f7bf1a733f974d4c60f048

Comment on attachment 8871220 [details] Bug 1367327 - stylo: compute and store SVG unitless length as a factor number. https://reviewboard.mozilla.org/r/142718/#review146816 ::: commit-message-2c8b6:3 (Diff revision 2) > +Bug 1367327 - stylo: compute and store SVG unitless length as a factor number. > + > +There 3 SVG properties that accept SVG unitless length, stroke-width, stroke-dasharray, There are ::: commit-message-2c8b6:5 (Diff revision 2) > +Bug 1367327 - stylo: compute and store SVG unitless length as a factor number. > + > +There 3 SVG properties that accept SVG unitless length, stroke-width, stroke-dasharray, > +and stroke-dashoffset. We compute and store SVG unitless length as a factor number for > +stroke-dasharray, but not for stroke-width and stroke-dashoffset. So, let's fix it. I don't think we need "So, let's fix it." in the commit message. :-)

Comment on attachment 8871598 [details] Bug 1367327 - stylo: re-enable SVG stroke-width related tetst. https://reviewboard.mozilla.org/r/143068/#review146818

Comment on attachment 8871220 [details] Bug 1367327 - stylo: compute and store SVG unitless length as a factor number. https://reviewboard.mozilla.org/r/142718/#review146816 > I don't think we need "So, let's fix it." in the commit message. :-) Yes, you're right. Thanks for the review. :)

Comments addressed!! Servo PR: https://github.com/servo/servo/pull/17053

Servo side merged: https://hg.mozilla.org/integration/autoland/rev/2e6408a2b2c7 Let's enable the test!!!

We're sorry, Autoland could not rebase your commits for you automatically. Please manually rebase your commits and try again. hg error in cmd: hg rebase -s ddd503c088e0 -d 2e6408a2b2c7: rebasing 398426:ddd503c088e0 "Bug 1367327 - stylo: re-enable SVG stroke-width related tetst. r=heycam" (tip) merging layout/reftests/svg/reftest.list warning: conflicts while merging layout/reftests/svg/reftest.list! (edit, then use 'hg resolve --mark') unresolved conflicts (see hg resolve, then hg rebase --continue)

Pushed by jichen@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/5f7aef97e7b7 stylo: re-enable SVG stroke-width related tetst. r=heycam