[webkit-reviews] review denied: [Bug 112054] SVG Marker orientation==?UTF-8?Q?auto=20has=20incorrect=20behavior=20around=200=C2=B0=20?=: [Attachment 193052] Correct bisector calculation for markers (2)

[bugzilla-daemon at webkit.org]

Stephen Chenney <schenney at chromium.org> has denied Philip Rogers
<pdr at google.com>'s request for review:
Bug 112054: SVG Marker orientation=auto has incorrect behavior around 0°
https://bugs.webkit.org/show_bug.cgi?id=112054

Attachment 193052: Correct bisector calculation for markers (2)
https://bugs.webkit.org/attachment.cgi?id=193052&action=review

------- Additional Comments from Stephen Chenney <schenney at chromium.org>
I think the code change is right. I don't think the test has optimal coverage
because the angle between the path segments in each case ranges within a
relatively narrow band. The global orientation changes for any given marker,
but the relative orientation of the in and out segments does not change
anywhere near as much.

I think you need a test with a fixed in angle and all possible out angles (at 5
degree increments or something). I would probably try 8 sets, each with a
different input angle: 0.1 deg, -0.1 deg, 89.9, 90.1, 179.9, -179.9, -89.9,
-90.1 and the range of output angles.