The test at canvas/philip/tests/2d.gradient.interpolate.colouralpha.html fails on macOS Sierra.
Created attachment 285636 [details]
Comment on attachment 285636 [details]
View in context: https://bugs.webkit.org/attachment.cgi?id=285636&action=review
> + Test 2d.gradient.interpolate.colouralpha.html is broken on macOS Sierra
There is another instance of this test in LayoutTests/imported/w3c/canvas/2d.gradient.interpolate.colouralpha.html.
Does that one also need the same fix? If it does, we presumably need to upstream the fix.
Created attachment 285763 [details]
Comment on attachment 285763 [details]
View in context: https://bugs.webkit.org/attachment.cgi?id=285763&action=review
> + When the test samples 25th pixel it expect the value to be exactly 25% interpolation of the color values.
> + However, the sampling location is the center of the pixel, i.e. it should correspond to 25.5% interpolation,
> + which would expect the color of (189.975, 189.75, 65.25, 65.25). The new values are a rounded values of
> + interpolation âby handâ at steps of 25.5%, 50.5% and 75.5%.
Can you describe what actually changed here then? And if this is only Sierra, won't this change cause failures on El Cap and Yosemite?
> + * canvas/philip/tests/2d.gradient.interpolate.colouralpha.html:
I wonder if we can remove all these tests if they are duplicated in imported/w3c/canvas
Created attachment 285824 [details]
Patch for landing
Comment on attachment 285824 [details]
Patch for landing
Clearing flags on attachment: 285824
Committed r204371: <http://trac.webkit.org/changeset/204371>
All reviewed patches have been landed. Closing bug.
Upstreaming the fix to the W3C web-platform-tests repository, see https://github.com/w3c/web-platform-tests/pull/3458.