Bug 190155 – [CG] Adopt CG SPI for non-even cornered rounded rects

• Component: Layout and Rendering
• Status: RESOLVED FIXED

[Said Abou-Hallawa]

Currently we create bezier curves for the non-even corners of the rounded rects. This should be optimized by the CG. With this SPI, I could get 3% progression on the Multiply test on MacPro.

[Said Abou-Hallawa]

Created attachment 351297 [details]
Patch

[EWS Watchlist]

Comment on attachment 351297 [details]
Patch

Attachment 351297 [details] did not pass ios-sim-ews (ios-simulator-wk2):
Output: https://webkit-queues.webkit.org/results/9426505

New failing tests:
fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-bottom-right.html
compositing/video/video-border-radius.html
fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-bottom-left.html
fast/box-shadow/box-shadow-with-zero-radius.html
fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-top-left.html
fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-top-right.html
fast/borders/border-radius-inner-clip-vertical.html
fast/borders/border-radius-inner-clip.html
css3/filters/backdrop/backdrop-filter-uneven-corner-radii.html

[EWS Watchlist]

Created attachment 351399 [details]
Archive of layout-test-results from ews124 for ios-simulator-wk2

The attached test failures were seen while running run-webkit-tests on the ios-sim-ews.
Bot: ews124  Port: ios-simulator-wk2  Platform: Mac OS X 10.13.6

[Said Abou-Hallawa]

Created attachment 352410 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 352410 [details]
Patch

Clearing flags on attachment: 352410

Committed r237163: <https://trac.webkit.org/changeset/237163>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Radar WebKit Bug Importer]

<rdar://problem/45291653>

[Ryan Haddad]

(In reply to Build Bot from comment #2)
> Comment on attachment 351297 [details]
> Patch
> 
> Attachment 351297 [details] did not pass ios-sim-ews (ios-simulator-wk2):
> Output: https://webkit-queues.webkit.org/results/9426505
> 
> New failing tests:
> fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-bottom-
> right.html
> compositing/video/video-border-radius.html
> fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-bottom-
> left.html
> fast/box-shadow/box-shadow-with-zero-radius.html
> fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-top-left.
> html
> fast/shapes/shape-outside-floats/shape-outside-floats-inset-rounded-top-
> right.html
> fast/borders/border-radius-inner-clip-vertical.html
> fast/borders/border-radius-inner-clip.html
> css3/filters/backdrop/backdrop-filter-uneven-corner-radii.html
These tests are indeed failing on the iOS Simulator bots now that this patch has landed:
https://build.webkit.org/results/Apple%20iOS%2012%20Simulator%20Release%20WK2%20(Tests)/r237183%20(395)/results.html

[Ryan Haddad]

Reverted r237163 for reason:

Introduced layout test failures on iOS Simulator.

Committed r237201: <https://trac.webkit.org/changeset/237201>

[Ryan Haddad]

It looks like there were similar reftest failures seen on Mojave:
https://build.webkit.org/results/Apple%20Mojave%20Release%20WK2%20(Tests)/r237168%20(274)/results.html

[Simon Fraser (smfr)]

With the change, we're rendering the radii with incorrect vertical flipping.

[Simon Fraser (smfr)]

Created attachment 352569 [details]
Testcase

[Said Abou-Hallawa]

(In reply to Simon Fraser (smfr) from comment #11)
> With the change, we're rendering the radii with incorrect vertical flipping.

The problem is I was passing the corners in the following orders:
    { topLeftRadius, topRightRadius, bottomRightRadius, bottomLeftRadius }

But they should be in the following order:
    { bottomLeftRadius, bottomRightRadius, topRightRadius, topLeftRadius }

This is to be consistent with CGRect whose origin is s located in the lower-left corner of the rectangle.

[Said Abou-Hallawa]

Created attachment 352645 [details]
Patch

[EWS Watchlist]

Comment on attachment 352645 [details]
Patch

Attachment 352645 [details] did not pass ios-sim-ews (ios-simulator-wk2):
Output: https://webkit-queues.webkit.org/results/9644184

New failing tests:
fast/box-shadow/box-shadow-with-zero-radius.html

[EWS Watchlist]

Created attachment 352670 [details]
Archive of layout-test-results from ews125 for ios-simulator-wk2

The attached test failures were seen while running run-webkit-tests on the ios-sim-ews.
Bot: ews125  Port: ios-simulator-wk2  Platform: Mac OS X 10.13.6

[EWS Watchlist]

Comment on attachment 352645 [details]
Patch

Attachment 352645 [details] did not pass ios-sim-ews (ios-simulator-wk2):
Output: https://webkit-queues.webkit.org/results/9645666

New failing tests:
fast/box-shadow/box-shadow-with-zero-radius.html

[EWS Watchlist]

Created attachment 352676 [details]
Archive of layout-test-results from ews126 for ios-simulator-wk2

The attached test failures were seen while running run-webkit-tests on the ios-sim-ews.
Bot: ews126  Port: ios-simulator-wk2  Platform: Mac OS X 10.13.6

[Said Abou-Hallawa]

Created attachment 352752 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 352752 [details]
Patch

Clearing flags on attachment: 352752

Committed r237280: <https://trac.webkit.org/changeset/237280>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

[Simon Fraser (smfr)]

This patch added some PLATFORM(IOS) that should be PLATFORM(IOS_FAMILY)

[Alexey Proskuryakov]

And with that fixed, the patch still breaks box-shadow-with-zero-radius.html (fix in bug 190813).

Probably roll back again?

[Ryan Haddad]

Reverted r237280 for reason:

Causes fast/box-shadow/box-shadow-with-zero-radius.html to fail on iOS Simulator.

Committed r237359: <https://trac.webkit.org/changeset/237359>

[Said Abou-Hallawa]

Created attachment 353317 [details]
Patch

[EWS Watchlist]

Comment on attachment 353317 [details]
Patch

Attachment 353317 [details] did not pass ios-sim-ews (ios-simulator-wk2):
Output: https://webkit-queues.webkit.org/results/9772127

New failing tests:
fast/box-shadow/box-shadow-with-zero-radius.html

[EWS Watchlist]

Created attachment 353328 [details]
Archive of layout-test-results from ews125 for ios-simulator-wk2

The attached test failures were seen while running run-webkit-tests on the ios-sim-ews.
Bot: ews125  Port: ios-simulator-wk2  Platform: Mac OS X 10.13.6

[Said Abou-Hallawa]

Created attachment 353565 [details]
Patch

[Simon Fraser (smfr)]

Comment on attachment 353565 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=353565&action=review

> LayoutTests/platform/ios/TestExpectations:1939
> +fast/box-shadow/box-shadow-with-zero-radius.html

You should mark this as [ImageOnlyFailure] so that the test still runs (to detect crashes).

Also the reason this only fails on iOS is probably because we don't use accelerated drawing for macOS testing.

[Said Abou-Hallawa]

Created attachment 353605 [details]
Patch

[WebKit Commit Bot]

Comment on attachment 353605 [details]
Patch

Clearing flags on attachment: 353605

Committed r237691: <https://trac.webkit.org/changeset/237691>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.