Bug 201714 – REGRESSION? (r249786): inspector/canvas/context-attributes.html is failing on macOS WK1

• Component: Web Inspector
• Status: RESOLVED FIXED

[Ryan Haddad]

inspector/canvas/context-attributes.html is failing on macOS WK1 after r249786 with the following diff:

--- /Volumes/Data/slave/mojave-release-tests-wk1/build/layout-test-results/inspector/canvas/context-attributes-expected.txt
+++ /Volumes/Data/slave/mojave-release-tests-wk1/build/layout-test-results/inspector/canvas/context-attributes-actual.txt
@@ -32,7 +32,7 @@
   "antialias": true,
   "premultipliedAlpha": true,
   "preserveDrawingBuffer": false,
-  "powerPreference": "default",
+  "powerPreference": "low-power",
   "failIfMajorPerformanceCaveat": false
 }
 
@@ -46,7 +46,7 @@
   "antialias": true,
   "premultipliedAlpha": true,
   "preserveDrawingBuffer": false,
-  "powerPreference": "default",
+  "powerPreference": "low-power",
   "failIfMajorPerformanceCaveat": false
 }
 PASS: Canvas context should have attribute "alpha" with value "false".

https://webkit-test-results.webkit.org/dashboards/flakiness_dashboard.html#showAllRuns=true&tests=inspector%2Fcanvas%2Fcontext-attributes.html

[Radar WebKit Bug Importer]

<rdar://problem/55290473>

[Ryan Haddad]

Devin, should the test be rebaselined for WK1, or is this unexpected?

[Joseph Pecoraro]

(In reply to Ryan Haddad from comment #2)
> Devin, should the test be rebaselined for WK1, or is this unexpected?

WK1 (or perhaps specifically Mac WK1) will apparently need its own results since WebKitLegacy forces low-power:

    WebPreferences.mm:
    @YES, WebKitForceWebGLUsesLowPowerPreferenceKey,

WK2 does not force low-power and uses the default WebCore::Setting values.

[Devin Rousso]

(In reply to Ryan Haddad from comment #2)
> Devin, should the test be rebaselined for WK1, or is this unexpected?
No, this is unexpected, but not wrong.  Apparently WK1 forces any WebGL/WebGL2 <canvas> to be "low-power", regardless of whatever value is asked for.

I'll change the test to just filter out the `powerPreference` so long as it's one of the known possible values.

[Devin Rousso]

Created attachment 378638 [details]
Patch

[Joseph Pecoraro]

Comment on attachment 378638 [details]
Patch

I liked seeing default. We could check in WK1 results. But this is fine

[WebKit Commit Bot]

Comment on attachment 378638 [details]
Patch

Clearing flags on attachment: 378638

Committed r249809: <https://trac.webkit.org/changeset/249809>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.