Bug 246026 – REGRESSION (255063@main ): [ Monterey ] fast/forms/button-sizes.html is a consistent failure

• Component: New Bugs
• Status: RESOLVED CONFIGURATION CHANGED

[Karl Rackler]

Description:
fast/forms/button-sizes.html

The first failure I saw on the dashboard was on at 255063@main.

Reproduction steps:
I can reproduce this on 255063@main but cannot reproduce it on 255062@main or earlier.

Command: 
run-webkit-tests --exit-after-n-failures 1 --exit-after-n-crashes-or-timeouts 1 --iterations 50 -f --no-retry fast/forms/button-sizes.html

Result: 
Regressions: Unexpected text-only failures (1)
  fast/forms/button-sizes.html [ Failure ]

History:
https://results.webkit.org/?suite=layout-tests&test=fast%2Fforms%2Fbutton-sizes.html&version_name=Monterey

Diff:
--- /Volumes/Data/worker/Apple-Monterey-Release-WK2-Tests/build/layout-test-results/fast/forms/button-sizes-expected.txt
+++ /Volumes/Data/worker/Apple-Monterey-Release-WK2-Tests/build/layout-test-results/fast/forms/button-sizes-actual.txt
@@ -103,8 +103,8 @@
           RenderText {#text} at (0,0) size 100x23
             text run at (0,0) width 100: "Test Button"
       RenderText {#text} at (0,0) size 0x0
-      RenderButton {BUTTON} at (0,49) size 121x29 [color=#000000D8] [bgcolor=#C0C0C0] [border: none (2px outset #C0C0C0) none (2px outset #C0C0C0)]
-        RenderBlock (anonymous) at (8,2) size 105x24
-          RenderText {#text} at (0,0) size 105x24
-            text run at (0,0) width 105: "Test Button"
+      RenderButton {BUTTON} at (0,49) size 120x29 [color=#000000D8] [bgcolor=#C0C0C0] [border: none (2px outset #C0C0C0) none (2px outset #C0C0C0)]
+        RenderBlock (anonymous) at (8,2) size 104x24
+          RenderText {#text} at (0,0) size 104x24
+            text run at (0,0) width 104: "Test Button"
       RenderText {#text} at (0,0) size 0x0

[Radar WebKit Bug Importer]

<rdar://problem/100767507>

[Alexey Proskuryakov]

This probably just needs a new baseline, this PR changed many already.

Weird that a refactoring changes test results though - and some of them in much bigger ways.

[alan]

(In reply to Alexey Proskuryakov from comment #2)
> This probably just needs a new baseline, this PR changed many already.
> 
> Weird that a refactoring changes test results though - and some of them in
> much bigger ways.
255063@main is not a refactoring change. It diverts certain type of content from legacy to modern line layout (meaning that the content is not laid out by legacy anymore). 1px diffs (and sometimes even bigger) are expected due to different rounding techniques, choice of storage type (float vs LayoutUnit), progressions etc.

[Karl Rackler]

Pull request: https://github.com/WebKit/WebKit/pull/4993

[EWS]

Test gardening commit 255141@main (d6f6606ca8a8): <https://commits.webkit.org/255141@main>

Reviewed commits have been landed. Closing PR #4993 and removing active labels.

[Karl Rackler]

Verified landing 255141@main Monterey re-baseline resolved the issue.