Bug 280424 – [ Sequoia ] fast/table/col-and-colgroup-offsets.html is a constant failure.

• Status: RESOLVED
• Resolution: FIXED
• Component: New Bugs

[Anfernee Viduya]

fast/table/col-and-colgroup-offsets.html
is a constant failure. 

HISTORY:
https://results.webkit.org/?suite=layout-tests&test=fast%2Ftable%2Fcol-and-colgroup-offsets.html&platform=mac 

DIFF: 
 PASS document.getElementById("columnThatShouldNotBeRendered").offsetWidth is 0
 
 Tests for offsetHeight:
-FAIL document.getElementById("productNo").parentNode.offsetHeight should be 111. Was 119.
-FAIL document.getElementById("productNo").offsetHeight should be 111. Was 119.
-FAIL document.getElementById("productName").offsetHeight should be 111. Was 119.
-FAIL document.getElementById("hasMAndHasNAndHasO").offsetHeight should be 111. Was 119.
-FAIL document.getElementById("hasMAndHasN").offsetHeight should be 111. Was 119.
-FAIL document.getElementById("hasO").offsetHeight should be 111. Was 119.
-FAIL document.getElementById("hasP").offsetHeight should be 111. Was 119.
+FAIL document.getElementById("productNo").parentNode.offsetHeight should be 108. Was 116.
+FAIL document.getElementById("productNo").offsetHeight should be 108. Was 116.
+FAIL document.getElementById("productName").offsetHeight should be 108. Was 116.
+FAIL document.getElementById("hasMAndHasNAndHasO").offsetHeight should be 108. Was 116.
+FAIL document.getElementById("hasMAndHasN").offsetHeight should be 108. Was 116.
+FAIL document.getElementById("hasO").offsetHeight should be 108. Was 116.
+FAIL document.getElementById("hasP").offsetHeight should be 108. Was 116.
 PASS document.getElementById("columnThatShouldNotBeRendered").offsetHeight is 0
 
 When borderCollapse == "collapse"
@@ -79,13 +79,13 @@
 PASS document.getElementById("columnThatShouldNotBeRendered").offsetWidth is 0
 
 Tests for offsetHeight:
-PASS document.getElementById("productNo").parentNode.offsetHeight is 91
-PASS document.getElementById("productNo").offsetHeight is 91
-PASS document.getElementById("productName").offsetHeight is 91
-PASS document.getElementById("hasMAndHasNAndHasO").offsetHeight is 91
-PASS document.getElementById("hasMAndHasN").offsetHeight is 91
-PASS document.getElementById("hasO").offsetHeight is 91
-PASS document.getElementById("hasP").offsetHeight is 91
+PASS document.getElementById("productNo").parentNode.offsetHeight is 88
+PASS document.getElementById("productNo").offsetHeight is 88
+PASS document.getElementById("productName").offsetHeight is 88
+PASS document.getElementById("hasMAndHasNAndHasO").offsetHeight is 88
+PASS document.getElementById("hasMAndHasN").offsetHeight is 88
+PASS document.getElementById("hasO").offsetHeight is 88
+PASS document.getElementById("hasP").offsetHeight is 88
 PASS document.getElementById("columnThatShouldNotBeRendered").offsetHeight is 0
 PASS successfullyParsed is true

DIFF LINK: 
https://build.webkit.org/results/Apple-Sequoia-Release-AppleSilicon-WK2-Tests/284293%40main%20(79)/fast/table/col-and-colgroup-offsets-diff.txt

REPRODUCIBILITY: 
I was able to reproduce the issue on ToT using command: 
run-webkit-tests fast/table/col-and-colgroup-offsets.html --iterations=10

DESCRIPTION: 
This looks like it needs a re-baseline since most of the diff is offset by a couple pts.

[Radar WebKit Bug Importer]

<rdar://problem/136772789>

[Anfernee Viduya]

On a second look, a rebase does not seem to be solution. Investigating further.

[EWS]

Test gardening commit 284306@main (321082fa27eb): <https://commits.webkit.org/284306@main>

Reviewed commits have been landed. Closing PR #34315 and removing active labels.

[Jonathan Bedard]

Looking closely at the history of this test, there are a couple things going on.

First, this test absolutely needs a rebase, I’ll work on that (and it will be covered in this bug)
Second, these tests are confusing. The “Fail” in the HTML is deliberately checking that two values ARE NOT equal, so we need to make the test itself better.
Third, the method checking that values are equal dumps the values it’s comparing to test output. That’s really horrible, given how LayoutTests work. Even passing tests which do this will need to be constantly rebased. I would like to change the method we're using for comparison too.

[Jonathan Bedard]

Pull request: https://github.com/WebKit/WebKit/pull/38868

[EWS]

Committed 288816@main (f97e84989be4): <https://commits.webkit.org/288816@main>

Reviewed commits have been landed. Closing PR #38868 and removing active labels.