Cucumber and Minitest pass on my machine.

Hm, tests seem to fail due to 1 rdiscount failure. Maybe it's related to the i18n version? Hm. Also, I can't believe we still test rdiscount – we don't waste our energy on rdiscount in Jekyll 4, do we?

Failure:
TestRdiscount#test_: rdiscount should render toc.  [/home/runner/work/jekyll/jekyll/test/test_rdiscount.rb:47]
Minitest::Assertion:
--- expected
+++ actual
@@ -1,13 +1,13 @@
-"<a name=\"Header.1\"></a>
+"<a name=\"Header-1\"></a>
 <h1>Header 1</h1>
 
-<a name=\"Header.2\"></a>
+<a name=\"Header-2\"></a>
 <h2>Header 2</h2>
 
 <p><ul>
- <li><a href=\"#Header.1\">Header 1</a>
+ <li><a href=\"#Header-1\">Header 1</a>
  <ul>
-  <li><a href=\"#Header.2\">Header 2</a></li>
+  <li><a href=\"#Header-2\">Header 2</a></li>
  </ul>
  </li>
 </ul>

Rdiscount failure also occurred here: #9272

So it must be a breaking change. We should just update our test.

So it must be a breaking change. We should just update our test.

gem rdiscount only bumped from 2.2.0.2 to 2.2.7 to mirror underlying native extension discount which bumped from 2.2.0 to 2.2.7. So, assuming they follow Semantic Versioning, our test was depending on a buggy behavior.

I recommend updating the test (or locking to last known compatible version) via a separate pull request (or a direct commit to the 3.9-stable branch).

I can't believe we still test rdiscount – we don't waste our energy on rdiscount in Jekyll 4, do we?

Jekyll 4 dropped built-in support for rdiscount and redcarpet. So, related tests were removed as well.
The 3.x-stable branches still tests these Markdown processors.