Make JRuby 10 a required CI target

[etagwerker]

What

RubyCritic already runs end-to-end on JRuby, but CI listed jruby-10.0 only as an experimental continue-on-error target, so it never actually had to pass. This PR makes JRuby 10 a required CI target and fixes the one thing that was keeping the suite from going green.

The bug

The application was fine. The test suite was the blocker. When flog/ruby_parser loaded before reek, Ruby 3.4's bundled_gems require shim clobbered Zeitwerk's implicit-namespace autoload on JRuby, and reek's dry-schema Macros namespace failed to load:

LoadError: cannot load such file -- .../dry-schema-1.16.0/lib/dry/schema/macros

It only showed up under the full rake test load order (a single test file passed on its own), which is why it looked flaky at first.

The fix

Force reek's schema to load up front in test_helper.rb, guarded by defined?(JRUBY_VERSION), before any flog/ruby_parser require. The application itself is untouched because it already loads reek before flog.

• test/test_helper.rb: eager reek schema load on JRuby, with a comment explaining the ordering issue
• .github/workflows/main.yml: move jruby-10.0 from the experimental include: block into the required matrix across all four jobs (Specs, Rubocop, Reek, Minitest); ruby-head stays experimental
• README.md: note JRuby 10 support in the Compatibility section

Verification

Run locally on JRuby 10.0.4.0 (Ruby 3.4.5):

Check	JRuby 10	MRI 4.0
rake test	164 runs, 0 failures	164 runs, 0 failures
rake spec	14 runs, 0 failures	n/a
rubocop	0 offenses	n/a
rake reek	0 warnings	n/a
mdl	clean	clean

No MRI regression.

🤖 Generated with Claude Code

[faisal]

If we do this should we revert the change that made failing runs flag green but annotate problems?

[etagwerker]

@faisal Good question. I don't think we want to revert it entirely, since ruby-head is still experimental after this PR and relies on that same continue-on-error: ${{ matrix.experimental }} to stay non-blocking. ruby-head is a genuinely moving target, so I'd rather it annotate failures than block merges.

What this PR does is narrow that "green but annotated" behavior down to just ruby-head by graduating jruby-10.0 out of the experimental set into the required matrix. So the mechanism stays, but it only applies where we actually want it.

Were you thinking of something more specific that we'd want to clean up, like dropping the experimental include block altogether, or making ruby-head failures more visible than they are today? Happy to fold that in if so.

[faisal]

Before answering I reviewed the history of the changes here, and came to the conclusion that this PR is likely the right approach (so long as we're willing and able to keep things running on jruby).

[faisal]

Also does this mean we are promoting jruby to "fully supported?"

[Copilot]

Pull request overview

This PR makes JRuby 10 a required (non-experimental) CI target for RubyCritic and addresses a JRuby-only test-suite load-order failure by eagerly loading Reek’s configuration schema early in the test boot process.

Changes:

• Promote jruby-10.0 from an experimental to a required CI matrix entry across the main test/lint jobs.
• Add a JRuby-only eager load in test_helper.rb to prevent a dry-schema Macros namespace load failure under specific require ordering.
• Document JRuby 10 support in the README and record the change in the changelog.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
test/test_helper.rb	Eager-load Reek on JRuby to stabilize gem loading/namespace autoload behavior in the test environment.
README.md	Document JRuby 10 compatibility and that it’s covered by CI.
CHANGELOG.md	Note the CI policy change and accompanying JRuby test-suite fix.
.github/workflows/main.yml	Add jruby-10.0 to the required matrix for Specs, Rubocop, Reek, and Minitest jobs (keeping ruby-head experimental).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[etagwerker]

Also does this mean we are promoting jruby to "fully supported?"

@faisal Yes, I think so.