New state: integrate new state tests into CI

[brbrr]

No description provided.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 76.49%. Comparing base (d29fc03) to head (09d6758).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3043      +/-   ##
==========================================
+ Coverage   76.47%   76.49%   +0.01%     
==========================================
  Files         402      402              
  Lines       36770    36770              
==========================================
+ Hits        28120    28127       +7     
+ Misses       6673     6666       -7     
  Partials     1977     1977              

Flag	Coverage Δ
new-state	75.81% <ø> (?)
old-state	76.47% <ø> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

This pull request is stale because it has been open 35 days with no activity. Remove stale label or comment or this will be closed in 14 days.

[github-actions]

This pull request is stale because it has been open 35 days with no activity. Remove stale label or comment or this will be closed in 14 days.

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Integrates the “new state” test mode into CI by expanding the existing test job matrix to run both old and new state variants and tag Codecov uploads accordingly.

Changes:

• Add a juno-new-state boolean axis to the workflow matrix.
• Plumb the matrix value into JUNO_NEW_STATE and job naming for visibility.
• Add a Codecov flags value to distinguish old vs new state coverage uploads.

---

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

[github-actions]

Claude encountered an error —— View job

---

I'll analyze this and get back to you.

[github-actions]

Claude encountered an error —— View job

---

I'll analyze this and get back to you.

[github-actions]

Claude finished @brbrr's task in 3m 4s —— View job

---

PR Review: Integrate new state tests into CI

• Read the workflow file and PR diff
• Verify Copilot's concern about matrix.juno-new-state (already fixed in commit 1c63f0ba — keys use underscores)
• Cross-check Makefile (test, test-cover) against the new matrix axis and env var plumbing
• Post inline findings on the workflow file

Summary

Important

• On ubuntu-latest, the matrix axis is partly redundant and partly incorrect. make test-cover already runs both modes sequentially, and its first go test line does not clear JUNO_NEW_STATE — so when matrix.juno_new_state=true propagates through env:, the file written to coverage.old.out is actually produced under new-state mode. Net effect: ubuntu-latest runs the full coverage twice (2× wall time) and the per-flag separation in Codecov is muddled.
• Both matrix variants upload the same two files (coverage.old.out,coverage.new.out) and differ only in the flags value, so the new-state / old-state flag split on Codecov isn't actually a clean separation of modes.

See inline comments on .github/workflows/juno-test.yml for two suggested fixes (drop the axis on ubuntu-latest, or refactor make test-cover to honor JUNO_NEW_STATE and produce one file per pass).

Looks good

• The recent rename to juno_new_state (underscores) addresses Copilot's earlier concern.
• The ${{ matrix.juno_new_state && ... || ... }} expressions for the job name and the Codecov flag evaluate correctly given GH Actions' truthy fallback semantics.
• For non-ubuntu runners, make test does respect the inherited JUNO_NEW_STATE, so those matrix cells do exercise the new-state code path as intended.

Nits

• None blocking.
  · Branch

[github-actions]

Both matrix variants upload the same two files — files: coverage.old.out,coverage.new.out on line 73 — differing only by the flags value here. Combined with the env-leak issue noted on the env block above, the new-state flag in Codecov ends up labelling the same artifact pair the old-state flag also labels, so per-flag coverage on Codecov is not actually measuring the two modes separately.

Aligned with the suggestion above: upload one file per matrix run (e.g., coverage.new.out when juno_new_state is true, otherwise coverage.old.out), or drop the matrix axis on ubuntu-latest and do two uploads from a single job — one per flag — pointing at the file actually produced by that mode.

[github-actions]

Important — matrix axis is redundant and corrupts coverage on ubuntu-latest.

On ubuntu-latest the test step runs make test-cover, which already executes both modes back-to-back:

test-cover: clean-testcache rustdeps
    mkdir -p coverage
    go test $(GO_TAGS) -coverpkg=$(PKG) -coverprofile=coverage/coverage.old.out -covermode=atomic $(PKG)
    JUNO_NEW_STATE=true go test $(GO_TAGS) -coverpkg=$(PKG) -coverprofile=coverage/coverage.new.out -covermode=atomic $(PKG)

Two problems flow from this:

1. Double work. With juno_new_state: [false, true], ubuntu-latest now runs the full coverage matrix twice — 4 full test passes instead of 2 — for the same artifacts.
2. Env leak corrupts coverage.old.out when matrix.juno_new_state=true. The first line of test-cover does not unset JUNO_NEW_STATE, so it inherits true from the job env. The "old" coverage file then actually contains new-state coverage. The flags: new-state upload below will therefore include genuine new-state data, but the flags: old-state upload contains the correct mix — and both runs still upload the same two files, so the codecov flag split is mostly cosmetic.

Suggested fix — pick one:

• (a) Restrict the matrix axis to non-ubuntu runners (make test honors JUNO_NEW_STATE correctly), and keep make test-cover running once on ubuntu-latest with one Codecov upload per file/flag.
• (b) Refactor make test-cover to honor JUNO_NEW_STATE (single pass producing one file), then the matrix axis drives both modes uniformly and each upload carries the right flag.