CI: Run docstring examples via pytest --doctest-modules

[mmcky]

Closes #866.

Wires the docstring-example sweep into ci.yml so examples become executable specifications and cannot silently rot again after the fixes in #864.

Note

Stacked on #864 (docs/consistency-audit), since the doctest run only passes with those 31 example fixes in place. After #864 merges, this PR should be retargeted to main (GitHub does this automatically on branch deletion) — please don't merge it before #864 lands.

What this adds

A new quantecon/conftest.py resolving both blockers from #866:

• Blocker 1 (print-state leak): an autouse fixture snapshots np.get_printoptions() before each test and restores it after. The six game_theory examples that call np.set_printoptions(precision=4) stay exactly as published, and the 8 spurious failures they previously caused in markov, optimize and random disappear. This is the conftest route the issue weakly preferred: invisible to readers, with a docstring in the fixture explaining why it exists for future contributors.
• Blocker 2 (unpassable modules): collect_ignore excludes util/notebooks.py (fetches over the network) and util/timing.py (prints wall-clock durations). This keeps the rendered docs free of # doctest: +SKIP directives — no cost paid by readers to satisfy CI.
• The third, milder case (util/array.py::searchsorted emitting a DeprecationWarning) needs no handling: no strict warning filter is configured, so the warning is reported but does not fail the run. If a strict filter is ever adopted, this example will need a filterwarnings exemption.

A CI step in the existing tests job, after the regular pytest run:

- name: Run docstring examples (doctest)
  if: runner.os == 'Linux'
  shell: bash -l {0}
  run: |
    pytest --doctest-modules quantecon --ignore-glob='*/tests/*'

It runs on Linux only: one platform is enough to stop drift, float reprs are the platform-sensitive part of doctest output, and it sidesteps --ignore-glob path-separator issues on Windows. The --ignore-glob='*/tests/*' keeps the step from re-running the whole unit-test suite (that already ran in the previous step); the sweep itself takes ~8 seconds.

Verification

On this branch: pytest --doctest-modules quantecon --ignore-glob='*/tests/*' → 47 passed (was 11 failed, 39 passed before the conftest — the 8 print-option leaks plus notebooks/timing ×2, matching the issue's inventory). A regular pytest run over quantecon/tests, util/tests and game_theory/tests/test_pure_nash.py still passes with the new conftest in place.

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Adds automated execution of docstring examples as part of CI, turning doctests into continuously-validated specifications for the quantecon/ package. This aims to prevent future drift/rot in documented examples (per #866) by running pytest --doctest-modules in CI and addressing known collection/state-leak blockers.

Changes:

• Add a quantecon/conftest.py to (a) reset NumPy print options between tests and (b) exclude modules whose doctests can’t pass deterministically (network / wall-clock output).
• Extend the existing GitHub Actions test workflow to run a doctest sweep for quantecon on Linux.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
quantecon/conftest.py	Adds pytest config to ignore non-deterministic doctest modules and an autouse fixture to prevent NumPy print-option leaks across doctests.
.github/workflows/ci.yml	Adds a Linux-only CI step to run pytest --doctest-modules quantecon while ignoring */tests/* to avoid re-running unit tests.

[mmcky]

thanks @oyamad -- this was a good idea.