tests/tests.py (1)

1628-1639: Test validates the fix, but comments may be misleading.

The test correctly verifies that the workflow doesn't crash when checkpoint outputs are deleted between runs, which addresses issue #3879. However, the comments "should not fail (but nothing to do)" are potentially misleading — when checkpoint outputs are deleted, the checkpoint should be re-executed, so there is actually work to do.

Consider verifying that deleted outputs are actually regenerated after rerun to strengthen the test:

💡 Optional: Verify outputs are regenerated

 def test_checkpoint_missing_output():
     """test for issue 3879, also covers 3009"""
     # normal run to create the checkpoint output and final output
     tmpdir = run(dpath("test_checkpoint_missing_output"), cleanup=False)
     assert tmpdir
-    # should not fail (but nothing to do)
+    # Delete checkpoint output; rerun should regenerate it
     (tmpdir / "output" / "test_1.txt").unlink()
     run(dpath("test_checkpoint_missing_output"), cleanup=False, tmpdir=tmpdir)
-    # should not fail (but nothing to do)
+    assert (tmpdir / "output" / "test_1.txt").exists(), "test_1.txt should be regenerated"
+    # Delete another checkpoint output; rerun should regenerate it
     (tmpdir / "output" / "test_0.txt").unlink()
     run(dpath("test_checkpoint_missing_output"), cleanup=False, tmpdir=tmpdir)
+    assert (tmpdir / "output" / "test_0.txt").exists(), "test_0.txt should be regenerated"

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tests/tests.py` around lines 1628 - 1639, The test_comment is misleading and
the test should assert that deleted checkpoint outputs are regenerated: in
test_checkpoint_missing_output update the comments to reflect that deleting
checkpoint outputs should trigger re-execution, and after each
run(dpath("test_checkpoint_missing_output"), cleanup=False, tmpdir=tmpdir) add
assertions that (tmpdir / "output" / "test_1.txt") and (tmpdir / "output" /
"test_0.txt") respectively exist (or contain expected contents) to verify the
checkpoint was rerun; reference the test function name
test_checkpoint_missing_output, the run(...) helper, and the tmpdir / "output" /
"<file>" paths when adding the checks.

tests/tests.py (1)

1628-1639: Test lacks assertions verifying checkpoint outputs are regenerated.

The test validates that the workflow doesn't crash when checkpoint outputs are missing, but doesn't assert that the deleted files (test_1.txt, test_0.txt) were actually regenerated after each rerun. Given the PR objective is to ensure "checkpoint outputs are properly detected and regenerated when deleted," consider adding assertions to verify the files exist after each run.

Additionally, the comments "target file exists so nothing to do" are misleading since the files are being deleted and the checkpoint should be re-run to regenerate them.

💡 Suggested improvement

 def test_checkpoint_missing_output():
     """test for issue 3879, also covers 3009"""
     # normal run to create the checkpoint output and final output
     tmpdir = run(dpath("test_checkpoint_missing_output"), cleanup=False)
     assert tmpdir
-    # should not fail (target file exists so nothing to do)
+    # delete checkpoint output and verify it gets regenerated
     (tmpdir / "output" / "test_1.txt").unlink()
     run(dpath("test_checkpoint_missing_output"), cleanup=False, tmpdir=tmpdir)
-    # should not fail (target file exists so nothing to do)
+    assert (tmpdir / "output" / "test_1.txt").exists(), "test_1.txt should be regenerated"
+    # delete another checkpoint output and verify regeneration
     (tmpdir / "output" / "test_0.txt").unlink()
     run(dpath("test_checkpoint_missing_output"), cleanup=False, tmpdir=tmpdir)
+    assert (tmpdir / "output" / "test_0.txt").exists(), "test_0.txt should be regenerated"
+    shutil.rmtree(tmpdir, ignore_errors=ON_WINDOWS)

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@tests/tests.py` around lines 1628 - 1639, The test function
test_checkpoint_missing_output currently deletes checkpoint output files but
doesn't assert they are regenerated; update test_checkpoint_missing_output to
assert existence of (tmpdir / "output" / "test_1.txt") after the first rerun and
assert existence of (tmpdir / "output" / "test_0.txt") after the second rerun
(using the existing run(...) helper and dpath(...) to locate the files), and
revise the misleading comments to say that files are deleted and should be
regenerated rather than "target file exists so nothing to do".