)]}'
{
  "commit": "d37783b6e5b44aba63e2aa5208070688770b9caa",
  "tree": "f11efc9444b45a6809fce0f049a43149f7a9b7d7",
  "parents": [
    "411c896ec762a00e9376af9d931f0aeedf9069de"
  ],
  "author": {
    "name": "Ralf W. Grosse-Kunstleve",
    "email": "rwgkio@gmail.com",
    "time": "Sat Oct 31 11:10:26 2026 -0700"
  },
  "committer": {
    "name": "GitHub",
    "email": "noreply@github.com",
    "time": "Sat Oct 31 11:10:26 2026 -0700"
  },
  "message": "test: fix GIL subprocess test correctness (#6211)\n\n* test: distinguish GIL deadlocks from subprocess startup and shutdown\n\nWait for explicit child-start and function-return signals so slow imports\nor interpreter shutdown cannot masquerade as a blocked test. Replace the\nintentional-deadlock special case with a dedicated GIL-only detector\ncontrol, and add regressions for delayed startup and shutdown.\n\nValidation: CPython 3.14.4 focused GIL suite, 90 passed on regular Python;\n72 passed and 18 skipped on free-threaded Python. Spawn checks and the\nall-files pre-commit hooks passed. Both timing regressions fail when their\ncorresponding safeguards are removed.\n\n* test: rename free-threaded skip marker to match its condition\n\nRename skipif_not_free_threaded to skipif_free_threaded so the name states\nwhen pytest skips the test. The predicate and test behavior are unchanged.\n\nValidation: all repository pre-commit hooks passed.\n\n* test: check the internals count for every GIL operation pattern\n\n* test: propagate exceptions from GIL test worker threads\n\n* test: avoid semaphore finalizers in GIL subprocess tests\n\nReplace the started/finished multiprocessing.Events with ordered messages\non a one-way pipe. Each Event allocates named semaphores under forkserver;\nthe per-case subprocess matrix can accumulate hundreds before PyPy garbage\ncollection, triggering reentrant resource_tracker cleanup in later tests.\n\nPreserve separate startup, execution, and shutdown allowances. Keep the\nparent writer open until child cleanup so EOF cannot masquerade as a phase\nsignal, and close all connection endpoints deterministically.\n\nAdd a regression that disables multiprocessing.Event and runs a real\nsubprocess successfully. It fails with the previous harness.\n\nValidation: clean CPython 3.14.4 suites passed (1454 regular, 1438\nfree-threaded; Catch2 31 cases/1709 assertions each). Linux PyPy 3.11/8.0.1\nfull serial Python suite passed (1065 passed). Spawn checks and all\npre-commit hooks passed.",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "8c0ac9597dd18883a5ee0b3e74d1959c26aa2496",
      "old_mode": 33188,
      "old_path": "tests/test_gil_scoped.py",
      "new_id": "2072e2661f2362c765d1373c19aef049e4ba6deb",
      "new_mode": 33188,
      "new_path": "tests/test_gil_scoped.py"
    }
  ]
}
