gh-118846: Fix free-threading test failures when run sequentially by colesbury · Pull Request #118864 · python/cpython · GitHub

The free-threaded build currently immortalizes some objects once the first thread is started. This can lead to test failures depending on the order in which tests are run. This PR addresses those failures by suppressing immortalization or skipping the affected tests.

Issue: Freethreading build fails -m test --pgo: test_ordered_dict and test_struct failed #118846

The free-threaded build currently immortalizes some objects once the first thread is started. This can lead to test failures depending on the order in which tests are run. This PR addresses those failures by suppressing immortalization or skipping the affected tests.

@hroncok - this should address the remaining test failures in conjunction with #118862

LGTM!

I still have:

======================================================================
ERROR: test_builtin_with_more_than_four_children (test.test_pydoc.test_pydoc.PydocDocTest.test_builtin_with_more_than_four_children)
Tests help on builtin object which have more than four child classes.
----------------------------------------------------------------------
Traceback (most recent call last):
  File ".../cpython/Lib/test/test_pydoc/test_pydoc.py", line 537, in test_builtin_with_more_than_four_children
    text = doc.docclass(object)
  File ".../cpython/Lib/pydoc.py", line 1424, in docclass
    subclasses = sorted(
        (str(cls.__name__) for cls in type.__subclasses__(object)
         if not cls.__name__.startswith("_") and cls.__module__ == "builtins"),
        key=str.lower
    )
  File ".../cpython/Lib/pydoc.py", line 1426, in <genexpr>
    if not cls.__name__.startswith("_") and cls.__module__ == "builtins"),
                                            ^^^^^^^^^^^^^^
  File ".../cpython/Lib/test/test_inspect/test_inspect.py", line 775, in __module__
    raise AttributeError
AttributeError: . Did you mean: '__reduce__'?

----------------------------------------------------------------------
Ran 108 tests in 1.709s

FAILED (errors=1, skipped=3)
test test.test_pydoc.test_pydoc failed


======================================================================
FAIL: test_bug1055820c (test.test_gc.GCTogglingTests.test_bug1055820c)
----------------------------------------------------------------------
Traceback (most recent call last):
  File ".../cpython/Lib/test/test_gc.py", line 1361, in test_bug1055820c
    self.fail("gc didn't happen after 10000 iterations")
    ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: gc didn't happen after 10000 iterations

======================================================================
FAIL: test_bug1055820d (test.test_gc.GCTogglingTests.test_bug1055820d)
----------------------------------------------------------------------
Traceback (most recent call last):
  File ".../cpython/Lib/contextlib.py", line 85, in inner
    return func(*args, **kwds)
  File ".../cpython/Lib/test/test_gc.py", line 1429, in test_bug1055820d
    self.fail("gc didn't happen after 10000 iterations")
    ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AssertionError: gc didn't happen after 10000 iterations

----------------------------------------------------------------------
Ran 49 tests in 14.476s

FAILED (failures=2, skipped=4)
test test_gc failed

Not sure if the pydoc thing is related, but the gc might.

The pydoc failure is a funny interaction between tests. In test_inspect.py, we create a class that raises an error when you access __module__:

cpython/Lib/test/test_inspect/test_inspect.py

Lines 771 to 775 in 98ff3f6

    
           def test_getfile_class_without_module(self): 
        
               class CM(type): 
        
                   @property 
        
                   def __module__(cls): 
        
                       raise AttributeError

In the free-threaded build, this class is now immortalized so it's still alive later on when test_pydoc tries to generate the pydoc for all subclasses of object.

Ok, I think the test_pydoc and test_gc failures are fixed now too.

Thanks @colesbury for the PR 🌮🎉.. I'm working now to backport this PR to: 3.13.
🐍🍒⛏🤖

…ly (pythonGH-118864) The free-threaded build currently immortalizes some objects once the first thread is started. This can lead to test failures depending on the order in which tests are run. This PR addresses those failures by suppressing immortalization or skipping the affected tests. (cherry picked from commit b309c8e) Co-authored-by: Sam Gross <colesbury@gmail.com>

GH-118927 is a backport of this pull request to the 3.13 branch.

…lly (GH-118864) (#118927) The free-threaded build currently immortalizes some objects once the first thread is started. This can lead to test failures depending on the order in which tests are run. This PR addresses those failures by suppressing immortalization or skipping the affected tests. (cherry picked from commit b309c8e) Co-authored-by: Sam Gross <colesbury@gmail.com>

This is the only one I got now:

======================================================================
ERROR: test_expr_context (test.test_ast.ASTConstructorTests.test_expr_context)
----------------------------------------------------------------------
Traceback (most recent call last):
  File ".../cpython/Lib/test/test_ast.py", line 3042, in test_expr_context
    self.assertIsInstance(name.ctx, ast.Load)
                          ^^^^^^^^
AttributeError: 'Name' object has no attribute 'ctx'

----------------------------------------------------------------------
Ran 175 tests in 0.765s

FAILED (errors=1, skipped=1)

Just checking, this is from ./python -m test -W on the 3.13 branch? Is it a debug build?

Note that that AST test was only merged (and backported to 3.13) yesterday in #118854

I might have git pulled but forgot to rebuild it. Mea culpa. Trying fresh.

All is good now, thank you!

…ly (python#118864) The free-threaded build currently immortalizes some objects once the first thread is started. This can lead to test failures depending on the order in which tests are run. This PR addresses those failures by suppressing immortalization or skipping the affected tests.

colesbury added tests Tests in the Lib/test dir topic-free-threading needs backport to 3.13 bugs and security fixes labels May 9, 2024

colesbury requested a review from DinoV May 9, 2024 19:31

colesbury requested review from ericsnowcurrently, gvanrossum and markshannon as code owners May 9, 2024 19:31

bedevere-app Bot added the awaiting core review label May 9, 2024

bedevere-app Bot mentioned this pull request May 9, 2024

Freethreading build fails -m test --pgo: test_ordered_dict and test_struct failed #118846

Closed

colesbury added the skip news label May 9, 2024

DinoV approved these changes May 9, 2024

View reviewed changes

bedevere-app Bot added awaiting merge and removed awaiting core review labels May 9, 2024

Merge branch 'main' into pythongh-118846 -all-tests

923dd2d

colesbury added 2 commits May 9, 2024 22:21

Fix test_pydoc failure

f469a2e

Fix test_gc when run sequentially

0c0c9b4

colesbury merged commit b309c8e into python:main May 10, 2024

colesbury deleted the gh-118846-all-tests branch May 10, 2024 20:29

bedevere-app Bot removed the awaiting merge label May 10, 2024

bedevere-app Bot removed the needs backport to 3.13 bugs and security fixes label May 10, 2024

JelleZijlstra mentioned this pull request May 28, 2024

backport a9a74da 3.13 #119642

Closed

gh-118846: Fix free-threading test failures when run sequentially by colesbury · Pull Request #118864 · python/cpython · GitHub

Uh oh!

Uh oh!

gh-118846: Fix free-threading test failures when run sequentially - #118864

gh-118846: Fix free-threading test failures when run sequentially#118864colesbury merged 4 commits intopython:mainpython/cpython:mainfrom colesbury:gh-118846-all-testscolesbury/cpython:gh-118846-all-testsCopy head branch name to clipboard

Conversation

colesbury commented May 9, 2024 • edited by bedevere-app Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

colesbury commented May 9, 2024

Uh oh!

DinoV left a comment

Choose a reason for hiding this comment

Uh oh!

hroncok commented May 9, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

colesbury commented May 9, 2024

Uh oh!

colesbury commented May 10, 2024

Uh oh!

miss-islington-app Bot commented May 10, 2024

Uh oh!

bedevere-app Bot commented May 10, 2024

Uh oh!

hroncok commented May 10, 2024

Uh oh!

colesbury commented May 10, 2024

Uh oh!

AlexWaygood commented May 10, 2024

Uh oh!

hroncok commented May 11, 2024

Uh oh!

hroncok commented May 11, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

gh-118846: Fix free-threading test failures when run sequentially#118864
colesbury merged 4 commits into
python:mainfrom
colesbury:gh-118846-all-tests

colesbury commented May 9, 2024 •
edited by bedevere-app Bot
Loading

hroncok commented May 9, 2024 •
edited
Loading