Skip to content

Change always_forkserver_on_unix() to use_multiprocessing_forkserver_on_linux() #4577

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Mar 16, 2023
Merged

Change always_forkserver_on_unix() to use_multiprocessing_forkserver_on_linux() #4577

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 4 additions & 5 deletions tests/conftest.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,8 +8,8 @@
import difflib
import gc
import multiprocessing
import os
import re
import sys
import textwrap
import traceback

Expand All @@ -25,17 +25,16 @@


@pytest.fixture(scope="session", autouse=True)
def always_forkserver_on_unix():
if os.name == "nt":
def use_multiprocessing_forkserver_on_linux():
if sys.platform != "linux":
# The default on Windows and macOS is "spawn": If it's not broken, don't fix it.
return

# Full background: https://github.com/pybind/pybind11/issues/4105#issuecomment-1301004592
# In a nutshell: fork() after starting threads == flakiness in the form of deadlocks.
# It is actually a well-known pitfall, unfortunately without guard rails.
# "forkserver" is more performant than "spawn" (~9s vs ~13s for tests/test_gil_scoped.py,
# visit the issuecomment link above for details).
# Windows does not have fork() and the associated pitfall, therefore it is best left
# running with defaults.
multiprocessing.set_start_method("forkserver")


Expand Down