This issue tracker has been migrated to GitHub, and is currently read-only.
For more information, see the GitHub FAQs in the Python's Developer Guide.

Author vstinner
Recipients asheesh, asvetlov, bquinlan, davin, pitrou, python-dev, vstinner, yselivanov, zach.ware
Date 2020-06-03.13:52:31
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1591192351.75.0.369355045062.issue40692@roundup.psfhosted.org>
In-reply-to
Content
> Automated testing: In order to prevent regressions, I'm interested in hosting a buildbot for at least one year, hopefully in perpetuity, on which we call ./configure with ac_cv_posix_semaphores_enabled=no.

Can't you tune an unit test to prevent multiprocessing.synchronize to be imported? For example, sys.modules['multiprocessing.synchronize'] = None ensures that "import multiprocessing.synchronize" fails, even if the module exists and works.

$ python3
Python 3.8.3 (default, May 15 2020, 00:00:00) 
>>> import sys
>>> sys.modules['multiprocessing.synchronize'] = None
>>> import multiprocessing.synchronize
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: import of multiprocessing.synchronize halted; None in sys.modules

A whole buildbot sounds like an overkill solution to this problem.

On the other hand, multiprocessing tests are already ones of the slowest tests of the test suite *because* they try to test all possible combinations.
History
Date User Action Args
2020-06-03 13:52:31vstinnersetrecipients: + vstinner, bquinlan, pitrou, asvetlov, python-dev, zach.ware, yselivanov, davin, asheesh
2020-06-03 13:52:31vstinnersetmessageid: <1591192351.75.0.369355045062.issue40692@roundup.psfhosted.org>
2020-06-03 13:52:31vstinnerlinkissue40692 messages
2020-06-03 13:52:31vstinnercreate