homepage

Recent test failures have been noticed on Windows 10. The full test log has been attached. The tests test_distutils, test_pdb, test_signal and test_threading require a little bit of attention since there seems to be some kind of error.

I wonder if these four tests do something that affects all threads? The test driver module runs tests in parallel (if you ask it to run all or most tests, like the default).

Pdb error source - https://github.com/python/cpython/blob/9aea31deddf7458be3546f72185740f3cd06687f/Lib/test/test_pdb.py#L1270
This is where the error in Pdb is originating from.

The test expects the <built-in default_int_handler> but gets sigint_handler method of a pdb.Pdb instance instead.

Can you submit a PR that fixes it?

I expect parallelism is a red herring: early in the test output attached to this report:
0:00:04 Run tests sequentially
and there's no other evidence in the output that multiple tests are running simultaneously.
Also on Win10, the 4 failing tests here pass for me if I run them one at a time, so it's not obvious.
I _suspect_ that what's going wrong with test_pdb is the root cause: every now & again, for some weeks now, when I try to run tests on Windows I come back to the cmd.exe window and see that it's just sitting there, waiting at a pdb prompt.
In the output attached to this report, note that after test_threading starts, 
(Pdb) continue
(Pdb) continue
(Pdb) continue
(Pdb) continue
appears out of the blue. But test_pdb is long finished by that time.
But I'm clueless about current pdb internals.

Yes you're right. Running the failed tests individually does not trigger the errors. And the
(Pdb) continue
(Pdb) continue
(Pdb) continue
(Pdb) continue
actually is manually entered by me. My tests also stop at that prompt but somehow entering these commands again allow us to continue. Why are the errors triggering at all? The tests that are failing when ran alongside the test suite haven't been messed with for a very long time now and still somehow out of the blue they are failing. Any thoughts?

Yes, I also see that pdb prompt. I recommend looking at what different
configurations are used by test.libregrtest.main() when no tests are passed
on the command line.

File "C:\github\cpython\lib\test\test_distutils.py", line 15, in <module>
 import distutils.tests
 File "C:\github\cpython\lib\contextlib.py", line 140, in __exit__
 next(self.gen)
 File "C:\github\cpython\lib\test\support\warnings_helper.py", line 179, in _filterwarnings
 raise AssertionError("filter (%r, %s) did not catch any warning" %
AssertionError: filter ('The distutils package is deprecated', DeprecationWarning) did not catch any warning
Ah, test_distutils fails if distutils was already imported previously. The test should be fixed.
"""
test test_pdb failed -- Traceback (most recent call last):
 File "C:\github\cpython\lib\doctest.py", line 2205, in runTest
 raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for test.test_pdb.test_pdb_issue_20766
 File "C:\github\cpython\lib\test\test_pdb.py", line 1270, in test_pdb_issue_20766
"""
This is a surprising failure. No clue why test_pdb fails when other tests are run previously (when tests are run sequentially).
See also bpo-43960 

Shreyan Avigyan:
> And the "(Pdb) continue (...) actually is manually entered by me.
Do you mean that you modified the Python source code? Does the issue go away if you revert your change or if you test a newly installed Python?

I think the test failures cascade, and some state is not being cleaned up. I'm trying to work through them now, but if someone else is as well, do say so and I'll find something else to do.

Shreyan Avigyan:
> And the "(Pdb) continue (...) actually is manually entered by me.
Victor Stinner:
Do you mean that you modified the Python source code?
Me:
Doubt it. For me, with more words: the "(Pdb) " prompt appears all by itself, by magic, and the test run is stuck then. I bet Shreyan meant to say "so I manually entered 'continue [RETURN]' at the pdb prompt to get it unstuck again".
> Does the issue go away if you revert your change or
> if you test a newly installed Python?
For me, I was using Win10 x64 CPython built from yesterday's github master/main.
And thanks for the distutils clue! I bet that one is a distinct issue.
To make it more confusing, all 4 tests passed when I ran with `-j` (to force parallel test execution) - but then test_compileall failed instead :-)

> To make it more confusing, all 4 tests passed when I ran with `-j` (to force parallel test execution) - but then test_compileall failed instead :-)
bpo-37387 "test_compileall fails randomly on Windows when tests are run in parallel" about random PermissionError.

For test_logging test_namer_rotator_inheritance() error: I created bpo-43961 "[Windows] test_logging.test_namer_rotator_inheritance() logs a logging error". It doesn't like like an error and it only happens on Windows.

Oh. I also got this surprising (Pdb) prompt when running "python -m test" on Windows.
The prompt doesn't display any output, I only get the prompt display 
(...)
0:34:05 load avg: 1.06 [300/426/2] test_select
0:34:08 load avg: 1.03 [301/426/2] test_selectors
0:34:13 load avg: 0.94 [302/426/2] test_set
0:34:24 load avg: 0.80 [303/426/2] test_setcomps
0:34:26 load avg: 0.78 [304/426/2] test_shelve
0:34:30 load avg: 0.74 [305/426/2] test_shlex
0:34:32 load avg: 0.70 [306/426/2] test_shutil
0:34:36 load avg: 0.66 [307/426/2] test_signal
(Pdb) help
(Pdb) l
(Pdb) ?
(Pdb) help
(Pdb) where
(Pdb) down
(Pdb) p 1
(Pdb) p 2
(Pdb) 
*But* I managed to get more info about that issue:
(Pdb) f=open("debug.txt", "w")
(Pdb) import traceback
(Pdb) traceback.print_stack(file=f)
(Pdb) f.flush()
C:\vstinner\python\master\build\test_python_6260æ\debug.txt content:
(...)
 File "C:\vstinner\python\master\lib\test\libregrtest\runtest.py", line 154, in _runtest
 result = _runtest_inner(ns, test_name,
(...)
 File "C:\vstinner\python\master\lib\unittest\case.py", line 549, in _callTestMethod
 method()
 File "C:\vstinner\python\master\lib\test\test_signal.py", line 1345, in test_sigint
 signal.raise_signal(signal.SIGINT)
 File "C:\vstinner\python\master\lib\unittest\case.py", line 216, in __exit__
 def __exit__(self, exc_type, exc_value, tb):
 File "C:\vstinner\python\master\lib\bdb.py", line 92, in trace_dispatch
 return self.dispatch_call(frame, arg)
 File "C:\vstinner\python\master\lib\bdb.py", line 136, in dispatch_call
 self.user_call(frame, arg)
 File "C:\vstinner\python\master\lib\pdb.py", line 252, in user_call
 self.interaction(frame, None)
 File "C:\vstinner\python\master\lib\pdb.py", line 357, in interaction
 self._cmdloop()
 File "C:\vstinner\python\master\lib\pdb.py", line 322, in _cmdloop
 self.cmdloop()
 File "C:\vstinner\python\master\lib\cmd.py", line 138, in cmdloop
 stop = self.onecmd(line)
 File "C:\vstinner\python\master\lib\pdb.py", line 423, in onecmd
 return cmd.Cmd.onecmd(self, line)
 File "C:\vstinner\python\master\lib\cmd.py", line 216, in onecmd
 return self.default(line)
 File "C:\vstinner\python\master\lib\pdb.py", line 381, in default
 exec(code, globals, locals)
 File "<stdin>", line 1, in <module>
Using debug.txt, I also see that in test_signal, sys.gettrace() = <bound method Bdb.trace_dispatch of <pdb.Pdb object at 0x0000029D85AB3570>>.
Lib/test/libregrtest/save_env.py is supposed to mark a test as "env changed" (ENV_CHANGED) if sys.gettrace() is changed between two tests.

Doesn't the pdb prompt use the builtin input()?

See also bpo-43963: "test_interpreters has side effects on test_signal".

Running these 3 tests is enough to reproduce the issue:
-------------------
vstinner@DESKTOP-DK7VBIL C:\vstinner\python\master>python -m test test_interpreters test_pdb test_threading 
Running Debug|x64 interpreter...
0:00:00 Run tests sequentially
0:00:00 [1/3] test_interpreters
0:00:04 [2/3] test_pdb
test test_pdb failed -- Traceback (most recent call last):
 File "C:\vstinner\python\master\lib\doctest.py", line 2205, in runTest
 raise self.failureException(self.format_failure(new.getvalue()))
AssertionError: Failed doctest test for test.test_pdb.test_pdb_issue_20766
 File "C:\vstinner\python\master\lib\test\test_pdb.py", line 1270, in test_pdb_issue_20766
(...)
0:00:10 load avg: 0.03 [3/3/1] test_threading -- test_pdb failed
(Pdb) 
-------------------
Moreover, my PR 25674 (bpo-43963) fix these tests.

FYI I used the following patch to debug this issue:
diff --git a/Lib/unittest/case.py b/Lib/unittest/case.py
index 872f121127..ee2e1ff77f 100644
--- a/Lib/unittest/case.py
+++ b/Lib/unittest/case.py
@@ -546,6 +546,8 @@ def _callSetUp(self):
 self.setUp()
 def _callTestMethod(self, method):
+ if sys.gettrace() is not None:
+ raise Exception("sys.gettrace() is not None")
 method()
 def _callTearDown(self):
And I used the following script:
---------------
import random
import os.path
import sys
with open("tests") as fp:
 lines = list(filter(bool, (line.strip() for line in fp)))
print(len(lines))
filename = "test"
loop = 2
while os.path.exists(filename):
 filename = "test%s" % loop
 loop += 1
with open(filename, "w") as fp:
 for _ in range(5):
 test = random.choice(lines)
 print(test, file=fp)
 print("test_signal", file=fp)
with open(filename) as fp:
 for line in fp:
 print(line.rstrip())
args = [sys.executable, "-m", "test", "--fromfile=" + filename]
print("RUN", args)
os.execv(args[0], args)
---------------
I created the "tests" file using the command:
 python -m test --list-tests > tests
Then I removed all tests starting at test_signal, including test_signal. I removed slow tests like asyncio and multiprocessing tests.

Another issue with the tests is that it has become terribly slow. It's taking about 45 minutes to complete 1 run of the test suite. A few days ago, it took only 10 - 15 minutes. And moreover I think test_threading is also modifying sys.gettrace. The logs have the full error message.
--Snippet--
0:33:17 load avg: 0.80 [307/426/2] test_signal
(Pdb) continue
Warning -- sys.gettrace was modified by test_signal
 Before: None
 After: <bound method Bdb.trace_dispatch of <pdb.Pdb object at 0x0000021A5D2DCD70>>
test test_signal failed -- Traceback (most recent call last):
 File "C:\github\cpython\lib\test\test_signal.py", line 1345, in test_sigint
 signal.raise_signal(signal.SIGINT)
AssertionError: KeyboardInterrupt not raised
--Snippet--
0:38:22 load avg: 0.04 [353/426/3] test_threading
(Pdb) continue
(Pdb) continue
(Pdb) continue
(Pdb) continue
Warning -- sys.gettrace was modified by test_threading
 Before: None
 After: <bound method Bdb.trace_dispatch of <pdb.Pdb object at 0x0000021A5D2DCD70>>
--Snippet--

Shreyan Avigyan: please try an up to date version of the master branch, I fixed the Pdb issue with commit a09766deab5aff549f40f27080895e148af922ed.

Same errors. 
(Recompilation was done 1 or 2 hours ago against the then updated master branch. It may not include the latest commit)

FYI test_signal is no longer modifying sys.gettrace thanks to your fix. But test_threading still does. Do you have a fix for that?

I ran the test again and test_signal still modifies sys.gettrace()

All the test failures are side effects of other tests because when they are ran individually they do not trigger errors. Running sequentially causes the error.

I ran "python -m test" on an up to date master branch (commit fe52eb62191e640e720d184a9a1a04e965b8a062), there is a single remaining issue. test_distutils failed because the deprecation warning was already emited: issue fixed by PR 25675. The other issues are gone.
Shreyan Avigyan: "Same errors."
IMO it's a mistake on your side. You ran an outdated version or you didn't rebuild Python cleanly.
"FYI test_signal is no longer modifying sys.gettrace thanks to your fix. But test_threading still does. (...) I ran the test again and test_signal still modifies sys.gettrace()"
It's not my case on my machine, again it sounds like an issue on your side.

> Another issue with the tests is that it has become terribly slow. It's taking about 45 minutes to complete 1 run of the test suite. A few days ago, it took only 10 - 15 minutes.
It took 49 minutes on my Windows 10 VM. Are you sure that you ran the test suite sequentially when it was slower?
I advice you run using -j0 option to run tests in parallel: it's much faster, and it's also more reliable.
== Tests result: FAILURE ==
384 tests OK.
1 test failed:
 test_distutils
41 tests skipped:
 test_curses test_dbm_gnu test_dbm_ndbm test_devpoll test_epoll
 test_fcntl test_fork1 test_gdb test_grp test_ioctl test_kqueue
 test_multiprocessing_fork test_multiprocessing_forkserver test_nis
 test_openpty test_ossaudiodev test_pipes test_poll test_posix
 test_pty test_pwd test_readline test_resource test_smtpnet
 test_socketserver test_spwd test_syslog test_threadsignals
 test_timeout test_tix test_tk test_ttk_guionly test_urllib2net
 test_urllibnet test_wait3 test_wait4 test_winsound test_xmlrpc_net
 test_xxlimited test_xxtestfuzz test_zipfile64
Total duration: 49 min 12 sec
Tests result: FAILURE

New changeset c1a9535989cc7323099725503519a17f79d083f5 by Steve Dower in branch 'master':
bpo-43955: Handle the case where the distutils warning has already been triggered (GH-25675)
https://github.com/python/cpython/commit/c1a9535989cc7323099725503519a17f79d083f5

Thanks everyone for fixing these issues and thanks Shreyan for the bug report.
The main initial is now fixed, so I close the issue.
Let's continue the discussions in remaining open issues:
* bpo-37387: test_compileall fails randomly on Windows when tests are run in parallel
* bpo-43960: test_pdb fails when only some tests are run
* bpo-43961 [Windows] test_logging.test_namer_rotator_inheritance() logs a logging error