Re: python self-tests hanging?

[Adam <adam%netbsd.org@localhost>]

>>> I noticed https://nvd.nist.gov/vuln/detail/CVE-2018-1000030
>>> against python 2.7.14, which is supposedly fixed by 2.7.15 which
>>> is now out.
>>> 
>>> I have an update of the python27 package lined up, but thought I
>>> would make sure this doesn't introduce any regressions, so wanted
>>> to run the self-tests.  However, it appears that both versions'
>>> self-tests end up hanging; the last in my log is:
>>> 
>>> 0:06:28 load avg: 0.65 [298/403/7] test_shlex
>>> 0:06:28 load avg: 0.65 [299/403/7] test_shutil
>>> 0:06:29 load avg: 0.65 [300/403/7] test_signal
>>> 
>>> and there it sits until it's killed.
>>> 
>>> Anyone have any ideas or (better yet) fixes?
>>> 
>>> Should I just barge ahead and update it anyway?
>> 
>> What OS/version?
> 
> Sorry, should have said that.  NetBSD 7.1_STABLE/i386.
> 
> Regards,
> 
> - Håvard

Looking into that it turns out my NetBSD-current/amd64 also has problem with that.
But, the test works when launched manually:

	python2.7 Lib/test/test_signal.py

Or, when only this particular test is launched using regrtest.py. e.g. in Makefile:
	...
	TESTOPTS= -l $(EXTRATESTOPTS) -m test_signal
	...

You can try this patch:

--- Lib/test/test_signal.py.orig        2018-05-24 08:14:39.956948402 +0000
+++ Lib/test/test_signal.py
@@ -438,6 +438,8 @@ class ItimerTest(unittest.TestCase):
             self.assertRaises(signal.ItimerError,
                               signal.setitimer, signal.ITIMER_REAL, -1)
 
+    @unittest.skipIf(sys.platform.startswith('netbsd'),
+        'waits indefinitely when launched from regrtest.py')
     def test_itimer_real(self):
         self.itimer = signal.ITIMER_REAL
         signal.setitimer(self.itimer, 1.0)

It makes the tests proceed, but then, at least for me, they hang on test_smtpnet (test_sendall_interrupted).

Concluding, I wouldn't bother with regression tests, and do the update. :)

Kind regards,
Adam