When a thread touches such an object for the first time, a new thread-local __dict__ is created,
and the __init__ method is run.
But a thread switch can occur here; if the other thread touches the same object, it installs another
__dict__; when the first thread resumes, it updates the dictionary of the second...
This is the deep cause of the failures in test_multiprocessing involving "managers" objects.
Also a 2.5 backport candidate.
FreeBSD's P1003.1b semaphore support is highly experimental and
it's disabled by default. Even if a user loads the experimental
kernel module manually, _multiprocessing doesn't work correctly due
to several known incompatibilities around sem_unlink and sem_getvalue,
yet.
svn+ssh://pythondev@svn.python.org/python/branches/tnelson-trunk-bsddb-47-upgrade
........
r64368 | trent.nelson | 2008-06-17 23:13:44 -0500 (Tue, 17 Jun 2008) | 1 line
Initial commit of work pertaining to switching the Windows build from Berkeley DB 4.4.20 to 4.7.25. Note that I've deprecated the standalone '_bsddb44.vcproj' in lieu of adding the sources in a separate folder to the _bsddb project. This was a conscious decision and actually makes everything far more easier to manage. With this approach, entire test suite passed straight off the bat. Well, almost -- the timeout in bsddb/test/test_replication.py needed bumping up a little -- 2 seconds was too short. 10 seconds seems to be fine for me, but I'll make sure Jesus verifies. More documentation to come once I've been able to test out this approach on the buildbots (hence keeping the changes in a separate branch for now).
........
r64369 | trent.nelson | 2008-06-17 23:19:12 -0500 (Tue, 17 Jun 2008) | 1 line
Bump Berkeley DB version from 4.4.20 to 4.7.25.
........
bus errors or SystemError being raised. As a side effect of fixing this, a bad
DECREF that could be triggered when 'message' and 'category' were both None was
fixed.
Closes issue 3211. Thanks JP Calderone for the bug report.
behaviours. I left the original test commented out (note
that that test came from #2702, which seems to have a
problem in FreeBSD and Windows, but not in Linux).
I included a new test, to watch over the now-broken
behaviour, I took it from #3179.