[issue10512] regrtest ResourceWarning - unclosed sockets and files

Brett Cannon added the comment: r88496 for 3.3 r88497 for 3.2 -- resolution: -> fixed stage: -> committed/rejected status: open -> closed ___ Python tracker ___ __

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: No objections here. The other remaining leaks that I am aware of are covered by issue10883. -- ___ Python tracker ___ __

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Brett Cannon added the comment: Once Python 3.3 is open I will apply the cgi fix. Just to double-check, can I close this issue once the test_cgi patch goes in? -- assignee: georg.brandl -> brett.cannon ___ Python tracker

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: Looks good to me. My earlier patch was more defensive because I wasn't sure whether any of the other tests might be using cgi.log(), but it seems that this isn't the case. -- ___ Python tracker

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Brett Cannon added the comment: Attached is a simpler fix for test_cgi so it can get in for Python 3.2. You can reproduce the failure if you run ``./python -W error -m test test_cgi``. Georg, can I commit? -- assignee: -> georg.brandl nosy: +brett.cannon, georg.brandl Added file: htt

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: Good idea; they look like more work to fix than the warnings so far. Aside from those two, it looks like test_cgi is all that's left. Just to clarify, did you manage to reproduce the test_cgi warning? -- ___ Python t

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: I opened a separated issue for test_urllib and test_urllib2net: #10883. -- ___ Python tracker ___ _

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Changes by Éric Araujo : -- nosy: +eric.araujo ___ Python tracker ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.pyth

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: Sorry, scratch that - I misunderstood the semantics of SocketIO.close(). I hadn't realized that the underlying socket is supposed to stay open until it itself is also explicitly closed (as well as all SocketIO objects referring to it). I've been able to get ri

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: Le vendredi 07 janvier 2011 à 02:19 +, Nadeem Vawda a écrit : > Most of these leaks seem to stem from the fact that socket.SocketIO.close() > doesn't behave as documented. According to its docstring, it is meant to > decrement the underlying socket's refco

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: Looking at the warnings from test_urllib2net, it seems that they all originate in the FTP tests: * OtherNetworkTests.test_ftp() * TimeoutTest.test_ftp_basic() * TimeoutTest.test_ftp_default_timeout() Most of these leaks seem to stem from the fact that socket.Soc

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: Have you tried my patch (resourcewarning-fixes-3.diff)? It fixes the warning for me without breaking anything. I was just worried that the warning was something triggered by my specific system configuration when you said that you couldn't reproduce it. I was tr

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: test___all__ just loads ALL modules... and it doesn't unload them. I patched test___all__ to unload modules: the "ResourceWarning: unclosed file ... '/dev/null' ..." disappears, but a new error occurs. The multiprocessing module registers _exit_function() in

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: Awesome. That just leaves test_urllibnet, test_urllib2net, and test_cgi. I'm hoping to post patches for the first two tomorrow. About test_cgi, I've fiddled around with it a bit more. The leak manifests itself with any set of tests including test_cgi and test__

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: > Fix attached for test_imaplib Oh thanks! Commited as r8 and r87778. I just changed the name of the subfunction. -- ___ Python tracker _

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: Fix attached for test_imaplib. Most of the warnings were simply due to reap_server() not closing the server object correctly. The remaining warning was due a genuine leak in imaplib.IMAP4.__init__() - if an exception is raised after the connection is opened, th

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: > r87736 introduces another DeprecationError; this time in test_time (line 150; > s/assertEquals/assertEqual/). Fixed by r87759. -- ___ Python tracker __

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: > r87710 introduces a ResourceWarning in test_threading. Fix attached. Fixed by r87757 (I wrote a different patch). -- ___ Python tracker ___

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: r87736 introduces another DeprecationError; this time in test_time (line 150; s/assertEquals/assertEqual/). -- ___ Python tracker ___ __

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Changes by Nadeem Vawda : Added file: http://bugs.python.org/file20255/test_threading.diff ___ Python tracker ___ ___ Python-bugs-list mailing

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: r87710 introduces a ResourceWarning in test_threading. Fix attached. -- ___ Python tracker ___ ___ Py

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: DeprecationWarnings: * test_unittest: fixed by r87717 * test_array: fixed by r87719 * test_httplib: fixed by r87720 -- ___ Python tracker ___ _

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: The fix for test_normalization was committed as r87441. As for test_cgi, I still seem to get the leak (also on Linux; Ubuntu 10.10 64-bit). I'll poke around with it some more tomorrow. In addition to the ResourceWarnings, some of tests have been raising Deprec

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: * r87686 fixes multiprocessing * r87687 fixes pydoc * r87688 fixes test_subprocess Remaining ResourceWarning warnings: * test_imaplib * test_urllibnet * test_urllib2net -- ___ Python tracker

[issue10512] regrtest ResourceWarning - unclosed sockets and files

STINNER Victor added the comment: * r87680 fixes test_sockserver * r87681 fixes test_timeout * r87682 fixes test_tk * r87683 fixes test_xmlrpc * r87684 fixes test_socket r87682, r87683, r87684 are patches from Nadeem Vawda. On my Linux box, I am unable to get the warning on test_cgi or te

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: test_cgi causes a strange filehandle leak that only causes a warning when regrtest terminates, and for some reason doesn't show up if you run just test_cgi by itself. I've attached a patch that closes the filehandle. -- nosy: +lukasz.langa Added file:

[issue10512] regrtest ResourceWarning - unclosed sockets and files

Nadeem Vawda added the comment: Attached is a patch that fixes the warnings in test_xmlrpc, along with some other file- and socket-related warnings in test_normalization, test_timeout and test_tk that only show up when regrtest is run with -uall. The warning in test_timeout could be fixed wit