This is from the sparc Unbuntu 3.0 bot. I'm having trouble reproducing
it, though.

test test_smtplib produced unexpected output:
**********************************************************************
*** line 2 of actual output doesn't appear in expected output after line 1:
+ error: uncaptured python exception, closing channel
<test.test_smtplib.SimSMTPChannel 127.0.0.1:38832 at 0x268d0a8> (<class
'socket.error'>:[Errno 9] Bad file descriptor
[/home/pybot/buildarea/3.0.klose-ubuntu-sparc/build/Lib/asyncore.py|readwrite|100]
[/home/pybot/buildarea/3.0.klose-ubuntu-sparc/build/Lib/asyncore.py|handle_write_event|426])
**********************************************************************

The failure happens even when test_smtplib runs before both
test_asyncore and test_asynchat (*). So the latter can't be the culprit.


(*)
http://www.python.org/dev/buildbot/3.0.stable/sparc%20Debian%203.0/builds/419/step-test/0

Adding this:

    def handle_error(self):
        raise

...to SimSMTPChannel class would help to provide a clearer error message to understand where exactly EBADF comes from, altough I think this was an old asyncore bug which have already been fixed.
Is this still reproducible on Python 3.2?

If it would provide clearer error message in cases of failure, it sounds like a reasonable thing to add whether or not the failure in this issue is still happening.

Done in r80882 for ftplib and smtplib modules.

Looking back over the logs (via bbreport) I don't see any failures of test_smtplib that match the report in this issue.  There were three in the last while, two of which look like network issues on the same machine (and thus probably not a test problem per se) and one of which (solaris) had multiple errors but succeeded when rerun immediately.  

While there maybe a remaining flakiness issue with test_smtplib, it doesn't seem to be the error described here, and I don't think it rates critical.

Can this be closed as "out of date"?