test_os failing with errno 61: No Data Available

[beardedp]

BPO	14358
Nosy	@benjaminp
Files	test_os_support_ext.patch test_os_support_ext_return_false.patch

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = <Date 2012-03-19.02:26:16.941>
created_at = <Date 2012-03-18.07:16:51.769>
labels = ['tests', 'type-crash']
title = 'test_os failing with errno 61: No Data Available'
updated_at = <Date 2012-03-19.02:26:16.935>
user = 'https://bugs.python.org/beardedp'

bugs.python.org fields:

activity = <Date 2012-03-19.02:26:16.935>
actor = 'python-dev'
assignee = 'none'
closed = True
closed_date = <Date 2012-03-19.02:26:16.941>
closer = 'python-dev'
components = ['Tests']
creation = <Date 2012-03-18.07:16:51.769>
creator = 'beardedp'
dependencies = []
files = ['24921', '24934']
hgrepos = []
issue_num = 14358
keywords = ['patch']
message_count = 4.0
messages = ['156232', '156233', '156303', '156307']
nosy_count = 4.0
nosy_names = ['benjamin.peterson', 'beardedp', 'rosslagerwall', 'python-dev']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'crash'
url = 'https://bugs.python.org/issue14358'
versions = ['Python 3.3', 'Python 3.4']

[beardedp]

When running the test suite on Linux 3.2.9-1, I get the following error on the test_os suite:

test test_os crashed -- Traceback (most recent call last):
  File "/home/benhayden/Documents/cpython/Lib/test/regrtest.py", line 1229, in runtest_inner
    the_package = __import__(abstest, globals(), locals(), [])
  File "/home/benhayden/Documents/cpython/Lib/test/test_os.py", line 1767, in <module>
    @unittest.skipUnless(supports_extended_attributes(),
  File "/home/benhayden/Documents/cpython/Lib/test/test_os.py", line 1754, in supports_extended_attributes
    os.fsetxattr(fp.fileno(), b"user.test", b"")
OSError: [Errno 61] No data available

I don't know the best way to handle this error, but instead of raising it, I just let the decorator return False. This allowed test_os to finish (and finish successfully) on my machine.

Once again, I don't really know what ENODATA might mean, so I just tried the first thing that worked for me.

[rosslagerwall]

Perhaps any errors that occur during supports_extended_attributes should cause it to just return false?

[beardedp]

That was my thought, but again, I didn't really know if it was 'safe' to snuff the OSError all together. This patch just returns False if any OSError is raised.

[python-dev]

New changeset ada766b07686 by Benjamin Peterson in branch 'default':
all OSErrors should indicate there are no extended attributes (closes bpo-14358)
http://hg.python.org/cpython/rev/ada766b07686