Implement stat.st_dev and os.path.samefile on windows

[amauryfa]

BPO	11939
Nosy	@amauryfa, @briancurtin, @serhiy-storchaka
Files	issue11939.diff issue11939_v2.diff issue11939_v3.diff

^{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 = 'https://github.com/briancurtin'
closed_at = <Date 2012-12-26.13:12:24.228>
created_at = <Date 2011-04-27.16:59:42.339>
labels = ['type-feature', 'OS-windows']
title = 'Implement stat.st_dev and os.path.samefile on windows'
updated_at = <Date 2013-01-01.18:32:19.605>
user = 'https://github.com/amauryfa'

bugs.python.org fields:

activity = <Date 2013-01-01.18:32:19.605>
actor = 'brian.curtin'
assignee = 'brian.curtin'
closed = True
closed_date = <Date 2012-12-26.13:12:24.228>
closer = 'brian.curtin'
components = ['Windows']
creation = <Date 2011-04-27.16:59:42.339>
creator = 'amaury.forgeotdarc'
dependencies = []
files = ['28423', '28429', '28430']
hgrepos = []
issue_num = 11939
keywords = ['patch', 'needs review']
message_count = 20.0
messages = ['134595', '134602', '178090', '178101', '178102', '178103', '178104', '178105', '178108', '178206', '178207', '178208', '178211', '178218', '178220', '178712', '178727', '178739', '178740', '178742']
nosy_count = 4.0
nosy_names = ['amaury.forgeotdarc', 'brian.curtin', 'python-dev', 'serhiy.storchaka']
pr_nums = []
priority = 'normal'
resolution = 'fixed'
stage = 'resolved'
status = 'closed'
superseder = None
type = 'enhancement'
url = 'https://bugs.python.org/issue11939'
versions = ['Python 3.4']

[amauryfa]

Since 9cd1036455e7, os.stat() on Windows fills the st_ino member;
it could fill st_dev just as easily;
then os.path.samefile() could be shared with the posix implementation.

[briancurtin]

I created/assigned bpo-10646 to myself for other samefile issues - I can cover this as well unless someone beats me to it.

[briancurtin]

Here's a patch that fills st_dev, and while we're at it st_rdev (which is the same value).

I've moved the implementation of samefile/sameopenfile/samestat from Lib/posixpath.py over to Lib/genericpath.py and then removed the implementation from Lib/ntpath.py, so those functions are now common. The same goes for tests - I've rearranged tests towards test_genericpath. I also removed _getfileinformation from Modules/posixmodule.c because it's no longer being used.

[serhiy-storchaka]

What about macpath? I think test_macpath will fail.

[briancurtin]

Why do you think that? I don't have a mac so I can't test it.

[briancurtin]

Here is an updated patch addressing the sameopenfile that remained in Lib/ntpath.py, thanks to Sehriy's comment on the review.

[serhiy-storchaka]

Hmm. test_macpath will now run the tests for samefile etc. macpath doesn't contain os.path.samefile. But in Modules/posixmodule.c I don't see any special cases for classic MacOS. Actually, I don't know how tests behave on classic MacOS.

Please update the documentation (availability, versionchanged).

[serhiy-storchaka]

On other hand, you should add "samefile", "sameopenfile" and "samestat" back to __all__ in posixpath and ntpath. __all__ is a list of exported names, not a list of defined names.

[briancurtin]

Docs and the __all__ changes in V3 patch.

[python-dev]

New changeset 189b21f9bc0c by Brian Curtin in branch 'default':
Fix bpo-11939. Set st_dev attribute on Windows to simplify os.path.samefile.
http://hg.python.org/cpython/rev/189b21f9bc0c

[python-dev]

New changeset 82531b78b719 by Brian Curtin in branch 'default':
Add NEWS entry for fixing bpo-11939
http://hg.python.org/cpython/rev/82531b78b719

[briancurtin]

Thanks for the reviews.

[python-dev]

New changeset 9e980454b85e by Brian Curtin in branch 'default':
Add tests for Issue bpo-10646.
http://hg.python.org/cpython/rev/9e980454b85e

[serhiy-storchaka]

Try on Windows:

>>> import os
>>> from os.path import *
>>> samestat(os.stat('.'), os.stat('.'))
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
NameError: name 'samestat' is not defined

[briancurtin]

Separate issue. Fixed in bpo-16788.

[python-dev]

New changeset 3738d270c54a by Brian Curtin in branch 'default':
st_dev/st_rdev should be unsigned long as dwVolumeSerialNumber, which it is set to, is a DWORD. This was fixed in bpo-11939 and the overflow was mentioned in bpo-10657 and seen by me on some machines.
http://hg.python.org/cpython/rev/3738d270c54a

[serhiy-storchaka]

New changeset 3738d270c54a by Brian Curtin in branch 'default':
st_dev/st_rdev should be unsigned long as dwVolumeSerialNumber, which it is
set to, is a DWORD. This was fixed in bpo-11939 and the overflow was
mentioned in bpo-10657 and seen by me on some machines.
http://hg.python.org/cpython/rev/3738d270c54a

But than st_dev used as long in _pystat_fromstructstat(). Perhaps you should
check if st_dev is negative and then select
PyLong_FromLong/PyLong_FromLongLong or
PyLong_FromUnsignedLong/PyLong_FromUnsignedLongLong.

[python-dev]

New changeset 61bada808b34 by Brian Curtin in branch 'default':
Set st_dev on Windows as unsigned long to match its DWORD type, related to the change to fix bpo-11939.
http://hg.python.org/cpython/rev/61bada808b34

[serhiy-storchaka]

Now the code is wrong on non-Windows without PY_LONG_LONG and with signed st_dev.

[briancurtin]

Backed out the changeset. If you have a solution, feel free to fix it.