test_mv_change_case_dir fails on OS X 10.4.11
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Bazaar |
Fix Released
|
Undecided
|
Vincent Ladeuil |
Bug Description
make check ends with:
blackbox.
Unexpected return code
not equal:
a = 0
b = 3
=======
FAIL: test_mv_
vvvv[log from bzrlib.
633.611 created control directory in file://
633.642 creating repository in file://
633.678 creating branch <bzrlib.
633.836 trying to create missing lock '/private/
633.852 opening working tree '/private/
633.992 run bzr: ['mv', 'foo', 'Foo']
633.992 bzr arguments: ['mv', 'foo', 'Foo']
633.995 encoding stdout as sys.stdout encoding 'US-ASCII'
634.015 opening working tree '/private/
634.053 Traceback (most recent call last):
File "/Users/
return run_bzr(argv)
File "/Users/
ret = run(*run_argv)
File "/Users/
return self.run(
File "/Users/
self._run(tree, names_list, rel_names, after)
File "/Users/
tree.
File "/Users/
return unbound(self, *args, **kwargs)
File "/Users/
WorkingTree
File "/Users/
return unbound(self, *args, **kwargs)
File "/Users/
rename_entries = self._determine
File "/Users/
raise errors.
RenameFailedFil
634.055 errors:
'bzr: ERROR: Could not rename foo => Foo because both files exist. (Use --after to tell bzr about a rename that has already happened)\n'
634.101 opening working tree '/tmp/testbzr-
^^^^[log from bzrlib.
-------
Traceback (most recent call last):
File "/Users/
self.
AssertionError: Unexpected return code
not equal:
a = 0
b = 3
-------
Ran 342 tests in 597.724s
FAILED (failures=1, known_failure_
Missing feature 'Internally performed glob expansion' skipped 2 tests.
tests failed
bzrlib.
make: *** [check] Error 1
Note: the standard Mac OS X file system is case insensitive. It is in general not possible to directly mv foo to Foo. One has to move foo to foo.tmp and then move foo.tmp to Foo or use some other work around. I don't know how the test is done but it seems more likely that one file is being recognised twice than that "both files exist".
I'm wondering if this test should at least be flagged as an expected failure on case insensitive file systems.
Actually, I see it is meant to work around this, but apparently isn't...