[MERGE] OSX's mkdtemp assigns wrong group by default
John Arbash Meinel
john at arbash-meinel.com
Mon Sep 1 01:33:43 BST 2008
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Vincent Ladeuil wrote:
>>>>>> "john" == John Arbash Meinel <john at arbash-meinel.com> writes:
>
> <snip/>
>
> >>>> Conditional logic in tests ? >-/
>
> john> Well, one of your alternatives was to detect that the
> john> platform is 'darwin' and raise either
> john> UnavailableFeature or TestNotApplicable.
>
> Yes, either we run the test or we don't.
>
> john> That is *still* conditional logic in the test. I
> john> honestly don't see what it is such a big problem. You
> john> are just changing where the logic is.
>
> The difference is that we do something in the test to make it
> pass, kind of cheating.
>
> In that specific case (the OS has a strange idea about how to
> setup a tmp dir) I'm not that worried, yet, rules are rules and I
> like to have a good explanation when I violate one.
>
> Another alternative will be to trap the exception when we can't
> chmod, but that seems a bit late to decide if we should change
> the group...
>
> Anyway, I went the easy way and split the tests.
>
> The remaining failing tests on OSX are all weird and need further
> discussion, let's try to merge this patch first.
>
> Vincent
>
>
You have one too many spaces here:
+++ bzrlib/tests/test_permissions.py 2008-08-31 19:19:08 +0000
...
@@ -138,6 +138,19 @@
t.commit('new c')
check_mode_r(self, '.bzr', 0664, 0775)
+
+ def test_new_files_group_sticky_bit(self):
+ if sys.platform == 'win32':
But otherwise
BB:tweak
John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iD8DBQFIuzhnJdeBCYSNAAMRAuB9AJ9O/+riOCA1mmN6Jo5BwAVbB4kSOQCgw/1g
6+2ll8f704Jk+LKBowI6P80=
=Dec1
-----END PGP SIGNATURE-----
More information about the bazaar
mailing list