[merge] follow up, fix the rest of the mac tests

[John Arbash Meinel]

Robey Pointer wrote:
> 
> On 3 Jul 2006, at 11:48, John Arbash Meinel wrote:
> 
>> I went ahead and fixed up the rest of the mac tests. It turned out the
>> only thing left was to write a new getcwd() for Mac, which returns the
>> normalized path. And then track down WorkingTree, since it was directly
>> calling os.getcwdu() instead of osutils.getcwd().
>>
>> Anyway, combining the previous patch with this patch means all tests
>> pass on Mac.
> 
> The patches don't quite cleanly apply to bzr.dev anymore, so I may have
> messed up in applying them, but this *does* fix all but 2 tests on my mbp*:
> 

The lsprof test is fixed by Martin's pending lsprof fix.

And I did catch the EINVAL bug, but after I posted. So:

http://bzr.arbash-meinel.com/branches/bzr/mac

Only has the lsprof test failing, because I'm waiting for Martin to
merge it.


...

> 
> (22 is EINVAL.  This is because on osx, rename returns EINVAL when:
>      [EINVAL]           From is a parent directory of to, or an attempt is
>                         made to rename `.' or `..'.
> Changing the test to use 'x' instead of '.' fixes it; I just tried.)

Well, considering the test is explicitly testing that rename_cwd raise
ENOENT when trying to rename from a missing source location, I did need
to fix it. :)

> 
> The lsprof test fails because the expected string has "*" and stuff in
> it, which lsprof doesn't actually put in the output.  Probably the test
> is just wrong, not specific to osx.
> 
> robey
> 
> 
> * By mbp I mean "macbook pro", not our illustrious leader. :)
> 

It sounds like you're seeing the same things I was. I'm glad, because
win32 is being a real pain.

John
=:->