Issue2274

Depending on input, `os.path.normpath` turns Unicode strings to byte strings:

Jython 2.7b4 (default:3672e624962a, Feb 13 2015, 04:59:14) 
[Java HotSpot(TM) 64-Bit Server VM (Oracle Corporation)] on java1.7.0_60
Type "help", "copyright", "credits" or "license" for more information.
>>> from os.path import normpath
>>> normpath(u'.')
'.'
>>> normpath(u'/')
'/'
>>> normpath(u'..')
u'..'
>>> normpath(u'./')
'.'

How severe this is depends on do we consider `os.path.abspath` turning byte strings to Unicode a bug or not (I'm in directory `böö` in these examples):

>>> abspath(u'.')
u'/home/peke/b\xf6\xf6'
>>> abspath('.')
u'/home/peke/b\xf6\xf6'
>>>

With CPython the latter returns byte string. Jython nowadays handles environment variables and such as Unicode, so perhaps Unicode is fine here too.

If we agree abspath always returning Unicode is fine, normpath possibly returning bytes is just a minor inconsistency. If abspath should return bytes or Unicode depending on the type of the input, like CPython does, there would be problems with `abspath(normpath(u'.'))` if the cwd would contain non-ASCII characters.

This was a bug in CPython (http://bugs.python.org/issue5827), and can be readily be fixed by updating the normpath function (written in Python) to the latest in CPython 2.7 posixpath.py. It has already been updated in Jython's version of ntpath.py

We should also update the corresponding tests

Fixed by removing Jython-specific test_posixpath, posixpath, and minor update to pwd - always the best sort of fix, it means underlying Jython runtime is even more compatible

https://hg.python.org/jython/rev/8c1a3b72913e