On Windows, Py_UNICODE is wchar_t, so we can avoid the expensive Py_UNICODE*
=> wchar_t* conversion.

 * Don't define _Py_wstat() on Windows, Windows has its own _wstat() function
   with a different API (the stat buffer has another type)
 * Include windows.h

I suppose that I reapplied my local patch creating Python/fileutils.c whereas
the file already existed.

 * _Py_fopen() and _Py_stat() come from Python/import.c
 * (_Py)_wrealpath() comes from Python/sysmodule.c
 * _Py_char2wchar(), _Py_wchar2char() and _Py_wfopen() come from Modules/main.c
 * (_Py)_wstat(), (_Py)_wgetcwd(), _Py_wreadlink() come from Modules/getpath.c

Use _Py_wchar2char() to support surrogate characters in the input path.

This undoes a previous attempt to fix some of the skips. Value and copy
come from sharedctypes rather than plain ctypes, and the test skipping is
then done at the class level rather than test-by-test.

There are zero skipped tests after the fix.

All unicode functions uses PyObject* except PyUnicode_AsWideChar(). Fix the
prototype for the new function PyUnicode_AsWideCharString().

 * Create format_arg variable to use the right types
 * Strip trailing spaces

 * fix argv0 reference counter if PyList_SetItem() fails
 * don't use complex if conditions, but a simple indentation and "goto error"
 * simplify error handling (remove Py_XDECREF(importer) from the error label)
 * don't set sys_path to NULL (it's useless, sys_path is a borrowed reference
   and sys_path is not a static variable)
 * try to write only one instruction per line for better readability

Create sys_update_path() static function. Do nothing if argc==0.

after the file is closed.  Patch by Jeffrey Finkelstein.

Fix the suspicious builder not to write CRLF, update suspicious file and add instructions to Makefile.

This should fix some buildbot failures.

Fix Łukasz’ name and also remove trailing whitespace.  See thread at
http://mail.python.org/pipermail/python-dev/2010-September/103584.html

uname and gname field.

If tarfile creates a new archive and adds a file with a
uid/gid that doesn't have a corresponding name on the
system (e.g. because the user/group account was deleted) it
uses the empty string in the uname/gname field now instead
of "root". Using "root" as the default was a bad idea
because on extraction the uname/gname fields are supposed
to override the uid/gid fields. So, all archive members
with nameless uids/gids belonged to the root user after
extraction.

Fix Issue10012 - httplib headers, which are (sometimes mistakenly) int are explicitly cast to str (bytes - in py3k).