HAVE_FTRUNCATE because ftruncate does not exist on Windows.

self.assertEqual(x, y).  (Christian used self.failUnlessEqual(),
but the double negative makes it hard to grok, so I changed it.)

This bug should be fixed in trunk and release25-maint also.

to 'file', and fix a bunch of subtleties in the behavior of objects
related to overriding __str__.  Also disabled a few tests that I couldn't
see how to fix but that seemed to be checking silly stuff only.

(somehow OSX marks I/O devices as seekable).

a text string instead of bytes.  (This being hex digits it *should*
be a text string -- that it wasn't before was a misunderstanding on
my part of the C code.)

caused by not closing a file.

restoring a disabled test.
This was necessary to make test_pickle.py pass.

Also fix mac toolbox glue to accept str, str8, bytes for
255-byte strings.

to even the most basic file object (I also added readall() which may
be a better API).  Also, not all the tests requiring specific failure
modes could be saved.  And there were the usual str/bytes issues.
I made sure test_io.py still passes (io.py is now most thoroughly
tested by combining test_file.py and test_io.py).

attempts to delete all modules).  This makes test_subprocess work again.
(I can't quite figure out why stdout/stderr don't get deleted properly,
which would flush them anyway, but that's a separate issue.)

Change the UUID properties to use @property.
Change the UUID API so that .bytes and .bytes_le return, and the
corresponding arguments to __init__() require, values of type 'bytes'.

However, test_ctypes is still broken -- and apparently more than before.

the problem was writing a file in text mode instead of in binary mode.

- Use os.urandom() as the only source of random bytes.
- Don't reference socket._fileobject; use io.BufferedReader instead.

should treat the bytes as it treats a string.
Not doing this broke re.compile() of big charsets.