Unicode problem in Tkinter under Windows 2000
Keyboard-entered chars in ascii range > 128 mess up 
internal unicode encoding in text-widget leading to 
unicode errors

The following example should reproduce the bug:

>>> import Tkinter
>>> t=Tkinter.Text()
>>> t.pack()
>>> t.insert("1.0",u'\xe2\xee\xfb')

Now set the focus to the text-widget and via the 
keyboard enter an a umlaut into the text-widget 
(alternatively press ALT and enter 0228 on the Numpad 
of your Keyboard to simulate this)
Then test the result:
>>> t.get("1.0","end")
u'\xe2\xee\xfb\xe4\n'
This is what you get under Linux (I was told) and what 
it should be.
However, under Windows 2000 I get:
'\xc3\xa2\xc3\xae\xc3\xbb\xe4\n'
which is a mixture of UTF-8 and cp1252(?) leading to 
an Unicode-error, if I try e.g. to save it as a file.
(All characters of an 8-bit value > 128 (e.g. latin-1 
or cp1252) entered via keyboard into a text-widget 
cause such a weird behaviour, not just the a umlaut.)

A simple workaround (not thoroughly tested) could look 
like this:

def badkey(self, event):
    try:
        if ord(event.char) > 127:
            txt.insert("insert", unicode
(event.char,"cp1252"))
            return "break"
    except:
        pass

`txt` being the instance of a text-widget, that is 
bound to a callback for the key-press-event:

if sys.platform == "win32":
         txt.bind("<KeyPress>",badkey)