Description:
when using both the tkinter GUI and the locale.setlocale
(locale.LC_ALL, '') call on Py2.4, the following error is 
printed to stderr:

<pre>error reading package index 
file /usr/share/tcl/vfs1.3/pkgIndex.tcl:
syntax error in expression "[package provide Tcl] < 8.4": 
extra tokens at
end of expression
</pre>

in addition to this, the BackSpace key appears not to be 
handled by the tk input widgets (for example "Entry") 
after the setlocale call.

in Py2.4.1 the error is no longer printed, though the 
problem with the BS key persists.


Testcase: 
see attached file.


---
Originally posted to ALTLinux bugtracker (original text 
(Russian): https://bugzilla.altlinux.org/show_bug.cgi?
id=6339)

---
additionally tested on:
Py2.4.1 (builds for win32, cygwin, debian linux)

I can confirm that this bug is still present. After
locale.setlocale(locale.LC_ALL, '') Backspace in Tkinter.Entry is not
working anymore. 
There is no difference if the Backspace is issued by the keyboard or by
self.master.winfo_toplevel().event_generate('<BackSpace>')

No messages are written.
Tested on Linux (Ubuntu Gutsy)/Python 2.5.1 and Windows XP/Python 2.4.3

Attached a workaround for this problem, based on:
http://ml.osdir.com/games.mud.client.lyntin/2005-03/msg00005.html

I also found that the problem only appears when the LC_NUMERIC setting
is different to en_US. (for example if it is de_AT)

The problem with the backspace key was fixed in python 2.6 when using 
the windows installer at least.
I didn't check yet if it is specific to tcl/tk or if it involves 
tkinter.

The backspace issue seems to have been fixed.

I modified bug.py for 3.1 by removing the coding cookie, changing unicode("Ïðîáà", "KOI8-R") to just "Ïðîáà", and fixing print. Running from IDLE, it ran and displayed a tk window with text entry and labelled button that printed "Friday" in the shell window.

On the other hand, "I also found that the problem only appears when the LC_NUMERIC setting is different to en_US. (for example if it is de_AT)" suggests that other might have a problem (I am in the US).

Sidenote: Since the obscure koi8-R encoding (no longer valid in 3.x, the interpreter said) is not described as part of the bug, it should not be part of a minimal example.