[Toolserver-l] Re. national characters in database fields

[toolserver-l.wikipedia.org at publi.purodha.net]

I am having a problem similar to Mashiah Davisons.

I am Using http://vandyke.com/ SecureCRT which usually worked fine
for me. (I would like to give putty a test run, but it does not log
me in, see separate message, "Key poblem logging in with putty")

My problems are currently related to interwiki.py - a python
program dealing with links between languages.
It is:
A - reporting page titles in various languages/scripts, and
B - requiring command line parameters, and interactive responses,
     in UTF-8 encoding

Problem Group A - messages:
---------------------------
Messages do strange things to the terminal emulation, e.g. they
display garbage, scroll backwards, suppress output, trigger an
automated response (you can never type a correct answer to a
prompt, when the backspace key has been redefined at the same time,
too)  etc. 
I tried every terminal emulation, SecureSCT has to offer, I tried
with and without UTF-8 enabled. Results varied slightly, but the
general behaviour remains - it's not usable.

Solution 1:
----------
I pipe messages into a file which I read over the web with a
web brouser.
When the web server is instructed to tell that it sends UTF-8,
everything is marvellous.
Drawbacks: huge overhead, rather time consuming, pretty unsuited for
hihgly interactive tasks. While I can read <g!> every foreign
script, and copy it locally, I cannot paste it back to the
toolserver.

Solution 2:
----------
I tried to alter Linux terminal type settings, and other such
settings. I have not found any of general usefulness, but some
allow 'some' operability at times.
My knowledge and experience in this field is, however, very
limited. I would like to find a reference + tutorial on
character encodig issues and settings under Linux / bash,
so as to understand the logic at work,
but could not find any yet.
(-: neither logic, nor tutorial, that is :-)

Problem Group B - interactive input, and command lines.
-------------------------------------------------------
In short: with the SecureCRT Terminal set to UTF-8, you can
successfull copy and paste "some" (i.e. few) national characters
into command lines (luckily, the lower case German Umlauts are
among them, which I need often) but you cannot use bashs line
editing tool when you have exotic characters in a line buffer.

Everything else usually does not work; results are unpredictable.

Solution 0:
Avoid interactivity entierly. Find the needed strings somewhere,
and pipe them into files where they do what you need. When editing
is unavoidable, scp your data elsewhere, edit e.g. under Windows,
which is bad enough, and scp the result back.

Obviously, this is not a solution.

Summary:
--------
:-(

Purodha