this blog is related to OpenOpt module for scikits (see scipy.org) - free Python-based optimization alternative to TOMLAB, AMPL, GAMS etc.
I had some experience with OpenOpt for MATLAB/Octave before. You can take a look at reviews and download that one from here
Now Python version includes 2 non-smooth UC solvers: ralg (an implementation of Naum Z.Shor r-alg) and ShorEllipsoid (Naum Z.Shor modificated method of ellipsoids). Former is intended for medium-scale problems and is used very intensively in lots software solutions of our non-smooth optimization department, cybernetics institute, Ukrainian science academy; latter is very special solver for small-scale problems only (and knowing r0 required), I add the one only because it's very simple to implement.
There is basic graphic output implemented (provided you have matplotlib installed), see example.py
Also, connections to some LP solvers have been made, see exampleLP.py.
I shall connect openopt to scikits svn during 1-2 days (I need to decide how to split all my files into subdirectories very carefully and learn how to use svn - previously I used only cvs). Currently you can download the OpenOpt for Python from here