So... why changing?

I'm an IDL user since ... 1992, when I started my PhD thesis. I use it since then, I'm even teaching the use of it some times. I'm not really a "guru", but people in my Institute use to ask me when they have problems.
It's true that I'm using IDL for everything, even for a simple addition ;-) I use to write IDL programs to generate the figures, the LaTeX tables and the main results of my papers and presentations. I'm quite able to do anything I want in IDL, such as string manipulation, reading-writing data, interfacing with other programs. I have a lot of ready-to-use routines that I can apply to new problems. I also wrote 3 big programs using IDL (Cloudy_3D, X-SSN, 3MdB managing tools).
So... why changing in these conditions? Various reasons, without one dominating.

• I have problems since the beginning of my IDL story of using a commercial language. I tried GDL, but some important functions are still not implemented (for example SAVE applied to structure containing pointers to arrays...). It's not a problem in my every day work, I have access to IDL licenses at my Institute. But teaching to students a language that they cannot afford is going against my personal politic...
• It seems that the next generation softwares in my area will by in Python: data processing pipelines (satellite data), data analysing tools (IRAF is now PyRAF), are in Python. Dedicated library specific to my area are now available in Python (e.g. FITS reading tools)
• Python was not so useful for me 5 years ago, as some important functionalities were missing. It changed in the last years and now it seems to me that it's really powerful. Let's see...

The aim of this blog is to share my experience in this learning of Python, first to help me to organize my ideas and secondly to help others to see how hard it is.
Any tip, tool, link is really welcome.