In my latest project I've been messing around with pygame, which is a Python wrapper and extension for the SDL media library. It is great because it adds all sorts of object-oriented stuff on top of the original SDL functionality, without really sacrificing anything (at least as far as I've noticed).
You may wonder why I chose Python for writing a game over C or C++. Well to be honest, it was mainly just to learn Python, and Python just so happened to have this gaming library available. However the more I use it, I think it might actually be a viable strategy for simpler games. If something somewhere is going too slow or using too much memory, you can always just write a part in C++ using Boost.Python and import it in - I'll be honest, I haven't tried Boost.Python so I don't know if it is as seamless as they claim, there could be more on this later.
Python offers much faster development time over C++, I noticed this right away. I can add things very easily that weren't originally in the plan. If I have a bunch of classes with similar functionality, I can quickly refactor them to inherit from a base class which provides the shared functionality - while I can probably do this in C++ too, I'd have to go through the code everywhere and use references/pointers to base class objects instead of the original - bit of a pain.
You may also now wonder why I didn't write it in Ruby. Well, like I said, wanted to learn Python. At this point, I still want to learn Python - I haven't been turned off it yet like I was with Haskell or Scala either (not sure why I was turned off of these, maybe because I'm too dumb or entrenched in my ways to really grasp their more advanced concepts). Also, Ruby is nice and all, but it's what I use at work. I'm happy about that, I'd rather be doing Ruby than PHP or Java, but I'd like a little bit of change when I go home and do my own coding. Choosing another language helps satisfy that desire for change.