No, it's not easy, but neither is garbage collection, good optimizations, a clean interface, helpful documentation, or anything that makes a language really great. If I'm going to keep using Python in the future, I'm going to want threading, and right now it sucks.
I wonder how nasty it would be to do a new implementation of Python in C/C++. Maybe one that doesn't have the GIL, has a proper exact GC, and emits LLVM or something. PyPy seems to be addressing some of these issues.
I wonder what the GIL is actually being used for... I don't see why the interpreter needs so much damn locking.
What kind of nasty are you thinking about?
I just mean implementation difficulty to do a clean, efficient, extensible new implementation.
The 'open letter' whiner got more of a response than he deserved. I'm actually surprised at Guido.