What can PyPy do for you?
=========================

Peek under the hood
-------------------

 - sys.pypy_repr() with small ints::

    >>>> import sys
    >>>> sys.pypy_repr(1)
    '<unboxed 1>'
    >>>> sys.pypy_repr(42)
    '<unboxed 42>'
    >>>> sys.pypy_repr(sys.maxint)
    '<W_IntObject object>'
    >>>> sys.pypy_repr(sys.maxint / 2)
    '<unboxed 1073741823>'
    >>>> sys.pypy_repr(sys.maxint / 2 + 1)
    '<W_IntObject object>'

Stackless stuff
---------------

 - lightweight threadlets similar to what stackless python provides

 - tasklet cloning

 - gc: performance/memory/feature tradeoffs: e.g. (check if and) show an
   example where a refcounting pypy uses less Unix process memory than a Boehm
   one (but of course the latter is faster).  Mention that tasklet cloning
   requires a gc which is (ATM) a bit slower.
 
Lazy stuff
----------
 
 - lazy evaluation with the thunk object space
   try running demo/sharedref on pypy-c-thunk.
 
 - Oz-like dataflow variables with the logic object space
   py.py -o logic --usemodules=_stackless fibonacci{,2,3,4}.py

Compiling
---------

 - algo example, e.g. sieve of eratosthenes
 - ctypes example
 - Writing compatible extension modules for PyPy and CPython with the PyPy
   extension compiler

 - The Javascript backend and its uses