preventing this right now is that when i apply this to py3k today, it
fails miserably in a debug build.

first, my patch has an invalid assert(size > 0) in it in _PyLong_New as
0 size is used when the value is 0.  get rid of that line.

then things at least run but you'll end up in an infinite loop when the
interpreter exits at best if you've compiled in debug mode.

things work great in a non-pydebug build.

i believe the reason is this change is not properly looking at the
structure allocation sizes.  debug builds add extra structure fields.

i'm investigating.  the free_list code in floatobject.c does not have
this problem so at least there's a good example to go from.

and yet another reason why a more general free list library for various
internals to use would be useful...