Ok, put a fork in me, 'cuz I'm done with this.  The latest is that 
mmap.size() is defined to return the size of the file and not the size 
of the data.  It tries to return it as a ssize_t, which of course, on 
systems where off_t is 64-bits and ssize_t is 32-bits, won't work for 
sizes that won't fit in 32-bits.

Without the size of the data, it is unclear how one would traverse a 
file using the offset parameter.  As part of fixing mmap, I would 
suggest someone should write an example of how it should be used in 
these cases and use that as a test case.

I'm going to leave this up to someone with more knowledge of how this 
stuff *should* work in Python, but I'm going to go redo my program in C.
Thanks for listening and the future efforts.