This is a checkin of the main work I did for my thesis: 
an extension of the Jif compiler (http://www.cs.cornell.edu)
so that it could find information flows in Java programs.  
This information would be outputted to XML, which would then
be processed by an external graph cut program (simp or 
lemon-mincut in my github account).

This work was eventually published in ESOP10 and FSE08.

There's a lot of dead code around -- mostly because the 
environment of graduate school does not encourage best (or 
even good) code practices.  Writing unit tests for a compiler 
that was an extension of a compiler that was itself an extension 
of another compiler also seemed beyond me at the time.  There is 
a regression test suite that used phc, a polyglot-specific test
harness.

There are a few half-baked ideas in here that never ended up 
implemented, or never ended up in a published paper.  I'm not 
sure that increasing code quality would have made my life in 
graduate school any easier, but it probably would have made me 
feel a little better.