This submission comprises the LPSAT LCNF solver and planning problem
compiler. There are two subdirectories of src. The ipp directory
contains a modified version of the IPP compiler (tweaked to compile
metric planning problems rather than plain planning problems and to
produce LCNF rather than CNF statements), some example domains, and
some utilities for printing the solution (which are integrated into
the core executable). The rel_sat directory a modified version of the
RelSAT solver which, among other things, calls the Cassowary linear
constraint solver embedded in the cassowary subdirectory. (Of these,
only Cassowary is called as a purely external library.)

What follows is the standard warning to those who would compile the
LPSAT source code. Good luck!

-----------------------------------------------------------------

Here is the code as it stands. It's not in the very best of conditions
as it hasn't had any care and feeding for a couple of years, but you
can give a shot at it.

I've attached a compiled, gzipped copy of the binary to this file and
a tarred, gzipped copy of the source. Warning: the source is not in
the best of shape to be distributed. To (make a start on) making it,
you'll need to run make in the rel_sat directory (hopefully the
lib-cassowary.a in cassowary/c++ will be fine; otherwise, you'll need
to make that and possibly creader in that directory as well) then in
the ipp directory. That will make the lpsat executable in the ipp
directory. You might also have to hand-make the files in
ipp/print_solution and ipp/stats (g++ -c stats.cc, etc.).

Tacked on is this response from the person I sent that message to:
From: <SNIP>
To: Steve Wolfman <wolf@cs.washington.edu>
Subject: Re: LPSAT source

steve,

after some pain I've managed to recompile lpsat, but in order so to do
I've had to correct what is apparently a syntax error, which puzzles
me:

in directory lpsat/source/rel_sat/cassowary/c++
   in file ClLinearExpression.h
      at row 45

         typedef map<PconstClAbstractVariable,class T>
         ClVarToCoeffMap;

      must be substituted by

         typedef map<PconstClAbstractVariable,T> ClVarToCoeffMap;

...and analogously two rows below. Does this sound reasonable?