Release 0. (HOL only)
---------------------------------------------------------------------------
Release 1. (HOL only.)
              Patches and Bugfixes to Previous Release

* Several bugs in the rewriter have been fixed. The result is that
  normal rewriting, as well as termination-condition extraction, is
  faster and apt to return correct answers in more situations. In particular,
  TC extraction for nested recursions involving "let" was faulty.

* Michael Norrish pointed out a non-portabilism in the source: in rw.sml, I 
  was using the identifier "Term" as a term parser. This has been fixed.

* The termination relation for the "unify" example is more apt to be 
  correct.

* I added a new example - the "select" algorithm. This is fairly demanding on
  the machinery of TFL, so I included it even though I haven't proved that it 
  terminates, or that it is correct.


---------------------------------------------------------------------------
Release 2. This was an Isabelle version of release 1. It is noteworthy
because it has a complete and new proof of the Manna/Waldinger/Paulson
unification algorithm. 
---------------------------------------------------------------------------
Release 3.  This was an improved Isabelle release. In particular, it now 
builds on PolyML.
---------------------------------------------------------------------------
Release 4 (HOL and Isabelle)

* The expressiveness of patterns has been improved. Incomplete and
   overlapping patterns are handled. See hol/test2.sml and
   isabelle/test1.sml for examples.

* The isabelle port still does not interface with the ".thy" files. This 
  is being worked on. 

* The proof of the wellfounded recursion theorem in hol/ has been simplified.

* In hol/ I have moved "define" from "Hol_datatype" to "post", since
it's an ugly-but-useful hack that doesn't really have much to do with
datatypes (it uses the datatype info to look up prim_rec definition
principles to apply). This caused changes to some of the files in the
"examples/" directory. "define" should someday grow into a useful
entrypoint in a HOL release, but it's not yet ready for prime time.


Please mail any problems to slind@informatik.tu-muenchen.de.
