Mike Werner has a draft of his Ph.D. thesis 

Facilitating Schema Evolution With Automatic Program Transformations

at 

~werner/thesis.ps.gz

The thesis is of interest to our current work because
it also plays with two current themes:

1. It is useful to separate traversal code from other code
   (itineraries).
   Compare with the paper by Johan and Mitch:
   http://www.ccs.neu.edu/research/demeter/biblio/traversal-automata.html

2. It is useful to use views (again itineraries).
   Compare with:
   http://www.ccs.neu.edu/research/demeter/biblio/class-graph-views.html

In addition the thesis picks up on the theme of evolution 
by primitive transformation and points out some limitations
of AP.

Please take a look and give Mike feedback. 
Ken Baclawski is his advisor. Please note that this is a draft.

-- Karl