Here are the lecture notes that I use in my graduate course on
constraints. The approach used is to present algorithms for
constraint satisfaction problems from the unified point of view
of nogood processing. The notes contain a detailed development of
all of the popular backtracking CSP algorithms. Warning. I no longer teach my course using these notes. Although they
are correct, they develop things is a more tedious manner than is optimal. My
view of backtracking as doing resolution become clearer to me, and I now
present things in a simpler manner using this relationship.