So far, I have only one Java example on-line for this course:
the eight-queens puzzle in Chapter 5 of the Budd book.
I didn't quite understand (or believe in) Budd's solution, so I tried
writing it myself from scratch. After much gnashing of teeth, I got it
to run, but it was incredibly slow: it took c. 75 seconds (on a DEC
3000 Alphastation) to find the first solution. Then I realized that
even if the leftmost two queens were attacking one another, I was
wasting my time moving the rightmost in hopes that they would
eventually kiss and make up. I added an optimization: when the current
queen runs out of rows to try, she doesn't just ask her neighbor to
advance; rather, she asks her neighbor to keep advancing until
said neighbor and all the queens to her left constitute a partial
solution, because there's no point in moving the current queen until
they do. This improved speed by at least a factor of 100. Finally I
realized that I had re-discovered the algorithm in Budd's book.
I've implemented both the slow version
and the fast version. They differ in one
line of code; indeed, I've written things in such a way that you can switch
from one to the other by redefining the compile-time constant SLOW to
either "true" or "false".