At the moment, I'm leaning toward proposing a careful implementation of the
basic Alexander continent-generation algorithm in C++, using my own C++
stack class to avoid the "how big does the sparklist get?" issue of
previous attempts. I estimate this as a 5-10 hour project; if I complete it
according to this estimate, then maybe I'll think of something to add at
the end to go beyond previous efforts. I won't worry at all about display
issues, and this has the potential to finally give me a "rock-solid"
implementation of your algorithm from way-back-when.