- all threads should regularly poll a status variable
  this can be used to make ctrl-c working faster, and to print messages

- use optparse instead of getopt with more flexible commandline help

- replace the debug() function with the logging module
  we'll see how we can insert multiple debug levels into this thing

- use the Set type instead of hashmaps (did I use hashmaps for sets here?)

- the HTML parser should be even more forgiving with badly formatted html

include some web check and/or spider features:
- warn if overall size of page (including images/flash/etc.) is too big
- save downloaded pages
