= Open Adventure TODO =

* Use a real pseudorandom-number generator with a seed rather than just
  time-sampling.

* Add command log replay.  Note that the replay log needs to begin
  with the random-number seed.

* Use that feature to make regression tests from walkthroughs.

* Translate the FORTRANish mess to actual C.

* Inline the database so the code doesn't need an external file.