gnu1987/gnuchess/TODO

TODO LIST

BIG
  Rewrite move generator in assembly language or figure
   out a faster algorithm. The point is to speed up the move
   generation considerably.
 Enter all collected games of several great Chess-masters
   (Capablanca, Tal, Morphy). The program's "personality" will
   then be determined by the master's style. It is not clear
   whether selection of attackers or postional players would
   be better for our purposes. Think about it.

MEDIUM 
  Figure out how to speed up quiescence.
  Debug transposition table code and determine usefulness
  Add iterative deepening, if it useful
  See if aspiration is of any use.
  Modify makemove/unmakemove so that it incrementally updates
   the positional evaluation.

SMALL
  Program currently will move into a position that is in book,
   regardless of whose move it is. It should move into book positions
   only if the book position represents a position for the side to
   move that corresponds to the current game (after the program makes
   its move). Example: 1 e2e4 d7d6 2 d2d4 d7d5 (?)
  When program plays black, why does it oscillate
	1. e2e4 2. ... e7e5, since really e7e5 is
        the first move.
  Add tournament time controls.
  Makemove/Unmakemove should handle hashing correctly for
    promotions and castling.
  Separate history mechanism into its own file. Perhaps rename
  it to something else?
  Modify ply-1 preprocessor so it randomizes.
  Move out all random commands from main into separate files.
  History command does not handle 'undos' properly.
  Unmakemove does not seem to handle unmake's of pawn promotions
   that capture. That is c2xb1=BQ does not get unmade properly!.
   Actually, what seems to happen, is the unmake gets done but
   the captured piece is not restored to the square!!!
  Program will move into check if the checking piece is the
   enemy king.