6 Dec 2010 dennissheil   » (Observer)

Blunder, Chess, Java, Architecture and Construction

So, I put Blunder up on Sourceforge.

Blunder is a suite of chess-related tools. Primarily, it helps you go over your games, and see where you made mistakes or missed opportunities. You keep looking at the boards where you made your biggest and/or most recent mistakes, and keep testing that you now know what to do correctly. Most chess teachers say this is one of the main ways to improve your game, and with Blunder it is automated.

Anyhow, the program has been out for almost a year, particularly the main LAMP (Linux, Apache, MySQL, PHP) component. However, one necessary component has been converting files in PGN format (records of games) to FEN format (pictures of individual boards at a set point). I give pointers how to do this, but have not been happy with any of the existing tools, and have begun writing my own one in Java, with GPL version 3 licensing. This was the impetus to put it on Sourceforge actually.

As I said, Blunder is functional already, particularly the LAMP package for going over games. One necessary component for that to work is PGN to FEN conversion, for which there are tools out there. I am unhappy with them, so I am writing my own in Java. If any Java developers want to send git patches, I'd be happy to get them. This second package within the Blunder project is in pre-alpha right now.

While this has all been done pretty loosely, I decided to try for a little bit stricter good practices in the pre-construction part of the project. I have to report - it worked out very well! I began by cheating on the good practices a little - I coded a method that read the PGN file into an array. It was just a detail I didn't want to bother with once requirements and architecture was done as I'd want to get right into the construction beyond that first.

My requirements were:
Read in a pgn (from say, a file), output a series of FENs for every move, or a specific FEN for one move. I might tweak the input or output requirements later, but the middle part, converting one to the other, will remain the heart of the program.

I then did architecture. I sketched out the major classes, their responsibilities and their interactions. Initially there were three classes - Pgn, Board and Fen. I thought about it and realized Pgn should have a helper class, Move, and Pgn would have an array of Move objects. Board is primarily an array of characters representing the board, and Fen is an output String representing the FEN. I think it was helpful thinking about all of this beforehand more than I usually would have. It saved time in the long run. Every minute I spent doing this right off the bat probably saved a multiple of itself so far.

One mistake I made is instead of making the Board array something that would be intuitive to me, I tried to fit its data structure to the other existing data structures. I thought this would make "less work". The problem is, Board's data structure then became inscrutable to me, and I had to bend my mind to figure out what it was, and kept making mistakes. I then decided to rearrange Board's data structure to something I could intuitively understand, and then use methods to do the conversion between it and the other two major data structures. This has worked out much better for me.

Most of the work left to get the program from pre-alpha to alpha is doing the logic (methods) for the various chess moves. I already have methods for PawnMoveNoCapture and KnightMoveNoCapture. My next method will probably be PawnMoveWithCapture - a move where the pawn captures a piece. The program needs methods for all the various moves - Queen move (capture and no capture), Bishop move (capture and no capture), Castling (Queenside or Kingside) and so on. This will be the bulk of the work to get the program into alpha.

I am plowing ahead with those methods right now. There is some code duplication within existing methods, but my concern is not with that but code duplication between methods - I already created a method to convert the letters from the Pgn moves to the numbers the array in Board uses, which both existing chess move methods use. I would like to complete moves for all the pieces, when capturing or not. Anyone who wants to send in git patches for these Java methods should feel free, the two existing methods can serve as a base.

You can grab it from the project's git page on Sourceforge.

Latest blog entries     Older blog entries

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!