Author: Jon Dart
Date: 07:27:18 08/28/04
Go up one level in this thread
I haven't really had much benefit from making my program's source public. 99% of people who download it only want the binaries. I have gotten a handful of bug fixes from people who've looked at the source and a few suggestions for improvement (not any that made a huge difference). In the really early days when I started, Gnuchess 4 was about the only decent quality program with published source. I studied it quite a bit but didn't find it very useful. It is very hard to read and the source isn't well documented. Crafty came out later and is very cleanly written. I used a number of ideas from Crafty including rotated bitboards. I didn't use the Crafty implementation: frankly I couldn't understand it very well from the code, and anyway my data structures are different so it wouldn't just slot in. But I got the basic idea and did my own implementation. My SEE is very similar to Crafty's. So are parts of my eval (but overall mine is quite different). Bob has made a great contribution to computer chess by making a really strong program open source. I'm not philosophically opposed to borrowing a whole algorithm from someone else's program (but not the literal code if it is copyrighted). However, with a few execeptions, I just haven't found anything that worked especially well, and hasn't been adequately described in the computer chess literature. E.g. Phalanx and GnuChess have some interesting variable-depth check extensions but the simpler method I use works better for me, based on tests. --Jon
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.