Author: Steven Edwards
Date: 16:31:05 06/24/04
Symbolic: Status report 2004.06.24 Most of the recent work has been performed on Symbolic's underlying chess toolkit and not on Symbolic's Lisp source. Some of this was original work while most of it was fixing or extending toolkit code written many months ago. One area that needed cleaning was the toolkit's batch file processing code. This was motivated by revisions in the interactive command processor and also to better support the "call", "return", and "exit" commands; these now do the right thing no matter when or where they may appear. I have so far resisted the temptation to include support for parameterized batch files; simple recursion is enough for now. There is better support for multi-move input; a sequence of moves can be entered interactively on a single line. The "esv" (execute supplied variation) command was added; it causes a whole variation, possibly including move numbers, to be executed. It's quite handy for cut and paste operations. The toolkit's interactive auto response feature (i.e., search and play without an explicit command) is better implemented and can be deactivated if desired. However, I'm having some difficulty with controlling the opportunity search (others call this the ponder search or the permanent brain); I need to recover the file descriptor from an ifstream object to pass to select(), and the g++ 3.x libraries no longer support said recovery. It used to be "ifstreamObjPtr->rdbuf()->fd()", but no longer, as the fd() member function is buried a couple of levels deeper and is no longer public. My old and not so reliable 400 MHz AMD K6-2 Linux machine had caused a delay in testing due to random failures. These were finally traced to a finicky external cache; once it was disabled, the chess toolkit ran fine, although about 25 percent more slowly. Most of the near term work will be concentrated on getting the opportunity search to work correctly and efficiently in the three playing modes: manual, batch, and via server interface. Then the effort will move towards fully automating mulitgame testing against other programs; after this, I'll probably work a bit more on simple positional scoring before re-entering Symbolic's Lisp arena.
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.