I've been making some time to work on TuxedoCat, but I decided to go back and refactor the work I've already done to make it more maintainable and easier to understand. So, a lot of the info below is outdated. If you want to follow along with the refactoring work there is a "refactor" branch that I'm actively working on.
progress on refactor:
move generation is working
See the COPYING file.
A .sln file is included in the repository so building the code should be as simple as opening the solution in Visual Studio and building. Visual Studio 2015 is recommended, any version of 2015 should work however (I use the free edition).
Simply cd into the TuxedoCat directory and run
make. Tested with gcc 6.1 on
The program doesn't accept any commandline parameters at the moment, so simply invoke the executable from a shell or point Winboard to it. If you run the engine on its own, you can use Winboard commands to interact with it. For documentation on the Winboard protocal read the protocol description, particularly the section "Commands from xboard to the engine". Currently TuxedoCat only reacts to the most essential commands, specifically
In addition to the regular Winboard commands, several other commands are available if you run the engine separately. The commands and their usage are as follows:
perft measures the number of possible chess positions that can be reached after <depth> number of moves from the current position. For example, from the standard starting position there are 20 possible moves for White, so
perft 1would be 20. For each of those 20 White moves Black has 20 possible responses, so
perft 2(still from the starting position) would be 400 (20 x 20). For depths greater than 2 it starts getting more complicated as there are different numbers of possible responses for each different move. Of course, perft can be calculated from positions other than the start position, too.
When you start TuxedoCat the board initializes to the standard starting position, so running perft immediately will calculate perft for that position. To calculate perft for another position, use the
setboardcommand to setup a position and then run perft.
Similar to perft, except that instead of simply returning the number of possible positions after <depth> moves, it lists all the possible moves at the top level and then calls
perft <depth - 1>for each of those positions, thus "dividing" the total count into separate sums for each original move. Mainly useful for debugging perft.
TuxedoCat logs all communication between itself and Winboard/the user. The log can be accessed by opening log.txt, located either in the same directory as the executable or in the working directory. There might be a difference between Linux and Windows here.
None that I know of right now, except that the engine sucks. :)
Currently the search is pure brute-force; the next step is to implement alpha-beta pruning and halfway-intelligent move ordering to reduce the search space.
Thanks to Norbert Leisner for his work on the amazing engine logos, and Tamás Bajusz for providing a Linux build.
Some bit-twiddling code was copied from the chess programming wiki; all of this code is clearly outlined in comments in the source files.
To report bugs or provide other feedback you can contact me at firstname.lastname@example.org.