Legal details ------------- PolyGlot 1.4 Copyright 2004-2006 Fabien Letouzey. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA See the file "copying.txt" for details. General ------- PolyGlot 1.4 (2006/01/16). PolyGlot is a "UCI adapter". It connects a UCI chess engine to an xboard interface such as WinBoard. UCI2WB is another such adapter (for Windows). PolyGlot tries to solve known problems with other adapters. For instance, it detects and reports draws by fifty-move rule, repetition, etc ... Official distribution URL ------------------------- The official distribution web site is Leo Dijksman's WBEC Ridderkerk: http://wbec-ridderkerk.nl/ This is where you should be looking for PolyGlot updates in the future. Install ------- PolyGlot can be placed in its own directory, or anywhere it can access the DLL file from (on Windows). On Windows the files "polyglot.exe" and "cygwin1.dll" (which you can download from http://wbec-ridderkerk.nl/) are needed. On Linux and Mac OS X only the file "polyglot_linux" or "polyglot_mac" is required. Compiling --------- The distribution comes up with Windows, Linux and Mac OS X binaries. Compiling should therefore not be necessary on those systems unless you want to make a change in the program. In any case this section describes the compiling procedure, it is safe to skip it. PolyGlot is a POSIX application (Unix compatible), and was developed on Linux using g++ (the GNU C++ compiler). 1) Unix You should be able to compile it on any POSIX-compliant operating system (*not* Windows) with the following command line (or similar): > g++ -O2 -o polyglot *.cpp IMPORTANT: In "io.cpp", the variable "UseCR" should be set to "false". A Makefile is provided but might not work on your system ... 2) Windows On Windows, you *must* use Cygnus GCC to compile PolyGlot. IMPORTANT: In "io.cpp", the variable "UseCR" should be set to "true". Usage ----- PolyGlot acts as an xboard engine. There should be no difference with a normal chess program as far as the interface (e.g. WinBoard) is concerned. PolyGlot is invoked using "polyglot ". Note that PolyGlot will look for the INI file in the current directory. If no is given, "polyglot.ini" is selected. To use PolyGlot with XBoard, you would type something like this: > xboard -fd 'ini_dir' -fcp 'polyglot engine.ini' Quotes are important when there is a space in the argument. IMPORTANT: some users seem confused by the concept of "current directory". PolyGlot needs to know where to read (INI file) and write (log file) files. Although it's possible to specify the full path to each file, a better solution is to provide a directory when launching PolyGlot, e.g. with the "-fd" XBoard option above. The directory should be where the INI file is. INI file -------- There should be a different INI file for each engine. Sections are composed of "variable = value" lines. See the sample INI files in the "example" directory. NOTE: There can be spaces in variable names or values. Do not use quotes. 1) [PolyGlot] section This section is used by PolyGlot only. The engine is unaware of these options. The list of available options is detailed below in this document. 2) [Engine] section This section contains engine UCI options. PolyGlot does not understand them, but sends the information to the engine at startup (converted to UCI form). You can add any UCI option that makes sense to the engine (not just the common options about hash-table size and tablebases). NOTE: use INI syntax, not UCI. For example "OwnBook = true" is correct. It will be replaced by PolyGlot with "setoption name OwnBook value true" at engine startup. Standard UCI options are "Hash", "NalimovPath", "NalimovCache" and "OwnBook". Hidden options like "Ponder" or "UCI_xxx" are automatic and should not be put in an INI file. The other options are engine-specific. Check their name using a UCI GUI or launch the engine in a console and type "uci". Options ------- These should be put in the [PolyGlot] section. - "EngineName" (default: UCI name) This is the name that will appear in the xboard interface. It is cosmetic only. You can use different names for tweaked versions of the same engine. If no "Engine Name" is given, the UCI name will be used. - "EngineDir" (default: ".") Full path of the directory where the engine is installed. You can use "." (without the quotes) if you know that PolyGlot will be launched in the engine directory or the engine is in the "path" and does not need any data file. - "EngineCommand" Put here the name of the engine executable file. You can also add command-line arguments. Path searching is used and the current directory will be "EngineDir". NOTE: Unix users are recommended to prepend "./"; this is required on some secure systems. - "Log" (default: false) Whether PolyGlot should log all transactions with the interface and the engine. This should be necessary only to locate problems. - "LogFile" The name of the log file. Note that it is put where PolyGlot was launched from, not into the engine directory. WARNING: Log files are not cleared between sessions, and can become very large. It is safe to remove them though. - "Resign" (default: false) Set this to "true" if you want PolyGlot to resign on behalf of the engine. NOTE: Some engines display buggy scores from time to time although the best move is correct. Use this option only if you know what you are doing (e.g. you always check the final position of games). - "ResignMoves" (default: 3) Number of consecutive moves with "resign" score (see below) before PolyGlot resigns for the engine. Positions with only one legal move are ignored. - "ResignScore" (default: 600) This is the score in centipawns that will trigger resign "counting". - "ShowPonder" (default: true) Show search information during engine pondering. Turning this off might be better for interactive use in some interfaces. - "KibitzMove" (default: false) Whether to kibitz when playing a move. - "KibitzPV" (default: false) Whether to kibitz when the PV is changed (new iteration or new best move). - "KibitzCommand" (default: "tellall") xboard command to use for kibitzing, normally "tellall" for kibitzing or "tellothers" for whispering. - "KibitzDelay" (default: 5) How many seconds to wait before starting kibitzing. This has an affect only if "KibitzPV" is selected, move kibitzes are always sent regardless of the delay. Work arounds ------------ Work arounds are identical to options except that they should be used only when necessary. Their purpose is to try to hide problems with various software (not just engines). The default value is always correct for bug-free software. IMPORTANT: Any of these work arounds might be removed in future versions of PolyGlot. You are strongly recommended to contact the author of faulty software and truly fix the problem. PolyGlot 1.4 supports the following work arounds: - "UCIVersion" (default: 2) The default value of 2 corresponds to UCI+. Use 1 to select plain UCI for engines that have problems with UCI+. - "CanPonder" (*** NEW ***, default: false) PolyGlot now conforms to the documented UCI behaviour: the engine will be allowed to ponder only if it (the engine) declares the "Ponder" UCI option. However some engines which can actually ponder do not declare the option. This work around lets PolyGlot know that they can ponder. - "SyncStop" (*** NEW ***, default: false) When a ponder miss occurs, Polyglot interrupts the engine and IMMEDIATELY launches a new search. While there should be no problem with this, some engines seem confused and corrupt their search board. "SyncStop" forces PolyGlot to wait for the (now useless) ponder search to finish before launching the new search. - "PromoteWorkAround" (*** NEW ***, default: false) Some engines do not specify a promotion piece, e.g. they send "e7e8" instead of the correct "e7e8q". This work around enables the incorrect form (and of course promotes into a queen). Opening Book ------------ PolyGlot 1.4 provides a simplistic opening-book implementation. The following options can be added to the [PolyGlot] section: - "Book" (default: false) Indicates whether a PolyGlot book should be used. This has no effect on the engine own book (which can be controlled with the UCI option "OwnBook" in the [Engine] section). In particular, it is possible to use both a PolyGlot book and an engine book. In that case, the engine book will be used whenever PolyGlot is out of book. Remember that PolyGlot is unaware of whether the engine is itself using a book or not. - "BookFile" The name of the (binary) book file. Note that PolyGlot will look for it in the directory it was launched from, not in the engine directory. Of course, full path can be used in which case the current directory does not matter. Note that there is no option to control book usage. All parameters are fixed when compiling a PGN file into a binary book (see below). This is purposeful and is not likely to change. Using a book does not require any additional memory, this can be important for memory-limited tournaments. A default book "fruit.bin" is provided in the archive. Note that this book is very small and should probably not be used in serious games. I hope that users will make other books available in the future. Book Making ----------- You can compile a PGN file into a binary book using PolyGlot on the command line. At the moment, only a main (random) book is provided. It is not yet possible to control opening lines manually. I am working on it though. Usage: "polyglot make-book ". "make-book" options are: - "-pgn" Name of the input PGN file. PolyGlot should support any standard-conforming file. Let me know if you encounter a problem. - "-bin" Name of the output binary file. I suggest ".bin" as the extension but in fact PolyGlot does not care. - "-max-ply" (default: infinite) How many plies (half moves) to read for each game. E.g. if set to "20", only the first 10 full moves of each game will be scanned. - "-min-game" (default: 3) How many times must a move be played to be kept in the book. In other words, moves that were played too rarely will be left out. If you scan full games "2" seems a minimum, but if you selected lines manually "1" will make sense. - "-only-white" *** NEW *** Save only white moves. This allows to use different parameters for white and black books, and merge them into a single file with the "merge-book" command, see below. - "-only-black" *** NEW *** Same for black moves. - "-uniform" *** NEW *** By default, a probability is calculated by PolyGlot for each move depending on how popular it is (how often it was playing in the provided PGN file) and how much it "scored". This option bypasses the default mechanism and affects equal probability to all moves. This allows more variety of play. This option is normally used only with hand-selected lines (e.g. "user books"). --- Example: "polyglot make-book -pgn games.pgn -bin book.bin -max-ply 30". Building a book is usually very fast (a few minutes at most). Note however that a lot of memory may be required. To reduce memory usage, select a ply limit. Book Merging ------------ *** NEW *** Usage: "polyglot merge-book -in1 -in2 -out " Merge two bin files into a single one. has "priority"; this means that if a position is present in both input books, data from will be ignored for this position. The two main applications are: 1) combine a white book and a black book (in which case priority does not matter) 2) combine a "user book" of manually-selected lines with a broader one from a large game set What follows is an admitedly complicated example of how this can be used. DO NOT MAILBOMB ME IF YOU DO NOT UNDERSTAND! My hope is that at least one advanced user will get what I mean and writes a better explanation on a web page or forum thread (yes, that's YOU, thanks by the way) ... --- Imagine that we've got 4 PGN files as follows: w1.pgn: fixed white lines, all moves manually checked w2.pgn: selected games (for random book as with PolyGlot 1.3) b1.pgn and b2.pgn: same for black The first step is to build 4 .bin files with appropriate options. Lines starting with "> " indicate what is typed on the command line. > polyglot make-book -min-game 1 -uniform -only-white -pgn w1.pgn -bin w1.bin I added "-uniform" because it allows randomness in the fixed lines (e.g. d4+e4 at 50%). It has no effect if lines are deterministic (only one move for a given position). "-min-game 1" is characteristic for user books. All moves are supposed to be safe so there is no reason to filter them with other heuristics. > polyglot make-book -min-score 50 -only-white -pgn w2.pgn -bin w2.bin This shows how min-score can actually be different for white and black (as with multiple books). I don't use "max-ply" because "min-game" default value of 3 will limit depth somewhat. You are of course free to use it. Same for black: > polyglot make-book -min-game 1 -uniform -only-black -pgn b1.pgn -bin b1.bin > polyglot make-book -min-score 40 -only-black -pgn b2.pgn -bin b2.bin At this point we have 4 .bin files. Notice that different parameters were used for white and for black (not to mention that different PGN files can be used). --- Let's now merge the white books. > polyglot merge-book -in1 w1.bin -in2 w2.bin -out w.bin Input files are not symmetrical, "in1" has priority over "in2". "skipped xxx entries." message from PolyGlot means there were some position conflicts. This is normal since we want to overwrite some random moves with fixed lines instead. Same for black: > polyglot merge-book -in1 b1.bin -in2 b2.bin -out b.bin Now we can finally merge the white and black books. > polyglot merge-book -in1 w.bin -in2 b.bin -out book.bin It's important to check that there are no conflicts, otherwise something went wrong. Note that this last operation was only made possible thanks to colour filtering, otherwise nearly all positions would lead to conflicts. For this reason, it does not make much sense to mix old .bin files (which contain moves for both colours). All these command lines might seem numerous and complicated but they can be put together into batch files. Chess 960 --------- *** NEW *** PolyGlot now supports Chess 960. However note that most xboard interfaces like WinBoard do not (except perhaps on an Internet chess server)! Here are pointers to modified XBoard/WinBoard versions that are known to work with PolyGlot in Chess960 mode: http://www.ascotti.org/programming/chess/winboard_x.htm (Windows) http://www.glaurungchess.com/xboard-960.tar.bz2 (Unix) http://www.milix.net/aice (?) It is also possible that PolyGlot is useful in combination with Arena(!): Arena Chess960 works correctly in xboard mode but it seems not compatible with the official UCI standard. With PolyGlot it is possible to include Chess960 UCI engines by using the xboard protocol instead. History ------- 2004/04/30: PolyGlot 1.0 - first public release. 2004/10/01: PolyGlot 1.1 - added "StartupWait" and "PonderWorkAround" ("AutoQuit" was available in version 1.0 but not documented). - fixed a minor bug that could prevent "AutoQuit" from working with some engines. 2005/01/29: PolyGlot 1.2 - rewrote engine initialisation and UCI parsing to increase UCI-standard compliance - added multi-move resign - added an internal work around for engines hanging with WinBoard 2005/06/03: PolyGlot 1.3 - added opening book - added kibitzing - added "ShowPonder" option 2006/01/16: PolyGlot 1.4 - added Chess960 (requires "fischerandom" xboard variant) - added "-only-white", "-only-black" and "-uniform" book-making options - added "merge-book" command - added "CanPonder", "SyncStop" and "PromoteWorkAround" work arounds - fixed "Move Now" (the engine was interrupted but the move was ignored) - fixed an UCI+draw problem that could occur with some engines after a draw by 50 moves or repetition - fixed pondering behaviour: the engine will ponder only if it declares the "Ponder" UCI option Known problems -------------- The addition of Chess960 support lead to a change in internal-move representation for castling. This slightly affected the opening-book format. I recommend that you recompile books with this version. Fruit 2.2 and above handle both book formats though. --- Several users reported engines losing on time. The playing conditions always mixed playing on an Internet server with pondering. Early log-file analysis did not reveal any misbehaviour by PolyGlot, but I have others to study. It is not yet clear what the source of the problem is, but let me state one more time that there is a forever incompatibility between the xboard and UCI protocol regarding a complex pondering/remaining-time relation. I suspect this might be related to the problem described above and if so, it is possible that there is no clean solution to it! In any case I have other log file to study that might reveal something, stay tuned! Thanks ------ Big thanks go to: - Leo Dijksman for compiling, hosting the PolyGlot distribution on his web site (see Links) and also for thorough testing - Tord Romstad, Joshua Shriver and George Sobala for compiling and testing on Mac OS X - all those who reported problems or proposed improvements; I am not well organised enough to provide their names! Links ----- - Tim Mann's Chess Pages: http://www.tim-mann.org/xboard.html - Leo Dijksman's WBEC Ridderkerk: http://wbec-ridderkerk.nl/ - Volker Pittlik's Winboard Forum: http://wbforum.volker-pittlik.name/ Contact me ---------- You can contact me at fabien_letouzey@hotmail.com; expect SLOW answer, if at all! If I am not available, you can discuss PolyGlot issues in Volker Pittlik's Winboard Forum: http://wbforum.volker-pittlik.name/ In fact for questions regarding specific Windows-only engines, you are advised to ask directly in the WinBoard forum, as I don't have Windows myself. The end ------- Fabien Letouzey, 2006/01/16.