1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN"
\r
2 "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
\r
3 <html xmlns="http://www.w3.org/1999/xhtml" lang="en" xml:lang="en">
\r
5 <title>Chess Engine Communication Protocol</title>
\r
6 <meta http-equiv="Content-Type" content="text/html;charset=utf-8" />
\r
7 <style type="text/css">
\r
9 border-top:2px solid black;
\r
10 border-bottom:2px solid black;
\r
12 .version1 { color: red;}
\r
13 .version43 { color: green;}
\r
14 .version44 { color: blue; }
\r
16 table tr { text-align: left}
\r
17 tr > td:first-child { font-weight:bold;}
\r
18 dt { font-weight:bold;}
\r
25 <div class="header">
\r
26 <h1>Chess Engine Communication Protocol</h1>
\r
27 <h2><a href="http://www.tim-mann.org/">Tim Mann</a> & <a href="http://home.hccnet.nl/h.g.muller/winboardF.html">H.G.Muller</a></h2>
\r
29 Version 2; implemented in xboard/WinBoard 4.2.1 and later. (Sept 3, 2009)<br />
\r
30 Changes since version 1 are indicated in <span class="version1">red</span>.<br />
\r
31 Changes for WinBoard 4.3.xx are indicated in <span class="version43">green</span>.<br />
\r
32 Changes for WinBoard 4.4.xx are indicated in <span class="version44">blue</span>.
\r
37 <li><a href="#1">1. Introduction</a></li>
\r
38 <li><a href="#2">2. Connection</a></li>
\r
39 <li><a href="#3">3. Debugging</a></li>
\r
40 <li><a href="#4">4. How it got this way</a></li>
\r
41 <li><a href="#5">5. WinBoard requires Win32 engines</a></li>
\r
42 <li><a href="#6">6. Hints on input/output</a></li>
\r
43 <li><a href="#7">7. Signals</a></li>
\r
44 <li><a href="#8">8. Commands from xboard to the engine</a></li>
\r
45 <li><a href="#9">9. Commands from the engine to xboard</a></li>
\r
46 <li><a href="#10">10. Thinking Output</a></li>
\r
47 <li><a href="#11">11. Time control</a></li>
\r
48 <li><a href="#12">12. Analyze Mode</a></li>
\r
49 <li><a href="#13">13. Idioms and backward compatibility features</a></li>
\r
54 <h2><a name="1">1. Introduction</a></h2>
\r
57 This document is a set of rough notes on the protocol that xboard and
\r
58 WinBoard use to communicate with gnuchessx and other chess engines.
\r
59 These notes may be useful if you want to connect a different chess
\r
60 engine to xboard. Throughout the notes, "xboard" means both xboard
\r
61 and WinBoard except where they are specifically contrasted.
\r
65 There are two reasons I can imagine someone wanting to do this:
\r
69 <li>You have, or are developing, a chess engine but you don't want to
\r
70 write your own graphical interface. </li>
\r
71 <li>You have, or are developing,a chess engine, and you want to
\r
72 interface it to the Internet Chess Server.</li>
\r
76 In case (2), if you are using xboard, you will need to configure the
\r
77 "Zippy" code into it, but WinBoard includes this code already. See
\r
79 href="http://www.tim-mann.org/xboard/zippy.README">zippy.README</a>
\r
80 in the xboard or WinBoard distribution for more information.
\r
84 These notes are unpolished, but I've attempted to make them complete
\r
85 in this release. If you notice any errors, omissions, or misleading
\r
86 statements, let me know.
\r
90 I'd like to hear from everyone who is trying to interface their own
\r
91 chess engine to xboard/WinBoard. Please join the mailing list for
\r
92 authors of xboard/WinBoard compatible chess engines and post a message
\r
93 about what you're doing. The list is now hosted by Yahoo Groups; you
\r
94 can join at <a href="http://groups.yahoo.com/group/chess-engines"
\r
95 >http://groups.yahoo.com/group/chess-engines</a>, or you can read the
\r
96 list there without joining. The list is filtered to prevent spam.
\r
99 <p class="version43">
\r
100 Note that the WinBoard 4.3.xx line was developed independently of the
\r
101 original GNU project, by H.G.Muller.
\r
102 If you have questions about WinBoard 4.3.xx, or want to report bugs in it,
\r
103 report them in the appropriate section of the
\r
104 <a href="http://www.open-aurec.com/wbforum/">WinBoard forum</a>.
\r
107 <h2><a name="2">2. Connection</a></h2>
\r
110 An xboard chess engine runs as a separate process from xboard itself,
\r
111 connected to xboard through a pair of anonymous pipes. The engine
\r
112 does not have to do anything special to set up these pipes. xboard
\r
113 sets up the pipes itself and starts the engine with one pipe as its
\r
114 standard input and the other as its standard output. The engine then
\r
115 reads commands from its standard input and writes responses to its
\r
116 standard output. This is, unfortunately, a little more complicated to
\r
117 do right than it sounds; see <a href="#6">section 6</a> below.
\r
121 And yes, contrary to some people's expectations, exactly the same
\r
122 thing is true for WinBoard. Pipes and standard input/output are
\r
123 implemented in Win32 and work fine. You don't have to use DDE, COM,
\r
124 DLLs, BSOD, or any of the other infinite complexity that
\r
125 Microsoft has created just to talk between two programs. A WinBoard
\r
126 chess engine is a Win32 console program that simply reads from its
\r
127 standard input and writes to its standard output. See sections
\r
128 <a href="#5">5</a> and <a href="#6">6</a> below for additional details.
\r
131 <h2><a name="3">3. Debugging</a></h2>
\r
134 To diagnose problems in your engine's interaction with xboard, use the
\r
135 -debug flag on xboard's command line to see the messages that are
\r
136 being exchanged. In WinBoard, these messages are written to the file
\r
137 WinBoard.debug instead of going to the screen.
\r
141 You can turn debug mode on or off while WinBoard is running by
\r
142 pressing Ctrl+Alt+F12. You can turn debug mode on or off while xboard
\r
143 is running by binding DebugProc to a shortcut key (and pressing the
\r
144 key!); see the instructions on shortcut keys in the xboard man page.
\r
148 While your engine is running under xboard/WinBoard, you can send a
\r
149 command directly to the engine by pressing Shift+1 (xboard) or Alt+1
\r
150 (WinBoard 4.0.3 and later). This brings up a dialog that you can type
\r
151 your command into. Press Shift+2 (Alt+2) instead to send to the
\r
152 second chess engine in Two Machines mode. On WinBoard 4.0.2 and earlier,
\r
153 Ctrl+Alt is used in place of Alt; this had to be changed due to a conflict
\r
154 with typing the @-sign on some European keyboards.
\r
157 <h2><a name="4">4. How it got this way</a></h2>
\r
160 Originally, xboard was just trying to talk to the existing
\r
161 command-line interface of GNU Chess 3.1+ and 4, which was designed
\r
162 for people to type commands to. So the communication protocol is very
\r
163 ad-hoc. It might have been good to redesign it early on, but because
\r
164 xboard and GNU Chess are separate programs, I didn't want to force
\r
165 people to upgrade them together to versions that matched. I
\r
166 particularly wanted to keep new versions of xboard working with old
\r
167 versions of GNU Chess, to make it easier to compare the play of old
\r
168 and new gnuchess versions. I didn't foresee the need for a clean
\r
169 protocol to be used with other chess engines in the future.
\r
173 Circumstances have changed over the years, and now there are many more
\r
174 engines that work with xboard. I've had to make the protocol
\r
175 description more precise, I've added some features that GNU Chess
\r
176 does not support, and I've specified the standard semantics of a few
\r
177 features to be slightly different from what GNU Chess 4 does.
\r
180 <p class="version1">
\r
181 This release of the protocol specification is the first to carry a
\r
182 version number of its own -- version 2. Previous releases simply
\r
183 carried a last-modified date and were loosely tied to specific
\r
184 releases of xboard and WinBoard. The version number "1" applies
\r
185 generally to all those older versions of the protocol.
\r
188 <p class="version1">
\r
189 Protocol version 2 remains compatible with older engines but has
\r
190 several new capabilities. In particular, it adds the
\r
191 "feature" command, a new mechanism for making backward-compatible
\r
192 changes and extensions to the protocol. Engines that do not support a
\r
193 particular new feature do not have to use it; new features are not
\r
194 enabled unless the engine specifically requests them using the feature
\r
195 command. If an engine does not send the feature command at all, the
\r
196 protocol behavior is nearly identical to version 1. Several new
\r
197 features can be selected by the feature command in version 2,
\r
198 including the "ping" command (recommended for all engines), the
\r
199 "setboard" command, and many optional parameters. Additional features
\r
200 will probably be added in future versions.
\r
203 <p class="version43">
\r
204 If it is necessary to have a separate name,
\r
205 it would be best to refer to the protocol including the green additions as version 2f.
\r
206 I really don't think it is a different protocol from version 2, though.
\r
207 I just tried to clarify some ambiguities in the original definition,
\r
208 now that the WinBoard 4.3.xx line has implemented them in a specific way.
\r
209 The hand-shaking protocol for features as defined in protocol 2 perfectly
\r
210 allows addition of an occasional new features without any need for stepping up the protocol version number,
\r
211 and I think refraining from the latter would enormously lower the barrier for actual
\r
212 implementation of these features in engines.
\r
214 The two really new things are the engine debug comments, and the "nps" command.
\r
215 The former merely tries to regulate an extremely common existing pactice
\r
216 of having engines dump debug messages on WinBoard in an unprotected way,
\r
217 as usually you get away with that.
\r
220 <h2><a name="5">5. WinBoard requires Win32 engines</a></h2>
\r
223 Due to some Microsoft brain damage that I don't understand, WinBoard
\r
224 does not work with chess engines that were compiled to use a DOS
\r
225 extender for 32-bit addressing. (Probably not with 16-bit DOS or
\r
226 Windows programs either.) WinBoard works only with engines that are
\r
227 compiled for the Win32 API. You can get a free compiler that targets
\r
228 the Win32 API from <a href="http://sources.redhat.com/cygwin/"
\r
229 >http://sources.redhat.com/cygwin/</a>. I think DJGPP 2.x should also
\r
230 work if you use the RSXNTDJ extension, but I haven't tried it. Of
\r
231 course, Microsoft Visual C++ will work. Most likely the other
\r
232 commercial products that support Win32 will work too (Borland, etc.),
\r
233 but I have not tried them. Delphi has been successfully used to write
\r
234 engines for WinBoard; if you want to do this, Tony Werten has donated
\r
235 some <a href="http://www.tim-mann.org/winboard/delphi.txt" >sample
\r
236 code</a> that should help you get started.
\r
239 <h2><a name="6">6. Hints on input/output</a></h2>
\r
242 Beware of using buffered I/O in your chess engine. The C stdio
\r
243 library, C++ streams, and the I/O packages in most other languages use
\r
244 buffering both on input and output. That means two things. First,
\r
245 when your engine tries to write some characters to xboard, the library
\r
246 stashes them in an internal buffer and does not actually write them to
\r
247 the pipe connected to xboard until either the buffer fills up or you
\r
248 call a special library routine asking for it to be flushed. (In C
\r
249 stdio, this routine is named <tt>fflush</tt>.) Second, when your engine tries
\r
250 to read some characters from xboard, the library does not read just
\r
251 the characters you asked for -- it reads all the characters that are
\r
252 currently available (up to some limit) and stashes any characters you
\r
253 are not yet ready for in an internal buffer. The next time you ask to
\r
254 read, you get the characters from the buffer (if any) before the
\r
255 library tries to read more data from the actual pipe.
\r
259 Why does this cause problems? First, on the output side, remember
\r
260 that your engine produces output in small quantities (say, a few
\r
261 characters for a move, or a line or two giving the current analysis),
\r
262 and that data always needs to be delivered to xboard/WinBoard for
\r
263 display immediately. If you use buffered output, the data you print
\r
264 will sit in a buffer in your own address space instead of being
\r
269 You can usually fix the output buffering problem by asking for the
\r
270 buffering to be turned off. In C stdio, you do this by calling
\r
271 <tt>setbuf(stdout, NULL)</tt>. A more laborious and error-prone
\r
272 method is to carefully call <tt>fflush(stdout)</tt> after every line
\r
273 you output; I don't recommend this. In C++, you can try
\r
274 <tt>cout.setf(ios::unitbuf)</tt>, which is documented in current
\r
275 editions of "The C++ Programming Language," but not older ones.
\r
276 Another C++ method that might work is
\r
277 <tt>cout.rdbuf()->setbuf(NULL, 0)</tt>. Alternatively, you can
\r
278 carefully call <tt>cout.flush()</tt> after every line you output;
\r
279 again, I don't recommend this.
\r
283 Another way to fix the problem is to use unbuffered operating system
\r
284 calls to write directly to the file descriptor for standard output.
\r
285 On Unix, this means <tt>write(1, ...)</tt> -- see the man page for write(2).
\r
286 On Win32, you can use either the Unix-like <tt>_write(1, ...)</tt> or Win32
\r
287 native routines like <tt>WriteFile</tt>.
\r
291 Second, on the input side, you are likely to want to poll during your
\r
292 search and stop it if new input has come in. If you implement
\r
293 pondering, you'll need this so that pondering stops when the user
\r
294 makes a move. You should also poll during normal thinking on your
\r
295 move, so that you can implement the "?" (move now) command, and so
\r
296 that you can respond promptly to a "result", "force", or "quit"
\r
297 command if xboard wants to end the game or terminate your engine.
\r
298 Buffered input makes polling more complicated -- when you poll, you
\r
299 must stop your search if there are <em>either</em> characters in the buffer
\r
300 <em>or</em> characters available from the underlying file descriptor.
\r
304 The most direct way to fix this problem is to use unbuffered operating
\r
305 system calls to read (and poll) the underlying file descriptor
\r
306 directly. On Unix, use <tt>read(0, ...)</tt> to read from standard input, and
\r
307 use <tt>select()</tt> to poll it. See the man pages read(2) and select(2).
\r
308 (Don't follow the example of GNU Chess 4 and use the FIONREAD ioctl to
\r
309 poll for input. It is not very portable; that is, it does not exist
\r
310 on all versions of Unix, and is broken on some that do have it.) On
\r
311 Win32, you can use either the Unix-like <tt>_read(0, ...)</tt> or the native
\r
312 Win32 <tt>ReadFile()</tt> to read. Unfortunately, under Win32, the function to
\r
313 use for polling is different depending on whether the input device is
\r
314 a pipe, a console, or something else. (More Microsoft brain damage
\r
315 here -- did they never hear of device independence?) For pipes, you
\r
316 can use <tt>PeekNamedPipe</tt> to poll (even when the pipe is unnamed).
\r
318 you can use <tt>GetNumberOfConsoleInputEvents</tt>. For sockets only, you can
\r
319 use <tt>select()</tt>. It might be possible to use
\r
320 <tt>WaitForSingleObject</tt> more
\r
321 generally, but I have not tried it. Some code to do these things can
\r
322 be found in Crafty's utility.c, but I don't guarantee that it's all
\r
323 correct or optimal.
\r
327 A second way to fix the problem might be to ask your I/O library not
\r
328 to buffer on input. It should then be safe to poll the underlying
\r
329 file descriptor as described above. With C, you can try calling
\r
330 <tt>setbuf(stdin, NULL)</tt>. However, I have never tried this. Also, there
\r
331 could be problems if you use <tt>scanf()</tt>, at least with certain patterns,
\r
332 because <tt>scanf()</tt> sometimes needs to read one extra character and "push
\r
333 it back" into the buffer; hence, there is a one-character pushback
\r
334 buffer even if you asked for stdio to be unbuffered. With C++, you
\r
335 can try <tt>cin.rdbuf()->setbuf(NULL, 0)</tt>, but again, I have never tried
\r
340 A third way to fix the problem is to check whether there are
\r
341 characters in the buffer whenever you poll. C I/O libraries generally
\r
342 do not provide any portable way to do this. Under C++, you can use
\r
343 <tt>cin.rdbuf()->in_avail()</tt>. This method has been reported to
\r
345 EXchess. Remember that if there are no characters in the buffer, you
\r
346 still have to poll the underlying file descriptor too, using the
\r
347 method described above.
\r
351 A fourth way to fix the problem is to use a separate thread to read
\r
352 from stdin. This way works well if you are familiar with thread
\r
353 programming. This thread can be blocked waiting for input to come in
\r
354 at all times, while the main thread of your engine does its thinking.
\r
355 When input arrives, you have the thread put the input into a buffer
\r
356 and set a flag in a global variable. Your search routine then
\r
357 periodically tests the global variable to see if there is input to
\r
358 process, and stops if there is. WinBoard and my Win32 ports of ICC
\r
359 timestamp and FICS timeseal use threads to handle multiple input
\r
363 <h2><a name="7">7. Signals</a></h2>
\r
365 <p>Engines that run on Unix need to be concerned with two Unix
\r
366 signals: <tt>SIGTERM</tt> and <tt>SIGINT</tt>. This applies both to
\r
367 engines that run under xboard and (the unusual case of) engines that
\r
368 WinBoard remotely runs on a Unix host using the -firstHost or
\r
369 -secondHost feature. It does not apply to engines that run on
\r
370 Windows, because Windows does not have Unix-style signals.
\r
371 <span class="version1">
\r
372 Beginning with version 2, you can now turn off the use of
\r
374 signals. See the "feature" command in <a href="#6">section 9</a> below.
\r
378 <p>First, when an engine is sent the "quit" command, it is also given
\r
379 a <tt>SIGTERM</tt> signal shortly afterward to make sure it goes away.
\r
380 If your engine reliably responds to "quit", and the signal causes
\r
381 problems for you, you should either ignore it by calling
\r
382 <tt>signal(SIGTERM, SIG_IGN)</tt> at the start of your program,
\r
383 or disable it with the "feature" command.</p>
\r
385 <p>Second, xboard will send an interrupt signal (<tt>SIGINT</tt>) at
\r
386 certain times when it believes the engine may not be listening to user
\r
387 input (thinking or pondering). WinBoard currently does this only when
\r
388 the engine is running remotely using the -firstHost or -secondHost
\r
389 feature, not when it is running locally. You probably need to know
\r
390 only enough about this grungy feature to keep it from getting in your
\r
395 The <tt>SIGINT</tt>s are basically tailored to the needs of GNU Chess 4
\r
396 on systems where its input polling code is broken or disabled.
\r
397 Because they work in a rather peculiar way, it is recommended that you
\r
398 either ignore <tt>SIGINT</tt> by having your engine call
\r
399 <tt>signal(SIGINT, SIG_IGN)</tt>, or disable it with the "feature"
\r
403 Here are details for the curious. If xboard needs to send a command
\r
404 when it is the chess engine's move (such as before the "?" command),
\r
405 it sends a <tt>SIGINT</tt> first. If xboard needs to send commands when it is
\r
406 not the chess engine's move, but the chess engine may be pondering
\r
407 (thinking on its opponent's time) or analyzing (analysis or analyze
\r
408 file mode), xboard sends a <tt>SIGINT</tt> before the first such command only.
\r
409 Another <tt>SIGINT</tt> is not sent until another move is made, even if xboard
\r
410 issues more commands. This behavior is necessary for GNU Chess 4. The
\r
411 first <tt>SIGINT</tt> stops it from pondering until the next move, but on some
\r
412 systems, GNU Chess 4 will die if it receives a <tt>SIGINT</tt> when not
\r
413 actually thinking or pondering.
\r
417 There are two reasons why WinBoard does not send the Win32 equivalent
\r
418 of <tt>SIGINT</tt> (which is called <tt>CTRL_C_EVENT</tt>) to local
\r
419 engines. First, the Win32 GNU Chess 4 port does not need it. Second, I
\r
420 could not find a way to get it to work. Win32 seems to be designed
\r
421 under the assumption that only console applications, not windowed
\r
422 applications, would ever want to send a <tt>CTRL_C_EVENT</tt>.
\r
425 <h2><a name="8">8. Commands from xboard to the engine</a></h2>
\r
428 All commands from xboard to the engine end with a newline (\n), even
\r
429 where that is not explicitly stated. All your output to xboard must
\r
430 be in complete lines; any form of prompt or partial line will cause
\r
435 At the beginning of each game, xboard sends an initialization string.
\r
436 This is currently "new\nrandom\n" unless the user changes it with the
\r
437 initString or secondInitString option.
\r
441 xboard normally reuses the same chess engine process for multiple
\r
442 games. At the end of a game, xboard will send the "force" command
\r
443 (see below) to make sure your engine stops thinking about the current
\r
444 position. It will later send the initString again to start a new
\r
445 game. If your engine can't play multiple games, you can disable reuse
\r
446 <span class="version1">
\r
447 either with the "feature" command (beginning in protocol version
\r
450 with xboard's -xreuse (or -xreuse2) command line
\r
451 option. xboard will then ask the process to quit after each game and
\r
452 start a new process for the next game.
\r
457 <dd>This command will be sent once immediately after your engine
\r
458 process is started. You can use it to put your engine into "xboard
\r
459 mode" if that is needed. If your engine prints a prompt to ask for
\r
460 user input, you must turn off the prompt and output a newline when the
\r
461 "xboard" command comes in.
\r
464 <dt class="version1">protover N</dt>
\r
465 <dd class="version1">
\r
466 <p>Beginning in protocol version 2 (in which N=2), this command will
\r
467 be sent immediately after the "xboard" command. If you receive some
\r
468 other command immediately after "xboard" (such as "new"), you can
\r
469 assume that protocol version 1 is in use. The "protover" command is
\r
470 the only new command that xboard always sends in version 2. All other
\r
471 new commands to the engine are sent only if the engine first enables
\r
472 them with the "feature" command. Protocol versions will always be
\r
473 simple integers so that they can easily be compared.
\r
476 <p>Your engine should reply to the protover command by sending the
\r
477 "feature" command (see below) with the list of non-default feature
\r
478 settings that you require, if any.</p>
\r
480 <p>Your engine should never refuse to run due to receiving a higher
\r
481 protocol version number than it is expecting! New protocol versions
\r
482 will always be compatible with older ones by default; the larger
\r
483 version number is simply a hint that additional "feature" command
\r
484 options added in later protocol versions may be accepted.
\r
488 <dt class="version1">accepted</dt>
\r
489 <dt class="version1">rejected</dt>
\r
490 <dd class="version1">
\r
491 These commands may be sent to your engine in reply to the "feature"
\r
492 command; see its documentation below.
\r
496 <dd>Reset the board to the standard chess starting position. Set
\r
497 White on move. Leave force mode and set the engine to play Black.
\r
498 Associate the engine's clock with Black and the opponent's clock with
\r
499 White. Reset clocks and time controls to the start of a new game.
\r
500 Use wall clock for time measurement.
\r
501 Stop clocks. Do not ponder on this move, even if pondering is on.
\r
502 Remove any search depth limit previously set by the sd command.
\r
505 <dt>variant VARNAME</dt>
\r
506 <dd>If the game is not standard chess, but a variant, this command is
\r
507 sent after "new" and before the first move or "edit" command. Currently
\r
508 defined variant names are:
\r
511 <tr><td>wildcastle</td><td>Shuffle chess where king can castle from d file</td></tr>
\r
512 <tr><td>nocastle</td><td>Shuffle chess with no castling at all</td></tr>
\r
513 <tr><td>fischerandom</td><td>Fischer Random</td></tr>
\r
514 <tr><td>bughouse</td><td>Bughouse, ICC/FICS rules</td></tr>
\r
515 <tr><td>crazyhouse</td><td>Crazyhouse, ICC/FICS rules</td></tr>
\r
516 <tr><td>losers</td><td>Win by losing all pieces or getting mated (ICC)</td></tr>
\r
517 <tr><td>suicide</td><td>Win by losing all pieces including king,
\r
518 or by having fewer pieces when one player has no legal moves (FICS)</td></tr>
\r
519 <tr class="version1"><td>giveaway</td><td>Win by losing all pieces including king,
\r
520 or by having no legal moves (ICC)</td></tr>
\r
521 <tr><td>twokings</td><td>Weird ICC wild 9</td></tr>
\r
522 <tr><td>kriegspiel</td><td>Kriegspiel (engines not supported)</td></tr>
\r
523 <tr><td>atomic</td><td>Atomic</td></tr>
\r
524 <tr><td>3check</td><td>Win by giving check 3 times</td></tr>
\r
525 <tr class="version43"><td>xiangqi </td><td>Chinese Chess (9x10 board)</td></tr>
\r
526 <tr class="version43"><td>shogi </td><td>Japanese Chess (9x9 bord)</td></tr>
\r
527 <tr class="version43"><td>capablanca</td><td>Capablanca Chess (10x8 board, with Archbishop and Chancellor)</td></tr>
\r
528 <tr class="version43"><td>gothic </td><td>Gothic Chess (10x8 board, same with better opening setup)</td></tr>
\r
529 <tr class="version43"><td>falcon </td><td>Falcon Chess (10x8 board, with two Falcon pieces)</td></tr>
\r
530 <tr class="version43"><td>shatranj </td><td>ancient Arabic Chess, with Elephants and General in stead of B and Q</td></tr>
\r
531 <tr class="version43"><td>courier </td><td>Courier Chess (12x8 board, a medieval precursor of modern Chess</td></tr>
\r
532 <tr class="version43"><td>knightmate </td><td>King moves as Knight and vice versa</td></tr>
\r
533 <tr class="version43"><td>berolina</td><td> Pawns capture straight ahead, and move diagonally</td></tr>
\r
534 <tr class="version43"><td>janus</td><td> Janus Chess (10x8, with two Archbishops)</td></tr>
\r
535 <tr class="version43"><td>caparandom </td><td>shuffle variant like FRC (10x8 board)</td></tr>
\r
536 <tr class="version43"><td>cylinder </td><td>Pieces wrap around between side edges, like board is a cylinder</td></tr>
\r
537 <tr class="version44"><td>super </td><td>Superchess: a shuffle variant with 4 fairy pieces on 8x8 board</td></tr>
\r
538 <tr class="version44"><td>great </td><td>Great Shatranj: sliders are replaced by corresponding short-range pieces on a 10x8 board</td></tr>
\r
539 <tr><td>unknown</td><td>Unknown variant (not supported)</td></tr>
\r
545 <dd>The chess engine should immediately exit. This command is used
\r
546 when xboard is itself exiting, and also between games if the -xreuse
\r
547 command line option is given (or -xreuse2 for the second engine).
\r
548 See also <a href="#7">Signals</a> above.
\r
552 <dd>This command is specific to GNU Chess 4. You can either ignore it
\r
553 completely (that is, treat it as a no-op) or implement it as GNU Chess
\r
554 does. The command toggles "random" mode (that is, it sets random =
\r
555 !random). In random mode, the engine adds a small random value to its
\r
556 evaluation function to vary its play. The "new" command sets random
\r
561 <dd>Set the engine to play neither color ("force mode"). Stop clocks.
\r
562 The engine should check that moves received in force mode are legal
\r
563 and made in the proper turn, but should not think, ponder, or make
\r
568 <dd>Leave force mode and set the engine to play the color that is on
\r
569 move. Associate the engine's clock with the color that is on move,
\r
570 the opponent's clock with the color that is not on move. Start the engine's
\r
571 clock. Start thinking and eventually make a move.
\r
574 <dt class="version1">playother</dt>
\r
575 <dd class="version1">
\r
576 (This command is new in protocol version 2. It is not
\r
577 sent unless you enable it with the feature command.)
\r
578 Leave force mode and set the engine to play the color that is <i>not</i> on
\r
579 move. Associate the opponent's clock with the color that is on move,
\r
580 the engine's clock with the color that is not on move. Start the opponent's
\r
581 clock. If pondering is enabled, the engine should begin pondering.
\r
582 If the engine later receives a move, it should start thinking and eventually
\r
588 <p><span class="version1">
\r
589 (This command is obsolete as of protocol version 2, but is still
\r
590 sent in some situations to accommodate older engines unless you disable it
\r
591 with the feature command.)
\r
593 Set White on move. Set the engine to play Black. Stop clocks.
\r
599 <span class="version1">
\r
600 (This command is obsolete as of protocol version 2, but is still
\r
601 sent in some situations to accommodate older engines unless you disable it
\r
602 with the feature command.)
\r
604 Set Black on move. Set the engine to play White. Stop clocks.
\r
607 <dt>level MPS BASE INC</dt>
\r
608 <dd>Set time controls. See the <a href="#11">Time Control</a> section below.
\r
612 <dd>Set time controls. See the <a href="#11">Time Control</a> section
\r
618 <p>The engine should limit its thinking to DEPTH ply.
\r
619 <span class="version43">The commands "level" or "st" and "sd" can be used together in an orthogonal way.
\r
620 If both are issued, the engine should observe both limitations:</span>
\r
621 In the protocol, the "sd" command isn't a time control. It doesn't
\r
622 say that your engine has unlimited time but must search to exactly the
\r
623 given depth. It says that you should pay attention to the time
\r
624 control as normal, but cut off the search at the specified depth even
\r
625 if you have time to search deeper. If you don't have time to search
\r
626 to the specified depth, given your normal time management algorithm,
\r
627 then you will want to stop sooner than the given depth.
\r
629 The "new" command should set the search depth back to unlimited. This
\r
630 is already stated in the spec. The "level" command should not affect
\r
631 the search depth. As it happens, xboard/WinBoard currently always
\r
632 sends sd (if needed) right after level, but that isn't part of the
\r
636 <dt><span class="version43">nps NODE_RATE</span></dt>
\r
637 <dd><span class="version43">The engine should not use wall-clock time to make its timing decisions,
\r
638 but an own internal time measure based on the number of nodes it has searched
\r
639 (and will report as "thinking output", see <a href="#10">section 10</a>),
\r
640 converted to seconds through dividing by the given NODE_RATE.
\r
641 Example: after receiving the commands "st 8" and "nps 10000",
\r
642 the engine should never use more that 80,000 nodes in the search for any move.
\r
643 In this mode, the engine should report user CPU time used (in its thinking output),
\r
644 rather than wall-clock time.
\r
645 This even holds if NODE_RATE is given as 0,
\r
646 but in that case it should also use the user CPU time for its timing decisions.
\r
647 The effect of an "nps" command should persist until the next "new" command.
\r
652 <dd>Set a clock that always belongs to the engine. N is a number in
\r
653 centiseconds (units of 1/100 second). Even if the engine changes to
\r
654 playing the opposite color, this clock remains with the engine.
\r
659 <dd><p>Set a clock that always belongs to the opponent. N is a number in
\r
660 centiseconds (units of 1/100 second). Even if the opponent changes to
\r
661 playing the opposite color, this clock remains with the opponent.
\r
663 If needed for purposes of board display in force mode (where the
\r
664 engine is not participating in the game) the time clock should be
\r
665 associated with the last color that the engine was set to play, the
\r
666 otim clock with the opposite color.
\r
669 <span class="version43">This business of "clocks remaining with the engine" is apparently so ambiguous
\r
670 that many engines implement it wrong.
\r
671 The clocks in fact always remain with the color.
\r
672 Which clock reading is relayed with "time", and which by "otim", is determined by which side the engine plays.
\r
673 Note that the way the clocks operate and receive extra time (in accordance with the selected time control)
\r
674 is not affected in any way by which moves are made by the engine, which by the opponent, and which were forced.
\r
678 <span class="version1">
\r
679 Beginning in protocol version 2, if you can't handle the time and
\r
680 otim commands, you can use the "feature" command to disable them; see
\r
683 The following techniques from older protocol versions also
\r
684 work: You can ignore the time and otim commands (that is, treat them
\r
685 as no-ops), or send back "Error (unknown command): time" the first
\r
686 time you see "time".
\r
691 <p>See below for the syntax of moves. If the move is illegal, print
\r
692 an error message; see the section "<a href="#9">Commands from the engine to
\r
693 xboard</a>". If the move is legal and in turn, make it. If not in force
\r
694 mode, stop the opponent's clock, start the engine's clock, start
\r
695 thinking, and eventually make a move.
\r
697 When xboard sends your engine a move, it normally sends coordinate
\r
698 algebraic notation. Examples:
\r
701 <tr><td>Normal moves:</td><td>e2e4</td></tr>
\r
702 <tr><td>Pawn promotion:</td><td>e7e8q</td></tr>
\r
703 <tr><td>Castling:</td><td>e1g1, e1c1, e8g8, e8c8</td></tr>
\r
704 <tr><td>Bughouse/crazyhouse drop:</td><td>P@h3</td></tr>
\r
705 <tr><td>ICS Wild 0/1 castling:</td><td>d1f1, d1b1, d8f8, d8b8</td></tr>
\r
706 <tr><td>FischerRandom castling:</td><td>O-O, O-O-O (oh, not zero)</td></tr>
\r
709 <p class="version43">
\r
710 Note that on boards with more than 9 ranks, counting of the ranks starts at 0.
\r
712 <p class="version1">
\r
713 Beginning in protocol version 2, you can use the feature command
\r
714 to select SAN (standard algebraic notation) instead; for example, e4,
\r
715 Nf3, exd5, Bxf7+, Qxf7#, e8=Q, O-O, or P@h3. Note that the last form,
\r
716 P@h3, is a extension to the PGN standard's definition of SAN, which does
\r
717 not support bughouse or crazyhouse.
\r
721 xboard doesn't reliably detect illegal moves, because it does not keep
\r
722 track of castling unavailability due to king or rook moves, or en
\r
723 passant availability. If xboard sends an illegal move, send back an
\r
724 error message so that xboard can retract it and inform the user; see
\r
725 the section "<a href="#9">Commands from the engine to xboard</a>".
\r
728 <dt class="version1">usermove MOVE</dt>
\r
729 <dd class="version1">
\r
730 By default, moves are sent to the engine without a command name;
\r
731 the notation is just sent as a line by itself.
\r
732 Beginning in protocol version 2, you can use the feature command
\r
733 to cause the command name "usermove" to be sent before the move.
\r
734 Example: "usermove e2e4".
\r
738 <dd><p>Move now. If your engine is thinking, it should move immediately;
\r
739 otherwise, the command should be ignored (treated as a no-op). It
\r
740 is permissible for your engine to always ignore the ? command. The
\r
741 only bad consequence is that xboard's Move Now menu command will do
\r
744 It is also permissible for your engine to move immediately if it gets
\r
745 any command while thinking, as long as it processes the command right
\r
746 after moving, but it's preferable if you don't do this. For example,
\r
747 xboard may send post, nopost, easy, hard, force, quit,
\r
748 <span class="version1">
\r
751 while the engine is on move.
\r
755 <dt class="version1">ping N</dt>
\r
756 <dd class="version1">
\r
757 <p>In this command, N is a decimal number. When you receive the command,
\r
758 reply by sending the string <strong>pong N</strong>, where N is the
\r
759 same number you received. Important: You must not reply to a "ping"
\r
760 command until you have finished executing all commands that you
\r
761 received before it. Pondering does not count; if you receive a ping
\r
762 while pondering, you should reply immediately and continue pondering.
\r
763 Because of the way xboard uses the ping command, if you implement the
\r
764 other commands in this protocol, you should never see a "ping" command
\r
765 when it is your move; however, if you do, you must not send the "pong"
\r
766 reply to xboard until after you send your move. For example, xboard
\r
767 may send "?" immediately followed by "ping". If you implement the "?"
\r
768 command, you will have moved by the time you see the subsequent ping
\r
769 command. Similarly, xboard may send a sequence like "force", "new",
\r
770 "ping". You must not send the pong response until after you have
\r
771 finished executing the "new" command and are ready for the new game to
\r
775 The ping command is new in protocol version 2 and will not be sent
\r
776 unless you enable it with the "feature" command. Its purpose is to
\r
777 allow several race conditions that could occur in previous versions of
\r
778 the protocol to be fixed, so it is highly recommended that you
\r
779 implement it. It is especially important in simple engines that do
\r
780 not ponder and do not poll for input while thinking, but it is needed in all
\r
786 <dd>The engine's opponent offers the engine a draw. To accept the
\r
787 draw, send "offer draw". To decline, ignore the offer (that is, send
\r
788 nothing). If you're playing on ICS, it's possible for the draw offer
\r
789 to have been withdrawn by the time you accept it, so don't assume the
\r
790 game is over because you accept a draw offer. Continue playing until
\r
791 xboard tells you the game is over. See also "offer draw" below.
\r
794 <dt>result RESULT {COMMENT}</dt>
\r
795 <dd>After the end of each game, xboard will send you a result command.
\r
796 You can use this command to trigger learning. RESULT is either 1-0,
\r
797 0-1, 1/2-1/2, or *, indicating whether white won, black won, the game
\r
798 was a draw, or the game was unfinished. The COMMENT string is purely
\r
799 a human-readable comment; its content is unspecified and subject to
\r
800 change. In ICS mode, it is passed through from ICS uninterpreted.
\r
801 Example: <pre>result 1-0 {White mates}</pre>
\r
803 Here are some notes on interpreting the "result" command. Some apply
\r
804 only to playing on ICS ("Zippy" mode).
\r
808 If you won but did not just play a mate, your opponent must have
\r
809 resigned or forfeited. If you lost but were not just mated, you
\r
810 probably forfeited on time, or perhaps the operator resigned manually.
\r
811 If there was a draw for some nonobvious reason, perhaps your opponent
\r
812 called your flag when he had insufficient mating material (or vice
\r
813 versa), or perhaps the operator agreed to a draw manually.
\r
817 You will get a result command even if you already know the game ended
\r
818 -- for example, after you just checkmated your opponent. In fact, if
\r
819 you send the "RESULT {COMMENT}" command (discussed below), you will
\r
820 simply get the same thing fed back to you with "result" tacked in
\r
821 front. You might not always get a "result *" command, however. In
\r
822 particular, you won't get one in local chess engine mode when the user
\r
823 stops playing by selecting Reset, Edit Game, Exit or the like.
\r
827 <dt><span class="version1">setboard FEN</span></dt>
\r
829 <p><span class="version1">
\r
830 The setboard command is the new way to set up positions, beginning
\r
831 in protocol version 2. It is not used unless it has been selected
\r
832 with the feature command. Here FEN is a position in Forsythe-Edwards
\r
833 Notation, as defined in the PGN standard.</span>
\r
834 <span class="version43">Note that this PGN standard referred to here
\r
835 only applies to normal Chess;
\r
836 Obviously in variants that cannot be described by a FEN for normal Chess,
\r
837 e.g. because the board is not 8x8, other pieces then PNBRQK participate,
\r
838 there are holdings that need to be specified, etc.,
\r
839 xboard will use a FEN format that is standard or suitable for that varant.
\r
840 In particular, in FRC or CRC, WinBoard will use Shredder-FEN or X-FEN standard,
\r
841 i.e. it can use the rook-file indicator letter to represent a castling right
\r
842 (like HAha) whenever it wants, but if it uses KQkq, this will always refer
\r
843 to the outermost rook on the given side.</span>
\r
846 <p class="version1">
\r
847 <em>Illegal positions:</em> Note that either setboard or edit can
\r
848 be used to send an illegal position to the engine. The user can
\r
849 create any position with xboard's Edit Position command (even, say,
\r
850 an empty board, or a board with 64 white kings and no black ones).
\r
851 If your engine receives a position that it considers illegal,
\r
852 I suggest that you send the response "tellusererror Illegal position",
\r
853 and then respond to any attempted move with "Illegal move" until
\r
854 the next new, edit, or setboard command.
\r
860 <p><span class="version1">
\r
861 The edit command is the old way to set up positions. For compatibility
\r
862 with old engines, it is still used by default, but new engines may prefer
\r
863 to use the feature command (see below) to cause xboard to use setboard instead.
\r
865 The edit command puts the chess engine into a special mode, where
\r
866 it accepts the following subcommands:</p>
\r
868 <tr><td>c</td><td>change current piece color, initially white</td></tr>
\r
869 <tr><td>Pa4 (for example)</td><td>place pawn of current color on a4</td></tr>
\r
870 <tr><td>xa4 (for example)</td><td>empty the square a4 (not used by xboard)</td></tr>
\r
871 <tr><td>#</td><td>clear board</td></tr>
\r
872 <tr><td>.</td><td>leave edit mode</td></tr>
\r
874 <p class="version1">
\r
875 See the Idioms section below for additional subcommands used in
\r
876 ChessBase's implementation of the protocol.
\r
879 <p>The edit command does not change the side to move. To set up a
\r
880 black-on-move position, xboard uses the following command sequence:
\r
887 <edit commands>
\r
892 This sequence is used to avoid the "black" command, which is now
\r
893 considered obsolete and which many engines never did implement as
\r
894 specified in this document.
\r
898 After an edit command is complete, if a king and a rook are on their
\r
899 home squares, castling is assumed to be available to them. En passant
\r
900 capture is assumed to be illegal on the current move regardless of the
\r
901 positions of the pawns. The clock for the 50 move rule starts at
\r
902 zero, and for purposes of the draw by repetition rule, no prior
\r
903 positions are deemed to have occurred.
\r
904 <span class="version43">
\r
905 In FRC or CRC, any rook and king put on the back rank should be considered to
\r
906 have castling rights, even if it later becomes apparent that they cannot be both in the
\r
907 initial position, because the position just set up is asymmetric.
\r
908 It is upto WinBoard to find work-around in cases where this is not desired,
\r
909 similar to the "black kludge" shown above, by setting up an earlier position,
\r
910 and then do a move to destroy castling rights or create e.p. rights.
\r
911 (Don't bet your life on it...)
\r
917 <dd>If the user asks for a hint, xboard sends your engine the command
\r
918 "hint". Your engine should respond with "Hint: xxx", where xxx is a
\r
919 suggested move. If there is no move to suggest, you can ignore the
\r
920 hint command (that is, treat it as a no-op).
\r
924 <dd>If the user selects "Book" from the xboard menu, xboard will send
\r
925 your engine the command "bk". You can send any text you like as the
\r
926 response, as long as each line begins with a blank space or tab (\t)
\r
927 character, and you send an empty line at the end. The text pops up in
\r
928 a modal information dialog.
\r
932 <dd>If the user asks to back up one move, xboard will send you the
\r
933 "undo" command. xboard will not send this command without putting you
\r
934 in "force" mode first, so you don't have to worry about what should
\r
935 happen if the user asks to undo a move your engine made. (GNU Chess 4
\r
936 actually switches to playing the opposite color in this case.)
\r
940 <dd>If the user asks to retract a move, xboard will send you the
\r
941 "remove" command. It sends this command only when the user is on
\r
942 move. Your engine should undo the last two moves (one for each
\r
943 player) and continue playing the same color.
\r
947 <dd>Turn on pondering (thinking on the opponent's time, also known as
\r
948 "permanent brain"). xboard will not make any assumption about what
\r
949 your default is for pondering or whether "new" affects this setting.
\r
953 <dd>Turn off pondering.</dd>
\r
956 <dd>Turn on thinking/pondering output.
\r
957 See <a href="#10">Thinking Output</a> section.</dd>
\r
960 <dd>Turn off thinking/pondering output.</dd>
\r
963 <dd>Enter analyze mode. See <a href="#12">Analyze Mode</a> section.</dd>
\r
966 <dd>This command informs the engine of its
\r
967 opponent's name. When the engine is playing on a chess server, xboard
\r
968 obtains the opponent's name from the server.
\r
969 <span class="version1">
\r
971 playing locally against a human user, xboard obtains the user's login
\r
972 name from the local operating system. When the engine is playing
\r
973 locally against another engine, xboard uses either the other engine's
\r
974 filename or the name that the other engine supplied in the myname
\r
975 option to the feature command. By default, xboard uses the name
\r
976 command only when the engine is playing on a chess server. Beginning
\r
977 in protocol version 2, you can change this with the name option to the
\r
978 feature command; see below.
\r
983 <dd>In ICS mode, xboard obtains the ICS opponent's rating from the
\r
984 "Creating:" message that appears before each game. (This message may
\r
985 not appear on servers using outdated versions of the FICS code.) In
\r
986 Zippy mode, it sends these ratings on to the chess engine using the
\r
987 "rating" command. The chess engine's own rating comes first, and if
\r
988 either opponent is not rated, his rating is given as 0.
\r
989 <span class="version1">
\r
990 In the future this command may also be used in other modes, if ratings
\r
993 Example: <pre>rating 2600 1500</pre>
\r
996 <dt><span class="version1">ics HOSTNAME</span></dt>
\r
997 <dd class="version1">
\r
998 If HOSTNAME is "-", the engine is playing against a local
\r
999 opponent; otherwise, the engine is playing on an Internet Chess Server
\r
1000 (ICS) with the given hostname. This command is new in protocol
\r
1001 version 2 and is not sent unless the engine has enabled it with
\r
1002 the "feature" command. Example: "ics freechess.org"
\r
1006 <dd>The opponent is also a computer chess engine. Some engines alter
\r
1007 their playing style when they receive this command.
\r
1010 <dt class="version1">pause</dt>
\r
1011 <dt class="version1">resume</dt>
\r
1012 <dd class="version1">(These commands are new in protocol
\r
1013 version 2 and will not be sent unless feature pause=1 is set. At
\r
1014 this writing, xboard actually does not use the commands at all, but it
\r
1015 or other interfaces may use them in the future.)
\r
1016 The "pause" command puts the engine into a special state where it
\r
1017 does not think, ponder, or otherwise consume significant CPU time.
\r
1018 The current thinking or pondering (if any) is suspended and both
\r
1019 player's clocks are stopped. The only command that the interface may
\r
1020 send to the engine while it is in the paused state is "resume". The
\r
1021 paused thinking or pondering (if any) resumes from exactly where it
\r
1022 left off, and the clock of the player on move resumes running from
\r
1026 <dt class="version44">memory N</dt>
\r
1027 <dd class="version44">
\r
1028 This command informs the engine on how much memory it is allowed to use maximally, in MegaBytes.
\r
1029 On receipt of this command, the engine should adapt the size of its hash tables accordingly.
\r
1030 This command does only fix the total memory use,
\r
1031 the engine has to decide for itself
\r
1032 (or be configured by the user by other means)
\r
1033 how to divide up the available memory between the various tables it wants to use
\r
1034 (e.g. main hash, pawn hash, tablebase cache, bitbases).
\r
1035 This command will only be sent to engines that have requested it through the memory feature,
\r
1036 and only at the start of a game,
\r
1037 as the first of the commands to relay engine option settings just before each "new" command.
\r
1040 <dt class="version44">cores N</dt>
\r
1041 <dd class="version44">
\r
1042 This command informs the engine on how many CPU cores it is allowed to use maximally.
\r
1043 This could be interpreted as the number of search threads for SMP engines.
\r
1044 (Threads that do not consume significant amounts of CPU time, like I/O threads, need not be included in the count.)
\r
1045 This command will only be sent to engines that have requested it through the smp feature.
\r
1046 The engine should be able to respond to the "cores" command any time during a game,
\r
1047 but it is allowed to finish a search in progress before procesing the command.
\r
1048 (Obeying the command should take priority over finishing a ponder search, though.)
\r
1049 In any case it will be sent at the start of every game
\r
1050 as the last command to relay engine option settings before the "new" command.
\r
1053 <dt class="version44">egtpath TYPE PATH</dt>
\r
1054 <dd class="version44">
\r
1055 This command informs the engine in which directory (given by the PATH argument)
\r
1056 it can find end-game tables of the specified TYPE.
\r
1057 The TYPE argument can be any character string which does not contain spaces.
\r
1058 Currently <strong>nalimov</strong> and <strong>scorpio</strong> are defined types,
\r
1059 for Nalimov tablebases and Scorpio bitbases, respectively,
\r
1060 but future developers of other formats are free to define their own format names.
\r
1061 The GUI simply matches the TYPE names the engine says it supports
\r
1062 with those that the user supplied when configuring xboard.
\r
1063 For every match, it sends a separate "y" command.
\r
1064 The PATH argument would normally (for Nalimov) be the pathname of the directory the EGT files are in,
\r
1065 but could also be the name of a file, or in fact anything the particular EGT type requires.
\r
1066 It is upto the developer of the EGT format to specify the syntax of this parameter.
\r
1067 This command will only be sent to engines that have told the GUI they support EGTs of the given TYPE
\r
1068 through the egt feature.
\r
1069 It will be sent at the start of each game, before the "new" command.
\r
1072 <dt class="version44">option NAME[=VALUE]</dt>
\r
1073 <dd class="version44">
\r
1074 This command changes the setting of the option NAME defined by the engine
\r
1075 (through an earlier feature command)
\r
1076 to the given VALUE.
\r
1077 XBoard will in general have no idea what the option means,
\r
1078 and will send the command only when a user changes the value of this option through a menu,
\r
1079 or at startup of the engine
\r
1080 (before the first 'cores' command or, if that is not sent, the first 'new' command)
\r
1081 in reaction to command-line options.
\r
1082 The NAME echoes back to the engine the string that was identified as an option NAME
\r
1083 in the feature command defining the option.
\r
1084 The VALUE is of the type (numeric or text or absent) that was implied by the option type
\r
1085 specified in this feature command,
\r
1086 i.e. with 'spin' and 'check' options VALUE will be a decimal integer (in the latter case 0 or 1),
\r
1087 with 'combo' and 'string' options VALUE will be a text string,
\r
1088 and with 'button' and 'save' options no VALUE will be sent at all.
\r
1092 <h3>Bughouse commands:</h3>
\r
1095 xboard now supports bughouse engines when in Zippy mode. See
\r
1096 <a href="http://www.tim-mann.org/xboard/zippy.README"
\r
1097 >zippy.README</a> for information on Zippy mode and how to turn on the
\r
1098 bughouse support. The bughouse move format is given above. xboard
\r
1099 sends the following additional commands to the engine when in bughouse
\r
1101 Commands to inform your engine of the partner's game state may
\r
1102 be added in the future.
\r
1106 <dt>partner <player></dt>
\r
1107 <dd><player> is now your partner for future games. Example: <pre>partner mann</pre>
\r
1111 <dd>Meaning: You no longer have a partner.
\r
1114 <dt>ptell <text></dt>
\r
1115 <dd>Your partner told you <text>, either with a ptell or an ordinary tell.
\r
1118 <dt>holding [<white>] [<black>]</dt>
\r
1119 <dd>White currently holds <white>; black currently holds <black>.
\r
1120 Example: <pre>holding [PPPRQ] []</pre></dd>
\r
1122 <dt>holding [<white>] [<black>] <color><piece></dt>
\r
1123 <dd>White currently holds <white>; black currently holds <black>, after
\r
1124 <color> acquired <piece>. Example: <pre>holding [PPPRQ] [R] BR</pre></dd>
\r
1127 <h2><a name="9">9. Commands from the engine to xboard</a></h2>
\r
1129 <p class="version1">
\r
1130 In general, an engine should not send any output to xboard that is not
\r
1131 described in this document. As the protocol is extended, newer
\r
1132 versions of xboard may recognize additional strings as commands that
\r
1133 were previously not assigned a meaning.
\r
1138 <dt class="version1"> feature FEATURE1=VALUE1 FEATURE2=VALUE2 ... </dt>
\r
1139 <dd class="version1">
\r
1140 <p>Beginning with version 2, the protocol includes the "feature"
\r
1141 command, which lets your engine control certain optional protocol
\r
1142 features. Feature settings are written as FEATURE=VALUE, where
\r
1143 FEATURE is a name from the list below and VALUE is the value to be
\r
1144 assigned. Features can take string, integer, or boolean values; the
\r
1145 type of value is listed for each feature. String values are written
\r
1146 in double quotes (for example, <tt>feature myname="Miracle Chess
\r
1147 0.9"</tt>), integers are written in decimal, and boolean values are
\r
1148 written as 0 for false, 1 for true. Any number of features can be set
\r
1149 in one feature command, or multiple feature commands can be given.</p>
\r
1152 Your engine should send one or more feature commands immediately after
\r
1153 receiving the "protover" command, since xboard needs to know the
\r
1154 values of some features before sending further commands to the engine.
\r
1155 Because engines that predate protocol version 2 do not send "feature",
\r
1156 xboard uses a timeout mechanism: when it first starts your engine, it
\r
1157 sends "xboard" and "protover N", then listens for feature commands for
\r
1158 two seconds before sending any other commands. To end this timeout
\r
1159 and avoid the wait, set the feature "done=1" at the end of your last
\r
1160 feature command. To increase the timeout, if needed, set the feature
\r
1161 "done=0" before your first feature command and "done=1" at the end.
\r
1162 If needed, it is okay for your engine to set done=0 soon as it starts,
\r
1163 even before it receives the xboard and protover commands. This can be
\r
1164 useful if your engine takes a long time to initialize itself. It
\r
1165 should be harmless even if you are talking to a (version 1) user
\r
1166 interface that does not understand the "feature" command, since such
\r
1167 interfaces generally ignore commands from the engine that they do not
\r
1172 The feature command is designed to let the protocol change without
\r
1173 breaking engines that were written for older protocol versions. When
\r
1174 a new feature is added to the protocol, its default value is always
\r
1175 chosen to be compatible with older versions of the protocol that did
\r
1176 not have the feature. Any feature that your engine does not set in a
\r
1177 "feature" command retains its default value, so as the protocol
\r
1178 changes, you do not have to change your engine to keep up with it
\r
1179 unless you want to take advantage of a new feature. Because some
\r
1180 features are improvements to the protocol, while others are meant to
\r
1181 cater to engines that do not implement all the protocol features, the
\r
1182 recommended setting for a feature is not always the same as the
\r
1183 default setting. The listing below gives both default and recommended
\r
1184 settings for most features.
\r
1188 You may want to code your engine so as to be able to work with
\r
1189 multiple versions of the engine protocol. Protocol version 1 does not
\r
1190 send the protover command and does not implement the feature command;
\r
1191 if you send a feature command in protocol version 1, it will have no
\r
1192 effect and there will be no response. In protocol version 2 or later,
\r
1193 each feature F that you set generates the response "accepted F" if the
\r
1194 feature is implemented, or "rejected F" if it is not. Thus an engine
\r
1195 author can request any feature without having to keep track of which
\r
1196 protocol version it was introduced in; you need only check whether the
\r
1197 feature is accepted or rejected. This mechanism also makes it
\r
1198 possible for a user interface author to implement a subset of a
\r
1199 protocol version by rejecting some features that are defined in that
\r
1200 version; however, you should realize that engine authors are likely to
\r
1201 code for xboard and may not be prepared to have a feature that they
\r
1202 depend on be rejected.
\r
1203 <span class="version44">If the GUI rejects an option feature because of the
\r
1204 syntax of the value, it should print the value string with the
\r
1205 "rejected" command, e.g. "rejected option nonsense" in response
\r
1206 to receiving feature option="nonsense".</span>
\r
1210 Here are the features that are currently defined.
\r
1214 <dt class="version1">ping (boolean, default 0, recommended 1)</dt>
\r
1215 <dd class="version1">
\r
1216 If ping=1, xboard may use the protocol's new "ping" command;
\r
1217 if ping=0, xboard will not use the command.
\r
1220 <dt class="version1">setboard (boolean, default 0, recommended 1)</dt>
\r
1221 <dd class="version1">
\r
1222 If setboard=1, xboard will use the protocol's new "setboard" command
\r
1223 to set up positions; if setboard=0, it will use the older "edit" command.
\r
1226 <dt class="version1">playother (boolean, default 0, recommended 1)</dt>
\r
1227 <dd class="version1">
\r
1228 If playother=1, xboard will use the protocol's new "playother" command
\r
1229 when appropriate; if playother=0, it will not use the command.
\r
1232 <dt class="version1">san (boolean, default 0)</dt>
\r
1233 <dd class="version1">
\r
1234 If san=1, xboard will send moves to the engine in standard algebraic
\r
1235 notation (SAN); for example, Nf3. If san=0, xboard will send moves in
\r
1236 coordinate notation; for example, g1f3. See MOVE in
\r
1237 <a href="#8">section 8</a> above for more details of both kinds of notation.
\r
1240 <dt class="version1">usermove (boolean, default 0)</dt>
\r
1241 <dd class="version1">
\r
1242 If usermove=1, xboard will send moves to the engine with the
\r
1243 command "usermove MOVE"; if usermove=0, xboard will send just the move,
\r
1244 with no command name.
\r
1247 <dt class="version1">time (boolean, default 1, recommended 1)</dt>
\r
1248 <dd class="version1">
\r
1249 If time=1, xboard will send the "time" and "otim" commands to
\r
1250 update the engine's clocks; if time=0, it will not.
\r
1253 <dt class="version1">draw (boolean, default 1, recommended 1)</dt>
\r
1254 <dd class="version1">
\r
1255 If draw=1, xboard will send the "draw" command if the engine's opponent
\r
1256 offers a draw; if draw=0, xboard will not inform the engine about
\r
1257 draw offers. Note that if draw=1, you may receive a draw offer while you
\r
1258 are on move; if this will cause you to move immediately, you should set
\r
1262 <dt class="version1">sigint (boolean, default 1)</dt>
\r
1263 <dd class="version1">
\r
1264 If sigint=1, xboard may send SIGINT (the interrupt signal) to
\r
1265 the engine as <a href="#7">section 7</a> above; if sigint=0, it will
\r
1269 <dt class="version1">sigterm (boolean, default 1)</dt>
\r
1270 <dd class="version1">
\r
1271 If sigterm=1, xboard may send SIGTERM (the termination signal) to
\r
1272 the engine as <a href="#7">section 7</a> above; if sigterm=0, it will
\r
1276 <dt class="version1">reuse (boolean, default 1, recommended 1) </dt>
\r
1277 <dd class="version1">
\r
1278 If reuse=1, xboard may reuse your engine for multiple games. If
\r
1279 reuse=0 (or if the user has set the -xreuse option on xboard's command
\r
1280 line), xboard will kill the engine process after every game and start
\r
1281 a fresh process for the next game.
\r
1284 <dt class="version1">analyze (boolean, default 1, recommended 1)</dt>
\r
1285 <dd class="version1">
\r
1286 If analyze=0, xboard will not try to use the "analyze" command; it
\r
1287 will pop up an error message if the user asks for analysis mode. If
\r
1288 analyze=1, xboard will try to use the command if the user asks for
\r
1292 <dt class="version1">myname (string, default determined from engine filename)</dt>
\r
1293 <dd class="version1">
\r
1294 This feature lets you set the name that xboard will use for your
\r
1295 engine in window banners, in the PGN tags of saved game files, and when
\r
1296 sending the "name" command to another engine.
\r
1299 <dt class="version1">variants (string, see text below)</dt>
\r
1300 <dd><span class="version1">
\r
1301 This feature indicates which chess variants your engine accepts.
\r
1302 It should be a comma-separated list of variant names. See the table
\r
1303 under the "variant" command in <a href="#8">section 8</a> above. If
\r
1304 you do not set this feature, xboard will assume by default that your
\r
1305 engine supports all variants. (However, the -zippyVariants
\r
1306 command-line option still limits which variants will be accepted in
\r
1307 Zippy mode.) It is recommended that you set this feature to the
\r
1308 correct value for your engine (just "normal" in most cases) rather
\r
1309 than leaving the default in place, so that the user will get an
\r
1310 appropriate error message if he tries to play a variant that your
\r
1311 engine does not support.</span>
\r
1313 <span class="version43">If your engine can play variants on a deviating board size,
\r
1314 like capablanca on an 8x8 board, or capablanca crazyhouse,
\r
1315 it can list them amongst the variants with a prefix spcifying board size plus
\r
1316 holdings size, like 8x8+0_capablanca or 10x8+7_capablanca.
\r
1317 If it is capable of playing any variant with an arbitrary board size,
\r
1318 it should list "boardsize" as one of the variants.
\r
1319 If there is a maximum to the board size, this can be prefixed,
\r
1320 e.g. "12x10+0_boardsize".
\r
1324 <dt class="version1">colors (boolean, default 1, recommended 0) </dt>
\r
1325 <dd><span class="version1">
\r
1326 If colors=1, xboard uses the obsolete "white" and "black"
\r
1327 commands in a stylized way that works with most older chess engines
\r
1328 that require the commands. See the "<a href="#13">Idioms</a>" section
\r
1329 below for details. If colors=0, xboard does not use the "white" and
\r
1330 "black" commands at all.
\r
1334 <dt class="version1">ics (boolean, default 0)</dt>
\r
1335 <dd class="version1">
\r
1336 If ics=1, xboard will use the protocol's new "ics" command
\r
1337 to inform the engine of whether or not it is playing on a chess server;
\r
1338 if ics=0, it will not.
\r
1341 <dt class="version1">name (boolean, see text below)</dt>
\r
1342 <dd class="version1">
\r
1343 If name=1, xboard will use the protocol's "name" command
\r
1344 to inform the engine of the opponent's name; if name=0, it will not.
\r
1345 By default, name=1 if the engine is playing on a chess server; name=0 if not.
\r
1348 <dt class="version1">pause (boolean, default 0)</dt>
\r
1349 <dd class="version1">
\r
1350 If pause=1, xboard may use the protocol's new "pause" command;
\r
1351 if pause=0, xboard assumes that the engine does not support this command.
\r
1354 <dt class="version43">nps (boolean, default ?)</dt>
\r
1355 <dd class="version43">
\r
1356 If nps=1, it means the engine supports the nps command.
\r
1357 If nps=0, it means the engine does not support it, and WinBoard should refrain from sending it.
\r
1358 Default is that WinBoard sends it, in an attempt to try out if the engine understand it.
\r
1359 The engine should properly respond with "Error (unkown command): nps" if it does not implement it,
\r
1360 (as any protocol version pre-scribes),
\r
1361 or WinBoard might assume that the engine did understand the command.
\r
1362 In that case the use of different time standards that ensues could lead to time forfeits for the engine.
\r
1365 <dt class="version43">debug (boolean, default 0)</dt>
\r
1366 <dd class="version43">
\r
1367 If debug=1, it means the engine wants to send debug output prefixed by '#',
\r
1368 which WinBoard should ignore, except for including it in the winboard.debug file.
\r
1369 As this feature is added to protocol 2 ony late,
\r
1370 so that not all protocol-2 supporting versions of WinBoard might implement it,
\r
1371 it is important that engines check if WinBoard accepts the feature.
\r
1372 If the feature is rejected,
\r
1373 engines must refrain from sending the debug output,
\r
1374 or do so at their own risk.
\r
1377 <dt class="version44">memory (boolean, default 0)</dt>
\r
1378 <dd class="version44">
\r
1379 If memory=1, the size of the total amount of memory available for the memory-consuming tables of the engine
\r
1380 (e.g. hash, EGTB cache)
\r
1381 will be set by the GUI through the "memory" command.
\r
1384 <dt class="version44">smp (boolean, default 0)</dt>
\r
1385 <dd class="version44">
\r
1386 If smp=1, the GUI will send the "cores" command to the engine to inform it how many CPU cores it can use.
\r
1387 Note that sending smp=1 does not imply the engine can use more than one CPU;
\r
1388 just that it wants to receive the "cores" command.
\r
1391 <dt class="version44">egt (string, see text below)</dt>
\r
1392 <dd class="version44">
\r
1393 This feature indicates which end-game table formats the engine supports.
\r
1394 It should be a comma-separated list of format names.
\r
1395 See under the "egtpath" command in <a href="#8">section 8</a> above.
\r
1396 If you do not set this feature, xboard will assume the engine does not support end-game tables,
\r
1397 and will not send any "egtpath" commands to inform the engine about their whereabouts.
\r
1400 <dt class="version44">option (string, see text below)</dt>
\r
1401 <dd><span class="version44">
\r
1402 This feature is used by the engine to define an option command to appear in a GUI menu,
\r
1403 so that the user can change the corresponding setting of the engine through the GUI interactively.
\r
1404 The string describes the option by defining a name, type, current value and (sometimes) the acceptable value range.
\r
1405 Unlike other features, option features are accumulated by the GUI,
\r
1406 and the GUI must be able to add a new option to the list at any time,
\r
1407 even after having received feature done=1.
\r
1408 There are ten different options types, each requiring a slighly different syntax of the defining string:
\r
1410 feature option="NAME -button"
\r
1412 feature option="NAME -save"
\r
1414 feature option="NAME -reset"
\r
1416 feature option="NAME -check VALUE"
\r
1418 feature option="NAME -string VALUE"
\r
1420 feature option="NAME -spin VALUE MIN MAX"
\r
1422 feature option="NAME -combo CHOICE1 /// CHOICE2 ..."
\r
1424 feature option="NAME -slider VALUE MIN MAX"
\r
1426 feature option="NAME -file VALUE"
\r
1428 feature option="NAME -path VALUE"
\r
1430 NAME is an arbitrary alphanumeric string which can contain spaces;
\r
1431 the other words in capitals would be replaced by the current (default) setting of the option,
\r
1432 (a character string for -string options, a decimal number for -spin and -check options,
\r
1433 were the latter uses 1=checked, 0=unchecked),
\r
1434 the minimum or maximum value of numeric (-spin) options,
\r
1435 or arbitrary text labels (for -combo option).
\r
1436 In the latter case, the current value will be preceded by an asterisk.
\r
1437 The -file and -path options are similar to -string, but can be used to inform the GUI that
\r
1438 the text represents a file name or folder name respectively,
\r
1439 so the GUI dialog could add the appropriate browse button to the text-edit field.
\r
1440 Similarly, a -slider option is like a -spin, but the GUI might make a different
\r
1441 graphical representation for it.
\r
1442 A -save option is like a -button, and defines an immediate command to be sent by the engine.
\r
1443 With -save the GUI will make sure all current option settings are flushed to the engine
\r
1444 before it sends this command.
\r
1445 A -reset option is like a -button, but use of it purges the list of options before sending
\r
1446 the corresponding option command to the engine.
\r
1447 This enables the engine to completely redefine its options or their current settings,
\r
1448 by sending a new set of option feature commands to the GUI,
\r
1449 terminated by feature done=1.
\r
1450 (The effect of sending an option feature for an option with the same name as was defined before,
\r
1451 without first receiving a -reset option command, is undefined.)
\r
1455 <dt class="version1">done (integer, no default)</dt>
\r
1456 <dd><span class="version1">
\r
1457 If you set done=1 during the initial two-second timeout after
\r
1458 xboard sends you the "xboard" command, the
\r
1459 timeout will end and xboard will not look for any more feature
\r
1460 commands before starting normal operation.
\r
1461 If you set done=0, the initial timeout is increased to one hour;
\r
1462 in this case, you must set done=1 before xboard will enter normal operation.
\r
1469 <dt>Illegal move: MOVE</dt>
\r
1470 <dt>Illegal move (REASON): MOVE</dt>
\r
1471 <dd>If your engine receives a MOVE command that is recognizably a move
\r
1472 but is not legal in the current position, your engine must print an
\r
1473 error message in one of the above formats so that xboard can pass the
\r
1474 error on to the user and retract the move. The (REASON) is entirely
\r
1475 optional. Examples:
\r
1478 Illegal move: e2e4
\r
1479 Illegal move (in check): Nf3
\r
1480 Illegal move (moving into check): e1g1
\r
1483 Generally, xboard will never send an ambiguous move, so it does not
\r
1484 matter whether you respond to such a move with an Illegal move message
\r
1485 or an Error message.
\r
1489 <dt>Error (ERRORTYPE): COMMAND</dt>
\r
1490 <dd>If your engine receives a command it does not understand or does
\r
1491 not implement, it should print an error message in the above format so
\r
1492 that xboard can parse it. Examples:
\r
1494 Error (ambiguous move): Nf3
\r
1495 Error (unknown command): analyze
\r
1496 Error (command not legal now): undo
\r
1497 Error (too many parameters): level 1 2 3 4 5 6 7
\r
1501 <dt>move MOVE</dt>
\r
1502 <dd>Your engine is making the move MOVE. Do not echo moves from
\r
1503 xboard with this command; send only new moves made by the engine.
\r
1505 <div class="version1">
\r
1506 <p>For the actual move text from your chess engine (in place of MOVE
\r
1507 above), your move should be either</p>
\r
1509 <li>in coordinate notation (e.g.,
\r
1510 e2e4, e7e8q) with castling indicated by the King's two-square move (e.g.,
\r
1512 <li>in Standard Algebraic Notation (SAN) as defined in the
\r
1513 Portable Game Notation standard (e.g, e4, Nf3, O-O, cxb5, Nxe4, e8=Q),
\r
1514 with the extension piece@square (e.g., P@f7) to handle piece placement
\r
1515 in bughouse and crazyhouse.</li>
\r
1518 xboard itself also accepts some variants of SAN, but for compatibility
\r
1519 with non-xboard interfaces, it is best not to rely on this behavior.
\r
1522 <p>Warning: Even though all versions of this protocol specification
\r
1523 have indicated that xboard accepts SAN moves, some non-xboard
\r
1524 interfaces are known to accept only coordinate notation. See the
\r
1525 Idioms section for more information on the known limitations of some
\r
1526 non-xboard interfaces. It should be safe to send SAN moves if you
\r
1527 receive a "protover 2" (or later) command from the interface, but
\r
1528 otherwise it is best to stick to coordinate notation for maximum
\r
1529 compatibility. An even more conservative approach would be for your
\r
1530 engine to send SAN to the interface only if you have set feature san=1
\r
1531 (which causes the interface to send SAN to you) and have received
\r
1532 "accepted san" in reply.
\r
1537 <dt>RESULT {COMMENT}</dt>
\r
1538 <dd>When your engine detects
\r
1539 that the game has ended by rule, your engine must output a line of the
\r
1540 form "RESULT {comment}" (without the quotes), where RESULT is a PGN
\r
1541 result code (1-0, 0-1, or 1/2-1/2), and comment is the reason. Here
\r
1542 "by rule" means that the game is definitely over because of what
\r
1543 happened on the board. In normal chess, this includes checkmate,
\r
1544 stalemate, triple repetition, the 50 move rule, or insufficient
\r
1545 material; it does not include loss on time or the like.
\r
1550 1/2-1/2 {Draw by repetition}
\r
1551 1/2-1/2 {Stalemate}
\r
1555 xboard relays the result to the user, the ICS, the other engine in Two
\r
1556 Machines mode, and the PGN save file as required.
\r
1557 <span class="version43">Note that "definitey over" above means that sending this command
\r
1558 will be taken by WinBoard as an unconditional refusal of the engine to play on,
\r
1559 which might cause you to forfeit if the game was in fact not over.
\r
1560 This command should thus not be used to offer draws, accept draws,
\r
1561 or make draw-by-rule claims that are not yet valid in the current position
\r
1562 (but will be after you move).
\r
1563 For offering and claiming such draws, "offer draw" should be used.</span>
\r
1566 <p class="version44">
\r
1567 Note that (in accordance with FIDE rules) only KK, KNK, KBK and KBKB with all bishops on the
\r
1568 same color can be claimed as draws on the basis of insufficient mating material.
\r
1569 The end-games KNNK, KBKN, KNKN and KBKB with unlike bishops do have mate positions,
\r
1570 and cannot be claimed.
\r
1571 Complex draws based on locked Pawn chains will not be recognized as draws by most interfaces,
\r
1572 so do not claim in such positions, but just offer a draw or play on.
\r
1575 <p class="version44">
\r
1576 Note to GUI programmers: RESULT commands that the engine sends immediately after its move
\r
1577 might be detected by the GUI only after the opponent has moved, because of communication
\r
1578 and scheduling delays, no matter how fast the engine sent it.
\r
1579 Any judgement of the validity of RESULT claims based on te "current" board position
\r
1580 will have to account for this uncertainty.
\r
1585 <dd>If your engine wants to resign, it can send the command "resign".
\r
1586 Alternatively, it can use the "RESULT {comment}" command if the string
\r
1587 "resign" is included in the comment; for example "0-1 {White
\r
1588 resigns}". xboard relays the resignation to the user, the ICS, the
\r
1589 other engine in Two Machines mode, and the PGN save file as required.
\r
1590 <span class="version44">Note that many interfaces work more smoothly if you resign <em>before</em>
\r
1594 <dt>offer draw</dt>
\r
1595 <dd>If your engine wants to offer a draw by agreement (as opposed to
\r
1596 claiming a draw by rule), it can send the command "offer draw".
\r
1597 xboard relays the offer to the user, the ICS, the other engine in Two
\r
1598 Machines mode, and the PGN save file as required. In Machine White,
\r
1599 Machine Black, or Two Machines mode, the offer is considered valid
\r
1600 until your engine has made two more moves.
\r
1601 <span class="version43">This command must also be used to accept a draw offer.
\r
1602 Do not use the 1/2-1/2 command for that, as the offer might be no longer valid,
\r
1603 in which case a refusal to play on implied by the RESULT command might make you forfeit the game.
\r
1604 "offer draw" should also be used to claim 50-move and 3-fold-repetition draws
\r
1605 that will occur <em>after</em> your move, by sending it <em>before</em> making the move.
\r
1606 WinBoard will grant draw offers without the opponent having any say in
\r
1607 it in situations where draws can be claimed.
\r
1608 Only if the draw cannot be claimed, the offer will be passed to your opponent after you make your next move,
\r
1609 just before WinBoard relays this move to the opponent.
\r
1613 <dt class="version1">tellopponent MESSAGE</dt>
\r
1614 <dd class="version1">
\r
1615 This command lets the engine give a message to its opponent,
\r
1616 independent of whether the opponent is a user on the local machine or
\r
1617 a remote ICS user (Zippy mode). MESSAGE consists of any characters,
\r
1618 including whitespace, to the end of the line. When the engine is
\r
1619 playing against a user on the local machine, xboard pops up an
\r
1620 information dialog containing the message. When the engine is playing
\r
1621 against an opponent on the ICS (Zippy mode), xboard sends "say
\r
1622 MESSAGE\n" to the ICS.
\r
1625 <dt class="version1">tellothers MESSAGE </dt>
\r
1626 <dd class="version1">This command lets the engine give a message to people watching the
\r
1627 game other than the engine's opponent. MESSAGE consists of any
\r
1628 characters, including whitespace, to the end of the line. When the
\r
1629 engine is playing against a user on the local machine, this command
\r
1630 does nothing. When the engine is playing against an opponent on the
\r
1631 ICS (Zippy mode), xboard sends "whisper MESSAGE\n" to the ICS.
\r
1634 <dt class="version1">tellall MESSAGE</dt>
\r
1635 <dd class="version1">This command lets the engine give a message to its opponent and
\r
1636 other people watching the game,
\r
1637 independent of whether the opponent is a user on the local machine or
\r
1638 a remote ICS user (Zippy mode). MESSAGE consists of any characters,
\r
1639 including whitespace, to the end of the line. When the engine is
\r
1640 playing against a user on the local machine, xboard pops up an
\r
1641 information dialog containing the message. When the engine is playing
\r
1642 against an opponent on the ICS (Zippy mode), xboard sends "kibitz
\r
1643 MESSAGE\n" to the ICS.
\r
1646 <dt>telluser MESSAGE</dt>
\r
1647 <dd>xboard pops up an information dialog containing the message.
\r
1648 MESSAGE consists of any characters, including whitespace, to the end
\r
1652 <dt>tellusererror MESSAGE</dt>
\r
1653 <dd>xboard pops up an error dialog containing the message.
\r
1654 MESSAGE consists of any characters, including whitespace, to the end
\r
1658 <dt>askuser REPTAG MESSAGE</dt>
\r
1659 <dd>Here REPTAG is a string containing no whitespace, and MESSAGE
\r
1660 consists of any characters, including whitespace, to the end of the
\r
1661 line. xboard pops up a question dialog that says MESSAGE and
\r
1662 has a typein box. If the user types in "bar", xboard sends "REPTAG
\r
1663 bar" to the engine. The user can cancel the dialog and send nothing.
\r
1666 <dt>tellics MESSAGE</dt>
\r
1667 <dd>In Zippy mode, xboard sends "MESSAGE\n" to ICS. MESSAGE consists
\r
1668 of any characters, including whitespace, to the end of the line.
\r
1671 <dt class="version1">tellicsnoalias MESSAGE</dt>
\r
1672 <dd class="version1">
\r
1673 In Zippy mode, xboard sends "xMESSAGE\n" to ICS, where "x" is a
\r
1674 character that prevents the ICS from expanding command aliases, if
\r
1675 xboard knows of such a character. (On chessclub.com and chess.net,
\r
1676 "/" is used; on freechess.org, "$" is used.) MESSAGE consists of any
\r
1677 characters, including whitespace, to the end of the line.
\r
1680 <dt class="version43"># COMMENT</dt>
\r
1681 <dd class="version43">
\r
1682 The engine can send any string of printable characters, terminated by a newline,
\r
1683 for inclusion in the winboard.debug file, provided the line starts with a '#' character.
\r
1684 If the engine has set feature debug=1,
\r
1685 it is guaranteed that WinBoard (and any future version of it) will completely ignore
\r
1686 these lines in any other respect.
\r
1690 <h2><a name="10">10. Thinking Output</a></h2>
\r
1693 If the user asks your engine to "show thinking", xboard sends your
\r
1694 engine the "post" command. It sends "nopost" to turn thinking off.
\r
1695 In post mode, your engine sends output lines to show the progress of
\r
1696 its thinking. The engine can send as many or few of these lines as it
\r
1697 wants to, whenever it wants to. Typically they would be sent when the
\r
1698 PV (principal variation) changes or the depth changes. The thinking
\r
1699 output should be in the following format:
\r
1702 <pre>ply score time nodes pv</pre>
\r
1706 <tr><td>ply</td><td>Integer giving current search depth.</td></tr>
\r
1707 <tr><td>score</td><td>Integer giving current evaluation in centipawns.</td></tr>
\r
1708 <tr><td>time</td><td>Current search time in centiseconds (ex:1028 = 10.28 seconds).</td></tr>
\r
1709 <tr><td>nodes</td><td>Nodes searched.</td></tr>
\r
1710 <tr><td>pv</td><td>Freeform text giving current "best" line.
\r
1711 You can continue the pv onto another line if you start each
\r
1712 continuation line with at least four space characters.</td></tr>
\r
1719 <pre> 9 156 1084 48000 Nf3 Nc6 Nc3 Nf6</pre>
\r
1726 9 ply, score=1.56, time = 10.84 seconds, nodes=48000, PV = "Nf3 Nc6 Nc3 Nf6"
\r
1730 Longer example from actual Crafty output:
\r
1734 4 109 14 1435 1. e4 d5 2. Qf3 dxe4 3. Qxe4 Nc6
\r
1735 4 116 23 2252 1. Nf3 Nc6 2. e4 e6
\r
1736 4 116 27 2589 1. Nf3 Nc6 2. e4 e6
\r
1737 5 141 44 4539 1. Nf3 Nc6 2. O-O e5 3. e4
\r
1738 5 141 54 5568 1. Nf3 Nc6 2. O-O e5 3. e4
\r
1742 You can use the PV to show other things; for instance, while in book,
\r
1743 Crafty shows the observed frequency of different reply moves in its
\r
1744 book. In situations like this where your engine is not really
\r
1745 searching, start the PV with a '(' character:
\r
1749 0 0 0 0 (e4 64%, d4 24%)
\r
1753 GNU Chess output is very slightly different. The ply number is
\r
1754 followed by an extra nonblank character, and the time is in seconds,
\r
1755 not hundredths of seconds. For compatibility, xboard accepts the
\r
1756 extra character and takes it as a flag indicating the different time
\r
1761 2. 14 0 38 d1d2 e8e7
\r
1762 3+ 78 0 65 d1d2 e8e7 d2d3
\r
1763 3& 14 0 89 d1d2 e8e7 d2d3
\r
1764 3& 76 0 191 d1e2 e8e7 e2e3
\r
1765 3. 76 0 215 d1e2 e8e7 e2e3
\r
1766 4& 15 0 366 d1e2 e8e7 e2e3 e7e6
\r
1767 4. 15 0 515 d1e2 e8e7 e2e3 e7e6
\r
1768 5+ 74 0 702 d1e2 f7f5 e2e3 e8e7 e3f4
\r
1769 5& 71 0 1085 d1e2 e8e7 e2e3 e7e6 e3f4
\r
1770 5. 71 0 1669 d1e2 e8e7 e2e3 e7e6 e3f4
\r
1771 6& 48 0 3035 d1e2 e8e7 e2e3 e7e6 e3e4 f7f5 e4d4
\r
1772 6. 48 0 3720 d1e2 e8e7 e2e3 e7e6 e3e4 f7f5 e4d4
\r
1773 7& 48 0 6381 d1e2 e8e7 e2e3 e7e6 e3e4 f7f5 e4d4
\r
1774 7. 48 0 10056 d1e2 e8e7 e2e3 e7e6 e3e4 f7f5 e4d4
\r
1775 8& 66 1 20536 d1e2 e8e7 e2e3 e7e6 e3d4 g7g5 a2a4 f7f5
\r
1776 8. 66 1 24387 d1e2 e8e7 e2e3 e7e6 e3d4 g7g5 a2a4 f7f5
\r
1777 9& 62 2 38886 d1e2 e8e7 e2e3 e7e6 e3d4 h7h5 a2a4 h5h4
\r
1779 9. 62 4 72578 d1e2 e8e7 e2e3 e7e6 e3d4 h7h5 a2a4 h5h4
\r
1781 10& 34 7 135944 d1e2 e8e7 e2e3 e7e6 e3d4 h7h5 c2c4 h5h4
\r
1783 10. 34 9 173474 d1e2 e8e7 e2e3 e7e6 e3d4 h7h5 c2c4 h5h4
\r
1787 <p>If your engine is pondering (thinking on its opponent's time) in post
\r
1788 mode, it can show its thinking then too. In this case your engine may
\r
1789 omit the hint move (the move it is assuming its opponent will make)
\r
1790 from the thinking lines <em>if and only if</em> it sends xboard the move in
\r
1791 the usual "Hint: xxx" format before sending the first line.
\r
1794 <h2><a name="11">11. Time control</a></h2>
\r
1797 xboard supports three styles of time control: conventional chess clocks,
\r
1798 the ICS-style incremental clock, and an exact number of seconds per move.
\r
1801 <p>In conventional clock mode, every time control period is the same.
\r
1802 That is, if the time control is 40 moves in 5 minutes, then after each
\r
1803 side has made 40 moves, they each get an additional 5 minutes, and so
\r
1804 on, ad infinitum. At some future time it would be nice to support a
\r
1805 series of distinct time controls. This is very low on my personal
\r
1806 priority list, but code donations to the xboard project are accepted,
\r
1807 so feel free to take a swing at it. I suggest you talk to me first,
\r
1812 The command to set a conventional time control looks like this:
\r
1821 The 40 means that there are 40 moves per time control. The 5 means
\r
1822 there are 5 minutes in the control. In the second example, the 0:30
\r
1823 means there are 30 seconds. The final 0 means that we are in
\r
1824 conventional clock mode.
\r
1827 <p class="version43">
\r
1828 Note that the time parameter in this command is not a pure numeric argument,
\r
1829 but in general is a character string, in order to pass the number of seconds.
\r
1830 Engines are encouraged to ignore any unexpected characters at the end of this string,
\r
1831 i.e. following the MIN or MIN:SEC specification.
\r
1832 Future protocol versions might (under control of an appropriate feature)
\r
1833 append such extra characters to this argument,
\r
1834 in order to inform the engine in advance of the time control it can expect after the current session completes.
\r
1835 E.g. "level 40 25+5 0" could mean that the engine has to play 40 moves in 25 minutes,
\r
1836 but should expect to get only 5 minutes for the entire remainder of the game after that,
\r
1837 rather than another 25 minutes for the next 40 moves.
\r
1838 When the time comes, (i.e. after the 40 moves),
\r
1839 it will be informed of the time-control change by receiving a new "level 0 5 0" command,
\r
1840 but engines with advanced time management might want to plan for this in advance.
\r
1844 The command to set an incremental time control looks like this:
\r
1852 Here the 0 means "play the whole game in this time control period",
\r
1853 the 2 means "base=2 minutes", and the 12 means "inc=12 seconds". As
\r
1854 in conventional clock mode, the second argument to level can be in
\r
1855 minutes and seconds.
\r
1859 At the start of the game, each player's clock is set to base minutes.
\r
1860 Immediately after a player makes a move, inc seconds are added to his
\r
1861 clock. A player's clock counts down while it is his turn. Your flag
\r
1862 can be called whenever your clock is zero or negative. (Your clock
\r
1863 can go negative and then become positive again because of the
\r
1867 <p class="version44">
\r
1868 The number of moves given in the level command (when non-zero) should
\r
1869 be taken as the number of moves still to do before the specified time
\r
1870 will be added to the clock, if the "level" command is received after
\r
1871 some moves have already been played.
\r
1872 The time given should be interpreted as the time left on its clock
\r
1873 (including any time left over from the previous sessions),
\r
1874 and not necessarily the time that will be added to the clock
\r
1875 after the specified number of moves has been played.
\r
1876 This is only relevant in WinBoard 4.3.xx, which might send the engine
\r
1877 "level" commands during a game,
\r
1878 just before the engine has to start thinking about the first move of
\r
1879 a new time-control session.
\r
1880 Example: if at the start of the game "level 40 60 0" was given
\r
1881 (40 moves per hour),
\r
1882 and the engine receives "level 20 22 0" just before move 41,
\r
1883 it should understand that it should do the next 20 moves in 22 minutes
\r
1884 (pehaps because the secondary session was 20 moves per 15 minutes,
\r
1885 and it had 7 minutes left on its clock after the first 40 moves).
\r
1889 A special rule on some ICS implementations: if you ask for a game with
\r
1890 base=0, the clocks really start at 10 seconds instead of 0. xboard
\r
1891 itself does not know about this rule, so it passes the 0 on to the
\r
1892 engine instead of changing it to 0:10.
\r
1896 ICS also has time odds games. With time odds, each player has his own
\r
1897 (base, inc) pair, but otherwise things work the same as in normal
\r
1898 games. The Zippy xboard accepts time odds games but ignores the fact
\r
1899 that the opponent's parameters are different; this is perhaps not
\r
1900 quite the right thing to do, but gnuchess doesn't understand time
\r
1901 odds. Time odds games are always unrated.
\r
1905 The command to set an exact number of seconds per move looks like this:
\r
1913 This means that each move must be made in at most 30 seconds. Time not used
\r
1914 on one move does not accumulate for use on later moves.
\r
1917 <h2><a name="12">12. Analyze Mode</a></h2>
\r
1919 <p>xboard supports analyzing fresh games, edited positions, and games
\r
1920 from files. However, all of these look the same from the chess
\r
1921 engine's perspective. Basically, the engine just has to respond to the
\r
1922 "analyze" command.
\r
1923 <span class="version1">
\r
1924 Beginning in protocol version 2,
\r
1925 if your engine does not support analyze mode, it should use
\r
1926 the feature command to set analyze=0.
\r
1928 The older method of
\r
1929 printing the error message "Error (unknown command): analyze" in
\r
1930 response to the "analyze" command will also work, however.
\r
1934 To enter analyze mode, xboard sends the command sequence "post", "analyze".
\r
1935 Analyze mode in your engine should be
\r
1936 similar to force mode, except that your engine thinks about what move
\r
1937 it would make next if it were on move. Your engine should accept the
\r
1938 following commands while in analyze mode:
\r
1942 <li>Any legal move, as in force mode</li>
\r
1943 <li><strong>undo</strong> Back up one move and analyze previous position.</li>
\r
1944 <li><strong>new</strong> Reset position to start of game but stay in analyze mode.</li>
\r
1945 <li><span class="version1"><strong>setboard</strong> if you have set feature setboard=1; otherwise <strong>edit</strong>. Exiting edit mode returns to analyze mode.</span></li>
\r
1946 <li><strong>exit</strong> Leave analyze mode.</li>
\r
1947 <li><strong>.</strong> Send a search status update (optional); see below.</li>
\r
1948 <li><span class="version1">
\r
1949 <strong>bk</strong> Show book moves from this position,
\r
1950 if any; see above.</span></li>
\r
1951 <li><span class="version1">
\r
1952 <strong>hint</strong> Show the predicted move from this
\r
1953 position, if any; see above.</span></li>
\r
1957 If the user selects "Periodic Updates", xboard will send the string
\r
1958 ".\n" to the chess engine periodically during analyze mode, unless the
\r
1959 last PV received began with a '(' character.
\r
1963 The chess engine should respond to ".\n" with a line like this:
\r
1967 stat01: time nodes ply mvleft mvtot <span class="version1">mvname</span>
\r
1972 <tr><td>time</td><td>Elapsed search time in centiseconds (ie: 567 = 5.67 seconds).</td></tr>
\r
1973 <tr><td>nodes</td><td>Nodes searched so far.</td></tr>
\r
1974 <tr><td>ply</td><td>Search depth so far.</td></tr>
\r
1975 <tr><td>mvleft</td><td>Number of moves left to consider at this depth.</td></tr>
\r
1976 <tr><td>mvtot</td><td>Total number of moves to consider.</td></tr>
\r
1977 <tr class="version1"><td>mvname</td><td>Move currently being considered (SAN or coordinate notation). Optional;
\r
1978 added in protocol version 2.</td></tr>
\r
1985 stat01: 1234 30000 7 5 30
\r
1986 stat01: 1234 30000 7 5 30 Nf3
\r
1993 <p>After 12.34 seconds, I've searched 7 ply/30000 nodes, there are a
\r
1994 total of 30 legal moves, and I have 5 more moves to search
\r
1995 before going to depth 8. In the second example, of the 30 legal
\r
1996 moves, the one I am currently searching is Nf3.</p>
\r
1999 Implementation of the "." command is optional. If the engine does not
\r
2000 respond to the "." command with a "stat01..." line, xboard will stop
\r
2001 sending "." commands. If the engine does not implement this command,
\r
2002 the analysis window will use a shortened format to display the engine
\r
2007 To give the user some extra information, the chess engine can output
\r
2008 the strings "++\n" and "--\n", to indicate that the current search is
\r
2009 failing high or low, respectively. You don't have to send anything
\r
2010 else to say "Okay, I'm not failing high/low anymore." xboard will
\r
2011 figure this out itself.
\r
2014 <h2><a name="13">13. Idioms and backward compatibility features</a></h2>
\r
2017 Some engines have variant interpretations of the force/go/white/black,
\r
2018 time/otim, and hard/easy command sets.
\r
2019 In order to accommodate these older engines, xboard uses these commands
\r
2020 only according to the stylized patterns ("idioms") given in this section.
\r
2021 The obsolete white and black commands
\r
2022 have historically been particularly troublesome, and it is recommended
\r
2023 that new engines set the feature colors=0 and/or ignore the commands.
\r
2031 <dd>Sent when the opponent makes a move and the engine is already
\r
2032 playing the opposite color.
\r
2036 <dd>Sent when the engine is in force mode or playing Black but should
\r
2037 switch to playing White. This sequence is sent only when White is
\r
2039 <span class="version1">
\r
2040 If you set the feature colors=0, "white" is not sent.
\r
2046 <dd>Sent when the engine is in force mode or playing White but should
\r
2047 switch to playing Black. This sequence is sent only when Black is
\r
2049 <span class="version1">
\r
2050 If you set the feature colors=0, "black" is not sent.
\r
2059 <dd>Sent when Black is on move, the engine is in force mode or playing
\r
2060 White, and the engine's clock needs to be updated before it starts
\r
2062 The initial "white" is a kludge to accommodate GNU Chess
\r
2063 4's variant interpretation of these commands.
\r
2064 <span class="version1">
\r
2065 If you set the feature colors=0, "white" and "black" are not sent.
\r
2074 <dd>Sent when White is on move, the engine is in force mode or playing
\r
2075 Black, and the engine's clock needs to be updated before it starts
\r
2076 playing. See previous idiom.
\r
2077 The initial "black" is a kludge to accommodate GNU Chess
\r
2078 4's variant interpretation of these commands.
\r
2079 <span class="version1">
\r
2080 If you set the feature colors=0, "black" and "white" are not sent.
\r
2086 <dd>Sent in sequence to turn off pondering if xboard is not sure
\r
2087 whether it is on. When xboard is sure, it will send "hard" or "easy"
\r
2088 alone. xboard does this because "easy" is a toggle in GNU Chess 4 but
\r
2089 "hard" is an absolute on.
\r
2094 To support older engines, certain additional commands from the engine
\r
2095 to xboard are also recognized. (These are commands by themselves, not
\r
2096 values to be placed in the comment field of the PGN result code.)
\r
2097 These forms are not recommended for new engines; use the PGN result
\r
2098 code commands or the resign command instead.
\r
2102 <tr><th>Command</th> <th>Interpreted as</th></tr>
\r
2103 <tr><td>White resigns </td><td>0-1 {White resigns}</td></tr>
\r
2104 <tr><td>Black resigns </td><td>1-0 {Black resigns}</td></tr>
\r
2105 <tr><td>White </td><td>1-0 {White mates}</td></tr>
\r
2106 <tr><td>Black </td><td>0-1 {Black mates}</td></tr>
\r
2107 <tr><td>Draw </td><td>1/2-1/2 {Draw}</td></tr>
\r
2108 <tr><td>computer mates </td><td>1-0 {White mates} or 0-1 {Black mates}</td></tr>
\r
2109 <tr><td>opponent mates </td><td>1-0 {White mates} or 0-1 {Black mates}</td></tr>
\r
2110 <tr><td>computer resigns </td><td>0-1 {White resigns} or 1-0 {Black resigns}</td></tr>
\r
2111 <tr><td>game is a draw </td><td>1/2-1/2 {Draw}</td></tr>
\r
2112 <tr><td>checkmate </td><td>1-0 {White mates} or 0-1 {Black mates}</td></tr>
\r
2116 Commands in the above table are recognized if they begin a line and
\r
2117 arbitrary characters follow, so (for example) "White mates" will be
\r
2118 recognized as "White", and "game is a draw by the 50 move rule" will
\r
2119 be recognized as "game is a draw". All the commands are
\r
2124 An alternative move syntax is also recognized:
\r
2128 <tr><th>Command </th><th>Interpreted as</th></tr>
\r
2129 <tr><td>NUMBER ... MOVE </td><td>move MOVE</td></tr>
\r
2133 Here NUMBER means any string of decimal digits, optionally ending in a
\r
2134 period. MOVE is any string containing no whitespace. In this command
\r
2135 format, xboard requires the "..." even if your engine is playing
\r
2136 White. A command of the form NUMBER MOVE will be ignored. This odd
\r
2137 treatment of the commands is needed for compatibility with gnuchessx.
\r
2138 The original reasons for it are lost in the mists of time, but I
\r
2139 suspect it was originally a bug in the earliest versions of xboard,
\r
2140 before I started working on it, which someone "fixed" in the wrong
\r
2141 way, by creating a special version of gnuchess (gnuchessx) instead of
\r
2146 Any line that contains the words "offer" and "draw" is recognized as
\r
2151 The "Illegal move" message is recognized even if spelled "illegal
\r
2152 move" and even if the colon (":") is omitted. This accommodates GNU
\r
2153 Chess 4, which prints messages like "Illegal move (no matching
\r
2154 move)e2e4", and old versions of Crafty, which print just "illegal move".
\r
2158 In Zippy mode, for compatibility with older versions of Crafty,
\r
2159 xboard passes through to ICS any line that begins "kibitz", "whisper",
\r
2160 "tell", or "draw". Do not use this feature in new code. Instead, use the
\r
2161 commands "tellall", "tellothers", "tellopponent", "tellics" (if needed),
\r
2162 "1/2-1/2 {COMMENT}", or "offer draw", as appropriate.
\r
2165 <p class="version1">
\r
2166 If the engine responds to the "sd DEPTH" command with an error message
\r
2167 indicating the command is not supported (such as "Illegal move: sd"),
\r
2168 xboard sets an internal flag and subsequently uses the command
\r
2169 "depth\nDEPTH" instead, for the benefit of GNU Chess 4. Note the
\r
2170 newline in the middle of this command! New engines should not rely on
\r
2174 <p class="version1">
\r
2175 If the engine responds to the "st TIME" command with an error message
\r
2176 indicating the command is not supported (such as "Illegal move: st"),
\r
2177 xboard sets an internal flag and subsequently uses the command "level
\r
2178 1 TIME" instead, for the benefit of GNU Chess 4. Note that this is
\r
2179 not a standard use of the level command, as TIME seconds are not added
\r
2180 after each player makes 1 move; rather, each move is made in at most
\r
2181 TIME seconds. New engines should not implement or rely on this
\r
2185 <div class="version1">
\r
2187 In support of the -firstHost/-secondHost features, which allow a chess
\r
2188 engine to be run on another machine using the rsh protocol, xboard recognizes
\r
2189 error messages that are likely to come from rsh as fatal errors. The following
\r
2190 messages are currently recognized:
\r
2194 <li>unknown host</li>
\r
2195 <li>No remote directory</li>
\r
2196 <li>not found</li>
\r
2197 <li>No such file</li>
\r
2198 <li>can't alloc</li>
\r
2199 <li>Permission denied</li>
\r
2203 <p class="version1">
\r
2204 ChessBase/Fritz now implements the xboard/winboard protocol and can use
\r
2205 WinBoard-compatible engines in its GUI. ChessBase's version of the
\r
2206 protocol is generally the same as version 1, except that they have
\r
2207 added the commands <strong>fritz</strong>, <strong>reset</strong>, and
\r
2208 <strong>ponder</strong>, and the edit subcommands
\r
2209 <strong>castle</strong> and <strong>ep</strong>. If you want your
\r
2210 engine to work well with the ChessBase/Fritz GUI, you may need to
\r
2211 implement these additional commands, and you should also be aware of
\r
2212 the peculiar way that ChessBase uses the protocol. See their <a
\r
2213 href="http://www.chessbase.com/Products/engines/winboard/tech.htm"
\r
2214 >web page</a> for documentation.
\r
2217 <p class="version1">
\r
2218 ChessMaster 8000 also implements version 1 of the xboard/winboard
\r
2219 protocol and can use WinBoard-compatible engines. The original
\r
2220 release of CM8000 also has one additional restriction: only pure
\r
2221 coordinate notation (e.g., e2e4) is accepted in the move command. A
\r
2222 patch to correct this should be available from The Learning Company
\r
2223 (makers of CM8000) in February 2001.
\r