version 1.4.53b

[polyglot.git] / ChangeLog

=========1.4.53b================\r
- Multipv code by HGM. \r
=========1.4.52b================\r
- Reexport SyncStop. It is apparently necessary for the Aquarium Book Adapter.\r
=========1.4.51b================\r
- Support for new nps command. Code provided by Harm Geert Muller. \r
=========1.4.50b================\r
- Change in architecture. Persistence is disabled. The user is required to push the "Save" button explicitly to save the settings. What is retained is that in case PG is invoked with -noini (or as a special exception for WB 4.4.0 when the config files are polyglot_1st.ini or polyglot_2nd.ini) then PG uses a default config file whose name is derived from the engine name. \r
- New option: OnlyWbOptions (default: true). Restrict the options that PG sends to those that are potentially useful for WinBoard.\r
=========1.4.47b================\r
- Try to detect if the engine is not an UCI engine.\r
=========1.4.46b================\r
- The default persistence directory on Windows is now _PG. The names of the engine specific ini files are mangled so as not to contain spaces. \r
- PG now tries to guess if a string option is really a file or a path (like in the Arena GUI). \r
=========1.4.45b================\r
- The engine specific config files (<myname>.ini) are now stored in a separate directory. $HOME/.polyglot on Linux and "./Polyglot Settings" on Windows. \r
- The engine specific config files are now complete config files. So they can be used in a stand alone fashion.\r
- Implemented a minor hack to make xboard show error messages during startup. \r
=========1.4.44b================\r
- "Implementation" of new buttons and controls.\r
- The "Persist" option is now persistent. That is if you turn of "Persist" then as a special case PG will remember this.\r
- There is now a "reset" button which restores the default options. The code is commented out since WB does not support it yet.\r
=========1.4.42b================\r
- SaveSettingsOnExit/SaveFile renamed as Persist/PersistFile.\r
- Many bug fixes in the persistence feature.\r
- The PG_<myname>.ini file now contains only options that have been changed. So the resulting ini files are much smaller. This makes this version of PG usable on WB 4.3.15 as now the engine specific ini file will be empty (the user has no way to change options).\r
=========1.4.41b================\r
- Some segfaults fixed in case of corrupt config files.\r
- Persistence of options. If the option SaveSettingsOnExit is true (the default) PolyGlot now saves its options in a file whose name is given by the option SaveFile which by default is equal to the engine name, prefixex by PG_ and suffixed by ".ini".\r
=========1.4.39b================\r
- Refactoring of config file parsing\r
- Fix crash bug in epd-test\r
- Do not export Chess960. This is handled completely internally. \r
- Do not export UCI_Opponent and UCI_Chess960. These are handled internally.\r
- PG can now also be comfortably configured through command line options (i.e. without using a config file). \r
- Allow quotes in configfile (as in Fonzy Bleumer's PG). \r
=========1.4.37b================\r
- Two new options: BookDepth and BookTreshold. BookDepth limits the number of moves the book is used and BookTreshold sets the minimum weight a book move should have.\r
=========1.4.36b================\r
- Bugfix: option = 0/1 was translated incorrectly.\r
- Default node count is 1.\r
- EOF received from the engine should now be logged correctly (i.e. after the buffer is empty).\r
- Bugfix: Nasty buffer overflow in the macro CONSTRUCT_ARG_STRING.\r
- Bugfix: pipex_writeln (win32) was logging at the wrong spot. \r
- Log board when an illegal move is received. \r
=========1.4.35b================\r
- New WbWorkAround for silly bug in WB. Depth <=1 clears the engine output window. Why should an engine not be allowed to send info at depth one?\r
=========1.4.34b================\r
- Warning: this release is less well tested as usual\r
- Allow spaces around "=" in "option =" commands.\r
- WbWorkArounds is now true by default. It decapitalizes Draw when this appears in engine option. \r
- Trying to set a non existent UCI option gives an error. \r
- PG will now resign in case of an illegal engine move. \r
- More code has been put in "pipex".\r
- info strings and unrecognized info lines are relayed to the GUI.\r
=========1.4.33b================\r
- New option: "WbWorkArounds" to work around bugs in WB. Currently it blocks options which contain "Draw" in their name from being sent to the GUI. Engines that have such options are Rybka and HIARCS. \r
- Bugfix for ASSERT failure in bright. Now another ASSERT failure has appeared which seems to be bright's fault however. \r
=========1.4.32b================\r
- Bugfix: the default setting for RepeatPV had become false again...\r
=========1.4.31b================\r
- Bugfix: improper handling of vararg function.\r
=========1.4.30b================\r
- Some more meaningful error messages added.\r
- Some buffer overflow checks added.\r
- Simplification of strange signal kludge in gui.c. \r
- Removal of obsolete code in engine.c and gui.c. Platform specific code has been abstracted and pushed into pipex_win32.c and pipex_posix.c\r
- The format of the version number has changed once again. Debian was unhappy with the previous one.\r
=========1.4b29================\r
- Conversion from C++ to C (suggested by E.M.)\r
- More refactoring. The win32 and posix I/O now have a uniform interface (see pipex.h). \r
=========1.4b28================\r
- Some comments added to explain the exact behaviour of some of the public functions in pipe.cpp. \r
- LineInput now returns a bool which is FALSE in case of EOF. \r
=========1.4b27================\r
- Option "ScoreWhite" : report score from white's point of view (suggested by E.M.).\r
- Option "KibitzInterval" : try to wait this many seconds between kibitzes (suggested by E.M.)\r
=========1.4b26================\r
- Contraction of version number.\r
- Log if SetProcessAffinityMask is not available (suggested by E.M.).\r
- pipe.cpp : Make PipeStruct a proper class with private/public members.\r
- Check for EOF in GUI input.\r
- pipe.cpp : delay reporting EOF until buffer is empty.\r
- Messages from the Engine to PG were not logged in the windows version.\r
=========1.4w10UCIb24===========\r
- Portability fixes for WIN9X (Eric Mullins).\r
- Portability fixes for MSVC++ (Eric Mullins).\r
- Default setting of RepeatPV is now true.\r
- Do not read data if input buffer is full (Windows).\r
=========1.4w10UCIb22===========\r
- Polyglot now completely poll free...\r
=========1.4w10UCIb21===========\r
- Crash bug in memory command fixed\r
- Small optimizations\r
=========1.4w10UCIb20===========\r
- More refactoring.\r
- No more polling for GUI input\r
=========1.4w10UCIb18===========\r
- Two new utilities: info-book and dump-book.\r
=========1.4w10UCIb17===========\r
- More refactoring. main.cpp is now #ifdef _WIN32 free.\r
- The main loop (previously in adapter.cpp) is now in its own file mainloop.cpp.\r
- adapter.cpp has been renamed into xboard2uci.cpp. It is #ifdef _WIN32 free.\r
- EOF from the engine no longer kills PG. It should now\r
be possible to close an engine and reopen another one (if that ever became useful) \r
=========1.4w10UCIb16===========\r
- xboard options commands\r
- correction of handling of combo boxes in UCI protocol\r
- "gui-mode". It is now possible to run PG without config file\r
- The polyglot options are not exported, so they are now settable by the GUI\r
=========1.4w10UCIb15===========\r
- Due to refactoring time stamps in Windows would be written twice. \r
This has been fixed. \r
- More refactoring; the number of #ifdef _WIN32's has again been\r
drastically reduced.\r
=========1.4w10UCIb14===========\r
- option to alter priority now also works on windows\r
(Eric Mullins)\r
- refactoring so that posix and win32 part share more\r
code\r
- a number of potential busy loop scenarios have been eliminated \r
in the Windows code (Windows uses polling)\r
- the return value of CreateProcess is now checked\r
=========1.4w10UCIb13===========\r
- perft mode exposed.\r
=========1.4w10UCIb12===========\r
- Replaced %lld everywhere by S64_FORMAT.\r
=========1.4w10UCIb11===========\r
- Crash bug fix (wrong parentheses in instruction that checked for help\r
option).\r
=========1.4w10UCIb10===========\r
- It is now possible to do "make rpm" to make an rpm binary package. \r
- Trivial bug fixes.\r
- Documentation added for epd-test functionality (man page).\r
- Cosmetic cleanups of output of epd-test.\r
- "polyglot --help" output added.\r
- Added polyglot book format specification to distribution.\r
=========1.4w10UCIb9===========\r
- Trivial cosmetic fixes\r
- Some engines do not support the "Hash" options.\r
So we do not send memory=1 in that case. \r
=========1.4w10UCIb8===========\r
- Support for egtpath command in xboard protocol\r
=========1.4w10UCIb7===========\r
- Debianization: it is now possible to do "make deb" to build a Debian/Ubuntu\r
package.\r
=========1.4w10UCIb6===========\r
- UCI GUI support added. This makes it possible to use PG as a book engine\r
on GUI's like ChessBase and Arena which normally only support their\r
own proprietary format. \r
\r
- Support for new xboard "cores N" and "memory N" commands. See here\r
for a complete description\r
\r
        http://home.hccnet.nl/h.g.muller/engine-intf.html\r
\r
- Some work arounds for engines like Toga which only allow some settings\r
to be set before the first "isready" command. \r
\r
- Autotooled distribution.\r
\r
- Proper manual page. The README file is generated automatically from this. \r
\r
- More sane naming of README files.\r
\r
- If BookLearn is false (the default) PG will open the book read only.\r
\r
- Windows and non-Windows log file formats are now the same.\r
\r
- Proper ChangeLog (you are reading it!).\r
=========1.4w10================\r
fixed analysis output for toga\r
=========1.4w9=================\r
fixed bug in 1.4w8\r
added RepeatPV workaround\r
        fixed disappearing output in engine-output window while in multipv mode\r
        when an engine sends its move to polyglot,\r
        polyglot normally repeats the last pv string(which also contains score,depth and time usage) it got from the engine.\r
        Some engines however do not send a new pv string just before sending the move and the now old pv string\r
        find might confuse debugtools that parse the winboard debug files.\r
        Adding "RepeatPV = false" to the [POLYGLOT] section of the ini file stops this repetition.\r
 \r
=========1.4w8=================\r
fixed multipv output\r
note that the pv with the worst score will be on top of the engine-output window.\r
added timestamp in logfile (Jaap Weidemann)\r
\r
=========1.4w7=================\r
compiles under linux/unix again\r
=========1.4w6=================\r
access to winboard draw/drawoffer and resign\r
\r
        1:to activate draw offers the engine has \r
           to define the "UCI_DrawOffers" parameter with the 'option" command at startup. \r
        2:to offer a draw or accept a draw offer:just send "info string DrawOffer" to polyglot. \r
        3:if winboard sends "draw" polyglot sends "setoption DrawOffer draw" to the engine.\r
        4.to resign: send "info string Resign" to polyglot.\r
        please check the winboard documentation for the draw/drawoffer and resign commands.\r
                \r
=========1.4w5:================\r
Fixed errors in SyncStop handling.\r
book building: the error message now also contains the game number \r
added Affinity option:\r
In the [PolyGlot] section(ini file):\r
- "Affinity" mask\r
mask is a bit vector in which each bit represents the processors that a process is allowed to run on.\r
\r
some minor bugs fixed\r
\r
checks if child did really quit.\r
\r
\r