Forum: rec.games.go Subject: Re: Linux Date: 10/22/1998 Author: Daniel Bump David Fotland (fotland@cup.hp.com) wrote: > Not so strange. Most go programmers would rather spend their time > improving the playing algorithm than writing another user interface. > And good go algorithms take a long time to write. Most of the best > ones are over 10 years old and Linux was not the natural choice to > start with 10 years ago :) It seems possible to me that you could do a Linux port. Since you have a HP-UX version of Many Faces, perhaps it could be recompiled for Linux fairly easily. It would be a few versions behind the DOS/Windows version, but people would buy it anyway, if like me and lots of other people they almost never boot Windows. How about it? > Also, the top programs are all for sale, and if you want to sell a > lot of copies you will write for Win-95 or DOS, not for Linux. > > David Fotland That seems to be changing. Intel just bought into Red Hat Linux (http://www.redhat.com/news/news-details.phtml?id=163). Oracle, Informix and Corel Wordperfect have all recently jumped to Linux. When will Many Faces join this list? Daniel Bump http://match.stanford.edu/bump/go.html Forum: rec.games.go Subject: Re: Linux Date: 10/23/1998 Author: David Fotland : On 22 Oct 1998 13:40:34 -0700, Daniel Bump wrote: : > : >It seems possible to me that you could do a Linux port. Since you : >have a HP-UX version of Many Faces, perhaps it could be recompiled for : >Linux fairly easily. It would be a few versions behind the DOS/Windows : >version, but people would buy it anyway, if like me and lots of other : >people they almost never boot Windows. : > It would be a lot of work to do a Linux product. The HP-UX version is old and buggy, and not something I feel comfortable charging money for. With installing and learning Linux, buying and learning a new set of compilers/editors, improving the GUI to get it up to commercial standards, recruiting beta testers, going through a beta cycle, paying the artists for the new packaging, it would take me 3 to 6 months of working on nothing else, and cost several thousand dollars. I doubt I would sell more than a handful of copies, so to me it is clear that the effort is not justified. I spend more time than I would like updating and supporting the Windows GUI already. There is a lot of literature available on computer go algorithms, so I'd encourage Linux users to put together a collaborative effort to write a Linux go program. It's certainly a lot easier than writing an operating system :) David Fotland Forum: rec.games.go Subject: Re: Linux Date: 10/26/1998 Author: Daniel Bump David Fotland (fotland@cup.hp.com) wrote: > There is a lot of literature available on computer go algorithms, so > I'd encourage Linux users to put together a collaborative effort to > write a Linux go program. It's certainly a lot easier than writing > an operating system :) This seems like a good idea. I'd like to point out that the user interface code for a Go program is already in place. Bill Shubert's client CGoban 1.9.2 understands the Go Modem Protocol which was developed by Bruce Wilcox with input from David Fotland, Anders Kierulf and others, according to the history in the file protocol.Z, available from ftp://igs.nuri.net/Go/programs/. Any Go program *should* use this protocol since it is standard. (For example the rules of the FOST Cup encourage entrants to use this protocol.) Thus the user interface for any Go program can be done entirely through CGoban. The programmer can concentrate on the real issues without worrying about drawing stones, resizing the board and other distracting issues. To write a Go program, all you have to do is to download Bill Shubert's goDummy program from http://www.hevanet.com/wms/comp/cgoban/ and rewrite the function playGame in the file dummy.c. Daniel Bump http://match.stanford.edu/bump/go.html Forum: rec.games.go Subject: Gnugo and CGoban [was: Linux] Date: 11/02/1998 Author: Daniel Bump Gnugo (http://www.gnu.ai.mit.edu/software/gnugo/gnugo.html) is a free Go program which can run under Linux. It is not very strong, but anyone is free to try to enhance it. I am not sure if there is a current maintainer---enhancing gnugo is on the Free Software Foundations tasklist (http://www.gnu.ai.mit.edu/prep/tasks_toc.html). Recently in this newsgroup I made the following assertions: > I'd like to point out that the user interface code for a Go program is > already in place. Bill Shubert's client CGoban 1.9.2 understands the Go > Modem Protocol which was developed by Bruce Wilcox with input from David > Fotland, Anders Kierulf and others, according to the history in the file > protocol.Z, available from ftp://igs.nuri.net/Go/programs/. Any Go > program *should* use this protocol since it is standard. (For example > the rules of the FOST Cup encourage entrants to use this protocol.) > Thus the user interface for any Go program can be done entirely through > CGoban. The programmer can concentrate on the real issues without > worrying about drawing stones, resizing the board and other distracting > issues. To prove this claim, I adapted Gnugo to run under CGoban. This is a much more enjoyable way of running it than its default ASCII interface. You get a beautiful board, entering your move with a mouse click. You can even make gnugo play against itself, or against William Shubert's goDummy program. The adaptations which I made were fairly minor, but depend upon files gmp.c and gmp.h which I borrowed from goDummy. I would like to make it clear that the hard work here was done not by myself, but by Bruce Wilcox, David Fotland and especially William Shubert, author of the file gmp.c. For more information, or to download, go to http://match.stanford.edu/bump/gnugo.html Daniel Bump http://match.stanford.edu/bump/go.html bump@math.stanford.edu Forum: rec.games.go Subject: Baduki 0.1 - Computer Baduk(Go) Program - is available Date: 01/05/1999 Author: Lim Jaebum Baduki is Computer Baduk(Go) Program for Linux. Features: Load/Save/Analyze SGF file. Suport GMP(go modem protocol) Easy to debug. (i hope) HomePage: http://soback.kornet21.net/~artist/baduk/baduki.html -- Artist Forum: rec.games.go Subject: Liberty, a Linux Go Program Date: 01/06/1999 Author: Daniel Bump Liberty is a Go program which runs on Linux machines. It should be easy to port to other Unixes. It requires CGoban and X-Windows. Liberty is under the Gnu General Public License, which guarantees that the source code will remain free. Liberty is derived from Gnugo 1.2, written by Man Lung Li and published by the Free Software Foundation. However I have made enough changes that I feel it should be regarded as a new program. It can give Gnugo 1.2 a nine stone handicap. You can download Liberty from: ftp://match.stanford.edu/pub/liberty-1.0.tar.gz - or - http://match.stanford.edu/bump/liberty.html Recently Lim Jaebum (``Artist'') released his own program baduki, based on Xamigo, at: http://soback.kornet21.net/~artist/baduk/baduki.html It is interesting to play these two programs against each other. They seem to be of comparable strength---both win some games. The games are bloody and exciting. Huge groups die. Both programs are far weaker than commercial programs such as Many Faces of Go (http://www.smart-games.com/manyfaces.html). I retained Gnugo's basic architecture. Beginning with Gnugo 1.2 I made the following changes: (1) Implemented Go Modem Protocol. It is fairly easy to adapt Gnugo itself to use this, and I have already issued a toolkit for this at http://match.stanford.edu/bump/gnugo.html. (2) Rewritten the life and death modules. This is the most important change which I've made. Unlike Gnugo, Liberty does some reading and can even read a ladder. It tries to determine whether a group can get three liberties. (3) I retained Gnugo's pattern matcher but expanded the number of patterns from 24 to 228. Moreover I've implemented a system for maintaining these patterns, with a ``human readable'' file patterns.db and a C program that compiles that into patterns.h. Even if you are not a programmer you can improve this program by modifying the pattern database. Fundamental improvements also seem very feasible. I am hoping other people will get excited about trying to make a stronger free Go program. Daniel Bump Forum: rec.games.go Subject: Re: Liberty, a Linux Go Program Date: 01/07/1999 Author: Daniel Bump > >The name of this program could cause some confusion. Jan van der Steen has > >written a go-editor for the Atari ST computer also called Liberty. > > > >I sugest you change the name > > > Do you really think that there are enough Atari ST left to cause > confusion...?! > ;-)) Jan tells me he may port Liberty (the editor) to Linux. So I suppose we need a new name for Liberty (the program). Daniel Bump Forum: rec.games.go Subject: Gnu Go Beta versions Date: 03/07/1999 Author: Daniel Bump Beta versions of a future Gnu Go 2.0 can be found at: http://www.gnu.org/software/gnugo/beta.html Recent versions are substantially stronger than the programs Liberty 1.0 and Freedom 1.8 which were circulated earlier. The Gnu Go beta versions are joint work of myself and David Denholm. They run under GNU/Linux using CGoban. Daniel Bump Forum: rec.games.go Subject: GNU Go 2.0 is released Date: 04/20/1999 Author: Daniel Bump GNU Go 2.0 is now available. See http://www.gnu.org/software/gnugo/gnugo.html It is quite a bit stronger than GNU Go 1.2. How strong is the new GNU Go? Thanks to Gunnar Farneback, beta versions have been playing on NNGS and after around 200 games the rating is stable between 16k* and 17k*. See: http://nngs.cosmic.org/cgi-bin/ngraph.cgi?n=gnugo Development versions of GNU Go are also available. They are likely to be stronger than GNU Go 2.0 but are not guaranteed to be stable. these may be found at http://www.gnu.org/software/gnugo/devel.html The current GNU Go maintainers are myself and David Denholm. Others contributing to or interested in the development of GNU Go are Gunnar Farneback, Nils Lohner and (original GNU Go author) Man Lung Li. You may reach all of us by writing to gnugo@gnu.org. Daniel Bump