-
I've just started to look at the analyze command and it looks very useful.
It seems the 'var' command with its variations can show a move-sequence. Can it also show move-trees?
What I'm thinking of is a command that would expect in return something that looks like SGF: (B[aa]C[Score: 0.8];W[ab]C[Score: -0.3](W[ac]C[Score: -0.2]))
and it would insert a tree of ghost stones that can be...
2008-10-29 11:37:42 UTC in GoGui
-
Since you say the games don't need to be numbered properly I figured I could accomplish what I want simply by concatenating all the .dat file. So I wrote a simple shell-script using 'cat' to merge the different .dat files before running the analyze command.
That works fine, so I got what I wanted. Thanks.
2008-10-29 11:20:57 UTC in GoGui
-
Well, adding machines would be nice, but not high on my list right now. I simply want to be able to make full use of the single machine I have.
Your game-A-1.sg, game-B-1.sgf scheme etc. sounds possible too. But the -analyze parameter takes a single .dat file as parameter. So I'd still like to have a way to merge data from several .dat files so that all the statistics get added up.
Mark.
2008-10-28 22:21:35 UTC in GoGui
-
How about the following, fairly simple scheme:
Add two parameters, one with the number of 'parallel' spawns and one parameter with it's starting number. A twogtp instance will then write SGF files with 'n' intervals, where 'n' is the number of processors, starting at the starting number.
So for example for 8 processors, the first twogtp spawned will write game-0.sgf, game-8.sgf...
2008-10-28 21:25:25 UTC in GoGui
-
I've now been using twogtp for more than a week and I like it a lot.
There's one thing that I think would be a nice addition: mulltiple process support. For example, you pass a parameter with the number of processors and it spawns that many pairs of programs. You'll probably have to make sure the file-numbering remains correct, but that should not be that hard.
On my 8-core computer that...
2008-10-28 15:52:23 UTC in GoGui
-
tesujisoftware committed revision 3 to the Plug-and-Go SVN repository, changing 1 files.
2008-10-28 13:43:46 UTC in Plug-and-Go
-
tesujisoftware committed revision 2 to the Plug-and-Go SVN repository, changing 1 files.
2008-10-28 13:41:16 UTC in Plug-and-Go
-
tesujisoftware committed revision 1 to the Plug-and-Go SVN repository, changing 1 files.
2008-10-28 12:27:43 UTC in Plug-and-Go
-
tesujisoftware registered the Plug-and-Go project.
2008-10-27 16:06:29 UTC in Plug-and-Go
-
You're right, when I use the -verbose option it shows that this is what my bot returns. However, by bots don't support final_score (yet) and don't return it in the list of commands. But the query for the final_score is still sent. Is it required? Or do I need a referee program when both bots don't implement final_score? I had kind of hoped gogui or twogtp already did that, it would save a little...
2008-10-21 19:02:30 UTC in GoGui
