Closed GoogleCodeExporter closed 9 years ago
Original comment by carsonday
on 4 Dec 2009 at 5:06
Original comment by carsonday
on 4 Dec 2009 at 5:07
Original comment by carsonday
on 5 Dec 2009 at 2:46
Perhaps it would be easier to maintain a "chess engine wizard" which downloads
the
appropriate Toga installer, allows the user to run the installer, and after
installation the user can specify where the engine was installed, what
parameters to
run it with, etc.
Original comment by DandyD...@gmail.com
on 6 Dec 2009 at 9:08
Original comment by carsonday
on 9 Dec 2009 at 9:17
Original comment by carsonday
on 9 Dec 2009 at 9:18
Original comment by carsonday
on 14 Dec 2009 at 10:24
Original comment by carsonday
on 16 Dec 2009 at 1:52
[deleted comment]
Comming soon. It is looking like there might be different open source engines
bundled with different installs.
Original comment by carsonday
on 24 Dec 2009 at 2:00
Wouldn't this sort of defeat the purpose of adding in your own UCI engines?
Why not
let the user define where the UCI engine is located (as is the case now in
v.93) and
leave it at that, instead of dealing with the headaches of adding in different
engines for different platforms?
fjgrafe: Maybe this is a bug in Raptor w/ how it handles UCI engines? You need
to do
"ucinewgame", position fen <fenstring> moves <moves in UCI long algebraic> --
then
"go infinite" (and if the UCI engine supports it, UCI_AnalyseMode must be
turned on.
Original comment by firepot...@gmail.com
on 25 Dec 2009 at 11:33
Original comment by carsonday
on 25 Dec 2009 at 11:34
The problem is many users have no idea what a UCI engine is.
I want to bundle one with each release so its just preconfigured and easy to
use. If
they want to add other ones they can. Raptor even lets you pick the one to use
when
you turn analysis on.
Original comment by carsonday
on 26 Dec 2009 at 12:30
Also its not a bug there is a place to add the "go command string". Look in the
engine properties. You can even bring them up when analysis is on.
It is a fruit/togaII bug imho. They should support go infinite its part of the
UCI spec.
Original comment by carsonday
on 26 Dec 2009 at 12:32
I see, well, you should then try to use Stockfish instead of Toga -- Stockfish
supports Chess960, can also specify the number of threads to use (currently Toga
requires a special patch/hack to do this), and is also the strongest open source
chess engine out there (check CCRL).
Original comment by firepot...@gmail.com
on 26 Dec 2009 at 1:03
Oh and also Stockfish supports "go infinite". From what I can tell Toga also
supports
"go infinite" -- I'm using Toga II 1.4.1SE (patched with the threads option
hack --
see http://alpha.uhasselt.be/Research/Algebra/Toga/linux_versions/ and
http://alpha.uhasselt.be/Research/Algebra/Toga/linux_versions/patch_posix.p for
the
patch).
Original comment by firepot...@gmail.com
on 26 Dec 2009 at 1:14
Oops, wrong link: this is the patch:
http://alpha.uhasselt.be/Research/Algebra/Toga/linux_versions/thread_patch.p
Original comment by firepot...@gmail.com
on 26 Dec 2009 at 1:18
Part of the problem with all of this is i need a bunch of versions:
Mac OS X:
need g4/g5 PPC version.
need intel version.
need intel x86_64 version.
Linux
need ppc version
need x86 version
need x86_64 version
solaris:
need x86 version
need SPARC version
windows:
need x86 version
(not sure what i need for windows 7)
Currently I can load up and compile:
Mac OS X ppc version,
linux x86_64
linux x86
Possibly windows if i am in a good mood.
Original comment by carsonday
on 26 Dec 2009 at 1:50
I'm looking at Stockfish 1.5.1's Makefile right now in front of me, and it has
build
targets for OSX PPC32/PPC64, x86 x86_64, etc. I'm quite sure Stockfish is one
of the
most portable UCI engines out there.
Glaurung (v. 2.2), which Stockfish is based on, came with binaries for Linux,
OS X,
and Windows. Link: http://www.glaurungchess.com/glaurung22.zip. So, Stockfish
shouldn't be any more different...
Original comment by firepot...@gmail.com
on 26 Dec 2009 at 5:49
ok I will check it out.
btw just for people who read this I added this to the FAQ:
I can't get TogaII or Fruit to work. They dont do anything.
TogaII and fruit do not accept the 'go infinite' command. You can fix this.
Open a game and click the
calculator icon to bring up the chess engine analysis. After that click the
settings button. Remove the infinite
parameter from the Go Analysis Parameters Field at the top and click save. You
should now recieve analysis.
Original comment by carsonday
on 26 Dec 2009 at 10:00
Here is Jim Ablett's page, which has the source for Stockfish (1.6 just came
out,
apparently): http://homepages.tesco.net/henry.ablett/jims.html
Original comment by firepot...@gmail.com
on 26 Dec 2009 at 7:36
RE: the "go infinite" command -- Just wanted to add, Toga II 1.4.1SE does
accept the
"go infinite" command. I tried it out by hand in the console, and it works. I
have
the source for Toga II 1.4 beta5c as well in front of me, and it also has code
to
handle the "infinite" command, so that version should also work as well... Did
you
try it out manually from the console?
Original comment by firepot...@gmail.com
on 26 Dec 2009 at 8:50
Hmm, I think there's a bug here, specifically:
UCIEngine.java, @ line 276:
if (StringUtils.isNotBlank(options)) {
send("go");
} else {
send("go " + options);
}
Raptor is sending the "go" command by itself. This, apparently, is exactly the
reason
for the bug reported in Issue 413 (if you type just "go" to Glaurung/Stockfish,
it
only analyzes to depth 2 -- this is after sending "setoption name OwnBook value
false", so it's not book-related).
I haven't looked at all of the code, but, sending "go" by itself should not be
the
default. I think "go infinite" should be the default command.
Original comment by firepot...@gmail.com
on 26 Dec 2009 at 9:05
[deleted comment]
Yes its a bug it should be isBlank. I will fix it in the next release.
Original comment by carsonday
on 27 Dec 2009 at 4:38
Original comment by carsonday
on 27 Dec 2009 at 4:38
Fixed in .94b along with the start button being mislabled in some situations.
Original comment by carsonday
on 27 Dec 2009 at 7:54
Original comment by carsonday
on 30 Dec 2009 at 7:59
I've tested Stockfish 1.6 with some positions, but currently it is probably not
what
we want because it seems to give a dead even "0" score for most positions
(probably a
bug with how it displays the eval). Rather, I recommend Glaurung 2.2
http://www.glaurungchess.com/glaurung22.zip for the time being (it comes with
binaries for Linux, Windows, and OS X -- for both 32 and 64-bit) right now as it
gives a more accurate score immediately upon analyzing a position (and it's
still
stronger than any Toga variant out there).
Original comment by firepot...@gmail.com
on 31 Dec 2009 at 12:44
Scratch that -- the authors of Stockfish have released a new version, 1.6.2,
which
fixes this bug. Link to Stockfish:
http://homepages.tesco.net/henry.ablett/jims.html
Original comment by firepot...@gmail.com
on 31 Dec 2009 at 10:01
Here is a link to probably the most up-to-date rating list of chess engines
(commercial, free, open source) out there:
http://www.computerchess.org.uk/ccrl/4040/
Notice how Stockfish dominates in the open source lists.
Original comment by firepot...@gmail.com
on 31 Dec 2009 at 10:03
[deleted comment]
Looks like the original question is answered: Raptor will not be bundled with
an engine.
Original comment by evil...@gmail.com
on 7 Nov 2014 at 9:42
FYI, I think the Stockfish zero evaluation bug was fixed in the last few months.
Original comment by DandyD...@gmail.com
on 8 Nov 2014 at 7:00
Original issue reported on code.google.com by
carsonday
on 4 Dec 2009 at 5:06