Polyglot queries...

Archive of the old Parsimony forum. Some messages couldn't be restored. Limitations: Search for authors does not work, Parsimony specific formats do not work, threaded view does not work properly. Posting is disabled.

Polyglot queries...

Postby Roger Brown » 18 Apr 2004, 06:14

Geschrieben von:/Posted by: Roger Brown at 18 April 2004 07:14:04:

Hello Fabien,
I am attempting to use your adapter with Shredder 6, 7.04 and Shredder 8. In a brief match consisting of 9 games at 5 minutes and 2 seconds each, there were three losses on time.
Even more important, there were two Shredders and four Polyglots hanging around in memory at the end of the tournament. I had to use Taskinfo to manually shut them down.
I am using Winboard through Galis with Bookthinker. I place /first Protocol 1 Polyglot in the command line of Galis to call Polyglot when Bookthinker - the book playing engine for Shredder -
runs out of book moves.
For every executable there are two instances of Polyglot which is also confusing to me.
Any help would be helpful.
My understanding is that the UCI executable should quit if the EOF command is sent, correct?
Later.


Winboard debug:

Interrupting first
740180 >first : time 5439
otim 4686
740180 >first : g2h2
740290 XBOARD 1/2-1/2 {Draw by repetition}
XBOARD 1/2-1/2 {Draw by repetition}
ENGINE quit
ENGINE EOF
XBOARD 0-1 {Black mates}
XBOARD 0-1 {Black mates}
ENGINE quit
ENGINE EOF
ENGINE stop
ENGINE quit
ENGINE EOF
POLYGLOT THINKING -> WAITING
< XBOARD result 1/2-1/2 {Draw by repetition}
POLYGLOT GAME END
POLYGLOT WAITING
< XBOARD force
POLYGLOT WAITING
< XBOARD quit
POLYGLOT *** QUIT ***


Shredder 7.04 log extract

POLYGLOT MOVE Qc2#
POLYGLOT WAITING
< XBOARD result 0-1 {Black mates}
POLYGLOT GAME END
POLYGLOT WAITING
< XBOARD force
POLYGLOT WAITING
< XBOARD quit
POLYGLOT *** QUIT ***

Shredder 8 log extract

< XBOARD result 1/2-1/2 {Draw by repetition}
POLYGLOT GAME END
POLYGLOT STOP SEARCH
POLYGLOT WAITING
< XBOARD force
POLYGLOT WAITING
< XBOARD quit
POLYGLOT *** QUIT ***
Roger Brown
 

Re: Polyglot queries...

Postby Fabien Letouzey » 19 Apr 2004, 10:43

Geschrieben von:/Posted by: Fabien Letouzey at 19 April 2004 11:43:26:
Als Antwort auf:/In reply to: Polyglot queries... geschrieben von:/posted by: Roger Brown at 18 April 2004 07:14:04:

Hello Fabien,
I am attempting to use your adapter with Shredder 6, 7.04 and Shredder 8. In a brief match consisting of 9 games at 5 minutes and 2 seconds each, there were three losses on time.
Even more important, there were two Shredders and four Polyglots hanging around in memory at the end of the tournament. I had to use Taskinfo to manually shut them down.
I am using Winboard through Galis with Bookthinker. I place /first Protocol 1 Polyglot in the command line of Galis to call Polyglot when Bookthinker - the book playing engine for Shredder -
runs out of book moves.
For every executable there are two instances of Polyglot which is also confusing to me.
My understanding is that the UCI executable should quit if the EOF command is sent, correct?
Send me full log files about a single game with loss on time. Delete the log files before the experiment, and please compress with Zip before sending it.
I knew about Shredder staying in memory, but I have never been reported that PolyGlot also stayed in memory. This is serious; I also need log files of course (not edited).
Try using the xreuse option with WinBoard (re-launch the engine after each game). This is not a fully acceptable solution (PolyGlot should work in all cases), but it might help me understanding what is going on. It might also be a work around for users.
I'm not sure which combination you used. But (WBTM +) WinBoard + BookThinker + PolyGlot + engine is quite a thing!!! Try to reproduce problems with the simplest combination.
This is due to CYGWIN.DLL internals. Consider it as "normal" behaviour. One PolyGlot instance should not do anything at all (hopefully not using memory either).
This is a tricky issue. Neither xboard or UCI specs mention the behaviour on receiving EOF. It is good programming practice to quit then (as I mentionned in a recent CCC post), but it is useful only in problematic cases.
So in short it would be very helpful if engines did that, but I can't blame them on not doing it. Unfortunately I suspect that Shredder and SOS (and many many others) don't care about it.
To make things clear; it would help if they did, but that is not the source of a problem.
Note that in the log files you posted, "quit" was always sent by PolyGlot first. I don't think those log files correspond to problem cases.
Fabien.
Fabien Letouzey
 


Return to Archive (Old Parsimony Forum)

Who is online

Users browsing this forum: No registered users and 23 guests

cron