Winboard Forum

I am trying new engines for the upcoming ChessWar XV Promo.

http://www.claudiocamacho.org/projects/chess0.php

This one does not seem to play as black under Winboard against another engine : it tries to switch sides.

You can play against it as white and black without problem. It will also play as black against another engine under Arena.

I don't see any email address where to complain at the website, so it would be nice if somebody finds a workaround or manages to hack the source

Olivier

I get immediately "White mates" when using chess0.

The relevant lines of xboard.debug seem to be

Code: Select all

790 <second: White (1):
GameEnds(37, White mates, 6)

The string "White (1)" seems to be interpreted by xboard as a mate claim.

This is on xboard 4.4.0 beta. Perhaps this issues has been fixed in more recent versions....

Ah I forgot : it needs xboard as a command line option.

Olivier

Ah I forgot : it needs xboard as a command line option.

Seems you are right!

Olivier Deville wrote:Ah I forgot : it needs xboard as a command line option.

Olivier

This won't help unless you have a newer version than I had in my report below.
http://www.open-aurec.com/wbforum/viewtopic.php?f=2&t=50361&p=190992&hilit=Chess0#p190992

Guenther

Michel wrote:I get immediately "White mates" when using chess0.

The relevant lines of xboard.debug seem to be

Code: Select all

790 <second: White (1):
GameEnds(37, White mates, 6)

The string "White (1)" seems to be interpreted by xboard as a mate claim.

This is on xboard 4.4.0 beta. Perhaps this issues has been fixed in more recent versions....

Indeed, WinBoard recognizes many such non-compliant commands, as a legacy of a policy to cater to non-compliant engines. Personally I think this is bad policy, but I guess in the early days engines were few and far in between. Nevertheless, kicking all that cr*p out is an almost certain invitation for a lynching mob that is angry their favorite old-timer engine is now broken with WinBoard...

I guess here the incompliancy was by mistake (forgetting an engine command-line argument), but in general it is pointless to try to fix things like this: it would just be favoring one form of non-compliancy over another. Officially the result of sending anything that is not in the specs is _undefined_. So if some GUI implementation takes it as a resign message, it is no fault of the GUI, and the solution is not to send it!