TOP 10 REASONS WHY YOU SHOULD STOP PROGRAMMIN

[Thomas Kerrigan]

Heh heh, yes, it really did play this move. It was a hashed move, and
evidently the promotion piece isn't included in the hash move record so
it defaulted to a N. Or at least this is what Stefan told me.

Hey! Don't be dissin Levy, even if he didn't let me restart my program on
round 1 despite precident. (Grrr)

Cheers,
Tom

------------------------------------------------------------------------------
Yea, though I walk through the valley of the shadow of APL, I shall
fear no evil, for I can string six primitive monadic and dyadic
operators together.
-- Steve Higgins

[chris whittington]

kerr...@alpha.pr1.k12.co.us (Thomas Kerrigan) wrote:
>
> Heh heh, yes, it really did play this move. It was a hashed move, and
> evidently the promotion piece isn't included in the hash move record so
> it defaulted to a N. Or at least this is what Stefan told me.
>
> Hey! Don't be dissin Levy, even if he didn't let me restart my program on
> round 1 despite precident. (Grrr)
>
> Cheers,
> Tom
>
>

Sorry to disagree, but on getting back from Paderborn, we obviously took
a good look at this bug.

It was caused as follows:

On return from normal search all promos from the engine get
handled just fine by the GUI.

Sometimes though, the search engine finds a mate (+99 score), if the
program is also in permanent brain (thinking on player time) it stops
searching and sits in a special case loop waiting for user input. If
the user input matches with the predicted permanent brian move, then
the program plays the move returned from the search engine. Except
that this special case loop manages to lose the promo information.

Result: If we find a mate line, where the first move is a promo,
in permanent brain, then the promo will be to a knight. In the game
in question Chess System Tal v. XXXX, CST score went from +99 pawns
to -8 pawns, can we claim the record for a score swing ?

AARRGGHH.

Needless to say this is now fixed !

On ICCA rules and precedents. Wouldn't it be nice if the rules were
explicit and the precedents all matched ?
Sometimes it seems they have a precedent to match whatever decision
they want. Without a written rule set the programmers and operators
are just helpless.

Or maybe just a simple rule: if your program screws up with a bug, or
hardware failure or whatever, you should get he chance to restart
and reset the position. The idea being to keep games moving forward
sensibly.

chris whittington