|
|
Post by alandau on May 28, 2014 21:32:30 GMT -8
|
|
levb
New Member
Posts: 18 
|
Post by levb on Jun 2, 2014 8:27:57 GMT -8
Yay! I'd noticed there was a bug long ago, e.g., in free-for-all games, some base propositions were missing. (But I haven't found it.)
The factory is non-deterministic simply because it uses a non-deterministic hash.
|
|
qfwfq
New Member
Posts: 29
|
Post by qfwfq on Jun 2, 2014 20:23:11 GMT -8
Pretty cool, thank you for working on Chess!
As there is no visualization, yet, I've tried the moves of a match on a board and noticed something weird: except for knights, all pieces (including rooks and queens) moved by at most one square per turn. I think I had already applied your fix to my player's code. Maybe a propnet generation or rule issue?
Update: the propnet passes verification. So far looks like a possible rule issue.
|
|
|
|
Post by Steve Draper on Jun 3, 2014 6:45:46 GMT -8
The visualization works now. Boy are the players bad at this! A new bar has been set in games-that-appear-to-be-played-terribly! This is a good thing.
|
|
|
|
Post by alandau on Jun 3, 2014 10:59:20 GMT -8
qfwfq: Interesting; you're right, this is probably a bug. I'll look over the game rules and do some tests later today or tomorrow.
|
|
|
|
Post by alandau on Jun 4, 2014 15:08:40 GMT -8
Looks like I omitted a definition for "coords", which I was relying on in the openStretch definition. I'll create a fix for that soon.
I'll have to rerun the performance tests after this.
|
|
|
|
Post by Steve Draper on Jun 5, 2014 13:18:31 GMT -8
Looks like I omitted a definition for "coords", which I was relying on in the openStretch definition. I'll create a fix for that soon. I'll have to rerun the performance tests after this. I assume it was the fixed version I played (against Galvanise) on Tiltyard last night (all pieces seemed to have full range of movement anyway). Sancho got around 5K playouts/sec near the start, trending upward gently as the game progressed, so performance is still decent. |
|
|
|
Post by alandau on Jun 5, 2014 20:30:04 GMT -8
Yep, the fixed version is up now.
Also, I've added warnings to the StaticValidator when a sentence name is referenced in the body of a rule but not defined anywhere. That should help prevent this kind of bug from happening again.
|
|
jackcs
New Member
never mind
Posts: 12
|
Post by jackcs on Jun 11, 2014 17:33:54 GMT -8
Probably way off topic here but speed chess allows illegal chess moves.
In particular, in chess the only legal moves when your king is in check are to move out of check, capture the checking piece, or block the check. In my speed chess loss, those moves, that were available, were not the only legal moves presented by the generate legal moves call.
|
|
|
|
Post by Steve Draper on Jun 11, 2014 17:56:18 GMT -8
Probably way off topic here but speed chess allows illegal chess moves. In particular, in chess the only legal moves when your king is in check are to move out of check, capture the checking piece, or block the check. In my speed chess loss, those moves, that were available, were not the only legal moves presented by the generate legal moves call. Yup, if you check out Alex's blog ( alloyggp.blogspot.com/2014/05/rewriting-chess-again.html) he notes that it's an approximation only in this version: I figured I would write a replacement for chess in two steps: first, write a "speed chess" version that ignores check and allows direct capture of the king; then, write a full version of chess. (Players' strategies should in theory be roughly the same as in normal chess, though this does depend on players reaching a certain level of play.) |
|
