Author: Uri Blass
Date: 11:11:36 08/14/03
Go up one level in this thread
On August 14, 2003 at 12:27:02, Robert Hyatt wrote: >On August 14, 2003 at 05:29:35, Uri Blass wrote: > >>There is already a free program that does it? >> >>The debugging function that I think about is writing the content of varaibles to >>files >> >>I should have 2 functions: > > >One thing you should do is write a "Validate()" procedure as I did for >Crafty. This procedure tests _all_ data structures. IE it makes sure >that the hash signature is correct for the current board position, that >the pieces are on the board correctly, that no two pieces occupy the same >square (possible with bitboards if you have a bug) and so forth. The last thing is theoretically possible also without bitboard. If bishops[i0][0]=rooks[i1][1] then it means that a white bishop and a black rook occupy the same square. I have an additional array that tells me for every square in the board what piece it has so if bishops[i0][1]=32 and info[32]!=blackbishop it means that I have a bug. I may add a function to check things but the problem is also to decide what things I need to check. At this point I do not know about specific bug based on test suites but I need to find the reason for the fact that the nodes are not the same before I continue. I already found a previous node that has a problem (node 213298536) when the arrays are not the same and probably the moves are not the same. I think at this point that the problem is something that should be expected and not a bug. The point is that if in case 1 king capture is in place 12 in the list of legal moves and in case 2 it is in place 11 then after I replace it with move number 1 in the list the place of move number 1 is different and it will continue to be different in case that the move is not a killer move and not one of the best moves based on history tables. different place for move number 1 means that the order of moves may be different and if move number 1 is considered(in most cases it is not the case because the good capture of the king generated a cutoff) then the number of nodes may be different. Uri
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.