Author: martin fierz
Date: 14:03:26 09/10/02
Go up one level in this thread
On September 10, 2002 at 16:36:08, Dave Gomboc wrote: >On September 10, 2002 at 13:04:25, martin fierz wrote: > >>i assume you are an intelligent young man. I also realized from your posts that >>you know nothing about checkers and were going to do just what you did - look it >>up on the nemesis website. and give the wrong position to schaeffer as a result >>:-) > >Well, you've got me there, then. :-) you had no chance on this... - it was perfectly normal that you would assume the info on the nemesis website to be correct :-) >So what is the right position? if you want to compare, only if you do it properly - so only if i get the logfile afterwards :-) and i hope you realize that even with the logfile, comparing one position is rather meaningless. >>i can tell you why i ask. you look at the code, and you see that there are some >>comments. but have you tried to figure out what they mean? have you tried to >>write access code of your own? have you tried to adapt that code to work with >>the 8-piece database? let me guess: no, no, and no? ed gilbert rewrote that code >>*for days* and got it wrong in the end. i've never even attempted to clean up >>that code because it i don't understand it. >>if you have ever tried to write efficient database access code for GB-sized >>databases, you know that this is not something you do in one afternoon. *that* >>is why i ask. why does schaeffer make everybody who wants to use this database >>in his program work so hard? he could just publish the "real" access code... it >>would cost him *5 minutes* and save everybody who wants to use it *5 days*. now >>that would be goodness of heart... > >The code at http://www.cs.ualberta.ca/~chinook/databases/code.c doesn't look too >tough to figure out to me. it's not hard to incorporate that code into your program as long as you don't have to change it... >Sure, looks can be deceiving, but I'm hacking crafty >at the moment to run an experiment, and I'd say that it's significantly harder >to understand than the database access code. ...and i find crafty much easier to understand :-) it probably all depends on your experience. maybe you have already done some endgame database stuff. then you have a much clearer picture of what is going on. a friend of mine explained to me how egdb indexing works in general, with the binomial coefficients etc. the guy was in nievergelt's games research group in zurich - for him that was easy, common knowledge. for me, it was completely unclear from reading the comments in the egtb access code, until he explained it to me. >Any newer code Jonathan has written probably looks similar to his old code! >He's a C hacker. i don't know if he wrote that code - the whole db thing was delegated to rob lake. i'd also guess that more than one person wrote that code. it's got good parts and awful parts. >You've already castigated the old code as basically unfit for >distribution (even for use with 6-piece tables) -- let's just say that I doubt >that any replacement code he'd write would give you a warm, fuzzy feeling in >your stomach. no, it wouldn't, but i could (and did) drop in the 6-piece code in my program easily with the 6-piece db. i can't do that with the 8-piece db. you have to modify all kinds of stuff just to make it work at all with the 8-piece db. and once you do that, caching as in the 6-piece case just doesnt work. so you really have to modify the whole caching thing too. and if you have to modify all these things, you really have to understand what's going on - else you break it like ed did. i don't care if it's unreadable, uncommented, and generally incomprehensible as long as i don't have to modify it & it works - which was the case with the 6-piece code. it was just ugly :-) BTW, according to the chinook webpage, the 6-piece db has been up there for a bit more than 5 years, not 10 as you said. they published it after chinook played it's final tournament. >But if the author of KingsRow has found the bug in the modified database access >code and fixed it, he could send it to Jonathan to put up on the website. i think he at least told schaeffer about the bug. i don't know if he sent him the code. aloha martin
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.