Author: Robert Hyatt
Date: 06:11:39 03/19/98
Go up one level in this thread
On March 19, 1998 at 04:40:45, Ed Schröder wrote: >>Posted by Bruce Moreland on March 18, 1998 at 20:44:15: > >>In Reply to: Re: SSDF and the programmers............ posted by Ed Schröder >>on March 18, 1998 at 18:01:26: > >>On March 18, 1998 at 18:01:26, Ed Schröder wrote: > >>>Right. > >>>Bruce, if the SSDF people ask you to test Ferret for the list would you >>>do it? I mean in principal? > >>Maybe, I don't know. They've never asked me. > >>I don't have an autoplayer version, secret or public, other than the ICC >>version, which plays its games in public. > >Adding the AUTO232 code is a one day job. You only have to sent the >moves played by Ferret to the printer. The supplied NONAME driver >does the rest. > > First time I've disagreed violently with Ed in a while, but the above is *pure bullshit*!!! It takes a day to add the code. But auto232 is so damned timing sensitive it may take months to get it to work right. IE I finally produced a crafty that would play using auto232, then Tony reported that as soon as the game gets into the endgame databases, it will hang. It turned out to be because Crafty was moving *too quickly* violating some bizarre timing requirement in auto232. I then added a delay before sending each move, which worked until someone tried it on a P6/200, where it failed. I had to make the delay variable for different platforms. This software is *not* written the way things ought to be written. I can't begin to imagine how someone wrote this and left such timing dependencies buried in it. But they are definitely there... >>I would have a lot of worries about getting on the Swedish list. You've >>had some problems with testers not using configurations you've >>specified. I would be worried about this and about the use of >>sub-optimal hardware, and about getting a wide variety of games versus a >>wide variety of opponents. > >Sometimes these things do happen. Only on very rare occasions. There >are now some complaints about wrong game scores. If so they will be >corrected. In general I would give the Swedish a 9.8 on this part. > > >>I don't know much about how they do their testing, but allowing anyone >>else to have some say about my reputation would worry me, obviously. > >There is indeed some risk :)) > > >>There's so much incentive to be high on the list that I would worry >>about my competitors doing strange things to skew the results. I don't >>know whether Ossi has a real complaint or whether he's just being >>paranoid, but the fear that something weird will happen to you has got >>to be quite real. > >All clear, thanks. > > >- Ed - > > >>bruce
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.