Author: Peter Skinner
Date: 15:49:56 11/22/00
Go up one level in this thread
>I have beta tested lots of hardware and software (no, not chess programs). In >every case, I was given a specific set of things to try, to test, to play with, >etc. And then a set of specific questions I had to answer, plus room for >any other comments I had. > >The alternative is just to give somebody a copy of something, say "use this for >a while and let me know what you think." You certainly get less info back that >way. But it takes less effort on both sides as well. > >Formulating a beta test plan, distributing the program, getting the results, >analyzing them, turns into a 'project'... I don't think chess engine authors >do this sort of detailed plan at all... I agree. In our case we were given the programs to evaluate, test in games, and report any bugs that we noticed. I personally played close to 1100 games via ICS during the beta period, and _never_ came up to an endgame of k vs kbn. So I could not see that it couldn't mate in that situation. As I personally liked testing the Rebel Tiger 13.0 and Gambit Tiger 1.0 better than the Rebel 3.0 package, I mainly focused on the engine within the interface, setting up Winboard engines in the Chess Partner interface, and tried as best I could to post games to the beta forum in which I thought one of the Tigers make a mistake in an endgame, or in an opening. With Rebel 3.0 I played quite a few auto232 games with various programs, to see book errors, and problems with the auto232 adapter. We had a great beta team this year, and I believe each had his or her own way of testing the product, and I truely believe that all the data collected from _all_ of the testers has truely made the Rebel 11.0 package as strong as it is. When we found errors, they were promptly fixed, and we were given the fixes to see if the error was still performed by the program. An author can only do so much to test his/her program, and a beta team like the Rebel group had more than served it's purpose, and I think the one or two possible bugs that are found were simply just not found. I think it is unreasonable for one to have the mind set that a beta tester should catch _everything_ wrong with a program. That would be nearly impossible.
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.