On Wednesday, January 5, 2005, 8:43:13 AM, Heimir wrote: HE> Something is not right with my rulebase I think.
HE> I went from 11mb to 23mb then down to 5mb. HE> Is that right? Finally, yes, this is correct. I began to optimize the weak-rule-removal mechanisms over the last two days. When I implemented the final piece of critical code it uncovered a bug in the rulebase compilers. (I call that a lens effect - when fixing one part of a system uncovers problems in another part). Around 0100 today (deep last night) the problem was discovered. Somehow, by deprecating the weak rules in the system the rulebase compilers became convinced to include nearly every rule that was ever created. So, rulebases shot up to ~24 Mbytes with 213000+ rules! I rolled back the changes and began hunting for the cause. A couple of hours later I had found what I thought was the problem and a way to work around it. I made an experiment of my fix and it worked, so I was able to roll back in the weak-rule-removal optimizations. As a result, we are now able to have the reduced size rulebases that we originally hoped for with all of the weak rules properly (aggressively) removed. Thus, rulebase sizes have dropped - yours to 5Mb. It was quite an adventure, but it seems to be working fine so far. Hope this helps, _M This E-Mail came from the Message Sniffer mailing list. For information and (un)subscription instructions go to http://www.sortmonster.com/MessageSniffer/Help/Help.html
