Hi Folks,
Dariush Navabi and Dan Larner have both recently reported bugs in Jess
5.0a2, the symptom of which is the always exciting "Corrupted negcnt"
message. Whenever you see this in a Jess exception, it always means
Jess's internals are messed up, due to a bug in Jess itself. Anyhow,
Dariush's nicely repeatable bug led me to the source of, I think, both
of these reports. It turns out to have been something I did in 5.0a2
that I quickly undid, so the bug was already fixed in my working
copy. The rub is that this bug was introduced by the fix to a
different bug... so there's no trivial patch to fix it without
breaking something else.
I think what this means is that I'd better do a 5.0a3 soon which
incorporates the fix for this as well as the other few bug fixes and
assorted goodies I've done since 5.0a2. I've not done as much as I
would have liked to, but Danielle is now 11 months old, and extremely
mobile: the days of the uninterrupted 11-hour hacking session are
unfortunately in my dimly-recalled past.
One thing I do have is an automated code generator that turns parts of
some Jess rules into compiled Java code. It works quite well on
things that do a lot of function calls and are mostly math; for
example the wordgame problem. Actually, I'm happy to say that you can
feed wordgame to Jess and (after the few seconds it takes for
compilation!) the runtime on my machine with JDK 1.2 is 4 seconds; the
runtime under CLIPS 6.04 is 5 seconds! Something like 'zebra' which
does -some- math, speeds up by about 30%. It's very rough, and doesn't
always gove perfect results, but it's very interesting to me, anyway.
One more tidbit: I've started using IBM's "jikes" Java compiler (see
www.alphaWorks.com.) It is -very- fast, and picks up on some subtle
style issues that javac misses. It seems very robust so far, and it
compiles Jess in less than 30 seconds on my machine (as opposed to
several minutes for javac.) Check it out, if you haven't already.
---------------------------------------------------------
Ernest Friedman-Hill
Distributed Systems Research Phone: (925) 294-2154
Sandia National Labs FAX: (925) 294-2234
Org. 8920, MS 9214 [EMAIL PROTECTED]
PO Box 969 http://herzberg.ca.sandia.gov
Livermore, CA 94550
---------------------------------------------------------------------
To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]'
in the BODY of a message to [EMAIL PROTECTED], NOT to the
list. List problems? Notify [EMAIL PROTECTED]
---------------------------------------------------------------------
