Subject: "fastfib" in the gf package faster than "fib"
From: Stavros Macrakis
Date: Sat, 19 Jul 2008 16:18:51 -0400
On Fri, Jul 18, 2008 at 10:59 PM, Richard Hennessy <rvh2007 at comcast.net>
wrote:
...This idea works for the assume database. You don't have to maintain more
> that one copy, just use an assume data structure like a tree instead of an
> assume "database" structure.
Richard,
Though I appreciate your enthusiasm, frankly I am getting tired of your
making naive suggestions for how *other people* could improve the Maxima
system. We certainly can use user feedback about areas to improve (and
there are lots of such areas), but telling us how easy it would be to do
this or that -- in complete ignorance of the existing code base -- is not
only unhelpful, but annoying.
This is a volunteer effort, and people contribute in areas they feel are
useful or interesting to them. The people here are pretty experienced, and
no doubt any of a number of us could spend a few weeks on the assume system
or a number of other subsystems and improve them in one way or another. The
assume system is weak in many areas, and my priority if I ever did decide to
spend some time on it would certainly not be making it thread-safe, but
making it mathematically more sophisticated. And by the way, the assume
database *is* tree-structured but not thread-safe.
Your hobbyhorse, for whatever reason, is multithreading. Fine. If you want
to improve Maxima in that area, then learn enough to contribute instead of
lecturing us with trivial observations like "most divide and conquer
algorithms would be good to run in separate threads" as though we were slow
students in a Computer Science for Poets class.
-s