On Feb 12, 2008 11:07 AM, Barton Willis <willisb at unk.edu> wrote:
> I would like to call attention to the SF bugs
>
> [ 1891216 ] is(if a>0 and a>0 and a>0 ...) => internal error
> [ 1891911 ] is(f(a+b+c)) --> Lisp error
> [ 1726550 ] not bugs
>
> I think these bugs are related--fix one bug and you'll get two bonus
> points! And they make it impossible for me to test some interesting
> (well, at least I think) experimental linearalgebra code. As a
> mathematician who has been known to program, the bugs in db.lisp and
> friends are beyond me.
Barton, I looked at these bugs in src/db.lisp recently and I couldn't
make any progress. I;ll try to take another look.
At least part of the problem is the circular property list which appears
to store declared numerical values. The circular lists cause trouble
elsewhere in the code. But there could well be more to the "is" bugs
-- I would have to look.
db.lisp is an incomprehensible mess. I don't think it is doing
anything very complicated (to judge by the weakness of the
assume system). Let's start thinking about how to replace it.
best
Robert