problem with sbcl when it thinks it's not interactive

Hello,

----- Original Message ----- 
From: "Andrey G. Grozin" 
To: "Siver Andrey" 
Cc: 
Sent: Tuesday, November 22, 2005 2:13 PM
Subject: Re: [Maxima] problem with sbcl when it thinks it's not interactive


> On Tue, 22 Nov 2005, Siver Andrey wrote:
> > Hello, Maxima Community,
> > I would like to say an opinion  (may be off topic here) that it's not
good
> > thing from methodological point of view when MAXIMA tries to ask
something
> > interactively.
> But what can maxima do if the form of the integral depends on the sign of
> a, and you have not supplied this information via
>
> assume(a>0);
>
> or something? I see 3 possibilities:

It seems to me, the general solution is: if some object can not be evaluated
it should be hold unevaluated until next stage of calculations.
If this rule would be applied to full calculational chain then there were no
buggs caused by this rule, as I could understand (only may be caused by some
enviromental limitations).

>
> 1. Refuse to calculate this integral
>
> 2. Return an unevaluated 'if
> This opens a *huge* can of worms
>
> 3. Somebody on this list suggested an interesting solution. Let the user
> to specify "sample" numerical values of all parameters. They will be used
> *only* for evaluating booleans like is(a>b) etc. The result produced will
> be correct in some neighbourhood of the specified point (whose boundaries
> are unknown).
>
> I think it is a *very* strong point of maxima that it takes mathematical
> correctness seriously, and always (tries to) return the result which is
> correct in the specified region of parameters. Other systems happily
> return integrals which are wrong in the region I am interested in.

This point is rather about correct or incorrect result but not about
calculational strategy.
By the way, different systems use different numerical domains for default
defined variables. For example, all variables in Mathematica are of type
complex by default.

>So,
> questions are an importand (and good) part of maxima design, and should be
> retained. But it should be possible to suppress them, thus losing a part
> of functionality (I hope *not* correctness!!)
>
> Andrey
>
> _______________________________________________
> Maxima mailing list
> Maxima@math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima
>
>

Regards,

-Andrey