Subject: Re: sum evals first argument inconsistently
From: Martin RUBEY
Date: Tue, 20 May 2003 18:05:46 +0200 (CEST)
On Tue, 20 May 2003, Richard Fateman wrote:
> The method of evaluation for the arguments to SUM have been hashed over
> repeatedly both in Macsyma (circa 1969?) and its competitors. It does
> not help to have the "right" answer if the users expect you to have a
> different answer.
Even if the answers are inconsistent? I think before 6.0 is the final
chance to change behaviour. I want to have lexical scoping and well
defined evaluation rules as birthday present :-)
Well, Christmas will be more realistic...
> I havent read / re-read the arguments recently, and I can't follow the
> link from Stavros' mail (it says ERROR - No group_id was chosen. )
Try to go the Bug report page and look out for bug number 740134.
> But an argument can be made to require that the first argument must be a
> function of one argument,
> lambda([index], expression).
Yes, I suggested some time ago to have a second set of functions that do
this, like fsum, fintegrate, ... whatever "should" take a function as an
argument...
> For what it is worth, I have been experimenting with Matlab (with
> symbolic toolbox), and it is really weird, with TWO value spaces and ONE
> namespace.
> e.g. command result
>
> pi 3.14159..
> x=3 sets x to 3
> simplify(pi+sym(0)) pi
> simplify(x+sym(0)) 3
>
> simplify(x) error
>
Sorry, I can't follow.
Martin