RE: [Axiom-developer] Expression problems

[Bill Page]

Hans Peter,

On November 7, 2005 1:28 AM you wrote:
> 
> Thank you for the quick answer. I had hoped for a different 
> one. S-expressions are what I would call a parse tree, but
> I feel that I have to leave thinking in Axiom and need
> rethinking in Lisp. Anyway, I will try.

I know what you mean, but I don't think that it is necessary
to "rethink in Lisp" as such. Structures like InputForm are
certainly Axiom domains in their own right. For example you can
write:

  A:=sin(2*x)::InputForm
  A.2.1

to get the "*". The InputForm domain includes a large number
of operations:

  )show InputForm

including one called 'interpret' that evaluates the InputForm
as an actual input.

> 
> I understand if you write that inMult does exactly what it 
> should, but somehow I would like to program without always
> having to have the precise documentation at hand (which in
> this case would have to be the source code, as in the
> HyperText documentation no type requirements are given, i.e.
> that "n in R").

Of course HyperDoc and the Axiom book need to be updated to
include information about these functions.

> 
> Just one example to explain, why I felt there was some
> inconsistency:
> 
> if isPower(x**y) returns [x**y,1] then I would expect 
> isMult(x*y) to return [x*y,1] as the power expression
> seems to be viewed as (x**y)**1, so one would 
> expect the second expression to be viewed as (x*y)*1.
> 

I agree with you that it is a little inconsistent. What do
you think would be better: 1) for isPower(x**y) to return
"failed" or 2) for isMult(x*y) to return 1?

If we check carefully how these are used in the Axiom library
code and the Axiom test input files now, then I expect that
it would be quite easy to make these operations consistent.

Regards,
Bill Page.