Re: [Maxima-discuss] subscripted functions

[Leo B. <l_b...@us...>]

Robert Dodier <rob...@gm...> writes:

> Leo Butler <l_butler <at> users.sourceforge.net> writes:
>
>> To be less cheeky, consider
>> 
>> f[n](e) := coeff(e,x,n)$
>> 
>> If f[n] := ... is a memoizing definition, I expect that
>> 
>> f[n](e) := ... is also a memoizing definition, which memoizes a function
>> of 1 variable (e) parameterized by n. That is, I expect it to be a
>> syntaxy equivalent to
>> 
>> f[n] := buildq([n:n],lambda([e],coeff(e,x,n)))$
>
> I've thought about this some more and at this point I believe
> the current behavior is better than the alternative. 
> When the function is f[n] := ... the body gets evaluated when
> the function is called. So when the function is f[n](x) := ...
> the body should likewise be evaluated (and then used to construct
> a lambda expression).
>
> As it happens, there is no documentation at all about functions
> like f[n](x) (and very little about f[n]). So there isn't any
> question of changing the documentation, or even of referring
> users to the documentation, because there isn't any.
> I am inclined to document the current behavior and let it stand.
>
> Thanks to all who have weighed in on this topic. It is a good one.

Robert, I think that it has been a mistake to say that f[n](x) :=
... *should* do X or Y based on the syntax. The point is that it does
*do* Y at the moment, and it has done this for a long time. Let's leave
it as is.

What I would like to do, then, is add a package in share/contrib that
defines a new infix operator, :≡, so that f[n](x) :≡ ... defines the
alternative behaviour.

I believe that this is consistent with current practice. For example,
contrast f(x) := ... and f(x) ::= ... .

Thoughts?


-- 
Leo Butler                <l_b...@us...>
SDF Public Access UNIX System -   http://sdf.lonestar.org