making for-loops more general. Was: Re: New version of makelist

Richard Fateman wrote:
> I added to the file mdo.lisp, referred to earlier the treatment of 
> sets. Now
>
> for i in [a,b] collect f(i)
>
> returns
> [f[a],f[b]]
>
> The display program in the command-line version of maxima displays 
> this new form of "do" properly.
> wxmaxima displays h(x):=for i in x collect f(i)   using 
> "mdoin-collect"  instead of  for ... collect.
>
> RJF.

As long as I was at it, I added to the file mdo.lisp  the additional 
grammatical and semantic routines to allow

for i:1 thru 10 summing f(i);

and  also

for i in L summing f(i);

This should make it perfectly clear what is being added, and that it is 
in fact being added. This
is not entirely clear if one writes sum(i , i,1,10). Is it being 
simplified first? What about
h(k):=i^2,
sum(h(k),i,1,10).  Is  h(k) evaluated to i^2 or not...

It is also perfectly clear that  for i:1 thru n summing f(i)  is NOT 
legitimate if n is just a "symbol".

I notice that the for-loop stuff does NOT permit the index to be (say)   
x[k].  Is this really
useful?  Indeed,  the for-loop index is locally bound, and

block([i:aha], for i:1 thru 10 do i,  print(i))   

prints aha,  not 10 or 11.  ( I think FORTRAN would return 11 in similar 
circumstances).

As for why I used "summing"   rather than just "sum" ...  it turns out 
that making "sum" a keyword in
the context of    "for ...."  makes
sum(i,i,1,10)  syntactically illegal.

Are there any problems with this?

Is anyone (else)  sufficiently interested and capable of put the pieces in
this file into the source tree?

RJF