how to pass the name of a variable as argument?

Ha, in my library I use quite a bit of dynamic scoping!
And since this is the only means in Maxima/Lisp, as
far as I see, to got some form of polymorphic behaviour,
it seems also necessary to me.

Only due to dynamic scoping you can make good use of "global"
variables, since when calling a function you can make
it act similar to some form of inheritance.

So dynamic scoping seems *essential* to me:
Back in the old days polymorphism wasn't much
of an issue, and structured programming was
the king, so I don't think those "good old
arguments" have such much relevance anymore.
Maxima without dynamic binding looks crippled to me.

Oliver


On Fri, May 02, 2008 at 01:44:23PM -0700, R Fateman wrote:
> Robert Dodier wrote:
> > On Thu, May 1, 2008 at 6:14 PM, Oliver Kullmann
> > <O.Kullmann at swansea.ac.uk> wrote:
> >
> >   
> >>  Since global variables should be avoided,
> >>     
> >
> > Bear in mind that, in Maxima, effectively ALL variables are global variables,
> > or, more precisely, global variables and local variables act the same.
> > All variables have dynamic scope, not lexical scope.
> >
> > As you doubtless know, dynamic scope can lead to subtle misbehavior.
> > A lexically-scoped user language could be very useful.
> > Instead of revising the existing Maxima language, at this point I think
> > the easiest way to get lexical scope is to run a different language
> > on top of Lisp. One possibility is Python; there is a CL implementation
> > of Python name CLPython. The developer is talking about porting it
> > to SBCL.
> >   
>  <snip>
> The Macsyma/ Maxima language is modelled on Algol 60, a language which 
> was sort of misunderstood when implemented at MIT.
> The mistake was to adopt the dynamic scope of Maclisp in the Macsyma 
> language.  The use of dynamic scope in lisp 1.5, according to John 
> McCarthy, was a mistake.  Subsequent experiments at MIT in lisp-like 
> languages got the Algol 60 lexical scope "right".  It resulted in  Scheme.
> 
> Now if you want to adopt lexical scope in Maxima,  I think it is not 
> necessary to change the syntax or language at all, just the implementation.
> (I am assuming that definition of procedures within procedures would be 
> handled properly...)
> That is, the surface syntax would look identical. There is no need to 
> adopt python, or java or C# or ....
> 
> Would this break existing programs?  Maybe some.  It is my understanding 
> that Maple, between versions 9 and 10 (I think) changed the language 
> implementation from dynamic to lexical scope.  And there is way way more 
> code in the Maple language than in Maxima top-level syntax.
> 
> Most programmers probably don't understand the scoping issues and use 
> only global variables or local variables, and do not nest procedures.  
> So it will not make any difference in their code.
> 
> There is another issue which has to do with 2-lisp vs 1-lisp semantics, 
> and whether you can have two bindings for the same name, a function and 
> a variable, or just one.
> That is, can you do f(x):=.....  
> and also f:3.
> or will the second assignment over-write the first.
> 
> Common Lisp,  and Maxima,  is a 2-language. Scheme is  a 1-lisp.
> 
> Now it may be appropriate to build another front end, interpreter, etc, 
> for Maxima, based on python.  Probably not too hard.  There is, for example,
> another front end for Macsyma that used Matlab syntax. It could entirely 
> co-exist with the existing one..
> 
> Or the alternative could be a front end for the lexically scoped 
> language that looks just the same,  but is subtlely different.
> 
> My 2 cents..
> 
> RJF
> 
> 
> _______________________________________________
> Maxima mailing list
> Maxima at math.utexas.edu
> http://www.math.utexas.edu/mailman/listinfo/maxima