Re: [Axiom-developer] Re: this is open source

[Ralf Hemmecke]

> > > They can do that privately in their own source code environments
> > > or they can do it more publicly via the Axiom Wiki. All types
> > > of Axiom-related programming are now possible through the web
> > > including Lisp, Boot, Spad, Aldor and Axiom interpreter scripts.
> > For a beginner these are too many languages to choose from.
> > People need guidance when to use which, and which language will
> > be supported on the long run.  (Nobody wants to invest time in a
> > language that will soon be discontinued.)

> Isn't that what they said about Fortran? :)

> As far as I am concerned none of these languages should be
> discontinued. They are all essential to the way Axiom is
> built. (Tim Daly disagrees about Boot, but that is a different
> story.)

Well, I think I rather agree with Matthias and Tim. The fewer languages 
I have to learn in order to get some mathematical stuff running, the better.

Boot is something that is maybe a bit easier to learn than LISP, but
it is not essential for Axiom in the long run. But I would rather have 
replaced BOOT by Aldor instead of LISP if this were possible.

> Learning Lisp is one thing in itself. If you program at all, you
> should at least know something about Lisp.

Yes, I know some basics in LISP, but I would not say I can speak it 
fluently. And although LISP is at the moment very important for Axiom, 
wouldn't it make sense for the future to replace more and more LISP by 
some higher level language? I consider LISP a kind of assembler 
language. Sure one can do all kinds of stuff with it, but Axiom actually 
exists because it is not LISP. Axiom provides a very nice language 
SPAD/Aldor. Anything below that is just hairy details for 
axiom-developers/maintainers. And if these hairy details could be made 
less hairy then that is certainly better. It is also better in view of 
getting more people become axiom-developers. As Matthias said, a new 
person just doesn't easily know where to start and thus is lost as an 
axiom-developer.

So making the build process simpler is essential to accelerate the 
development.

> Learning Axiom interpreter scripts is essential for all but
> trivial use of Axiom.

I hope some day we will just have Aldor and Bnatural (which I think 
would be a library written in Aldor).

> Learning Spad is essential for Axiom library programming.

Right, but unfortunately SPAD is not Aldor. So which language to choose?
Right, spad and aldor are very much the same, but unfortunately there 
are some tiny differences. It would make me somehow happy if Axiom 
states somewhere that SPAD is deprecated and for new code Aldor should 
be preferred.

> Learning Boot is simple because it is essentially a simplified
> form of Spad - an intermediate language between Lisp and Spad.
> Boot is used only internally in the Axiom compiler and interpreter.

Yes, Boot is internal, and until I really want to learn how the compiler 
and interpreter work I don't learn it. Well, I think it would also lead 
to a paragraph in the developer's guide. What language is used for what 
internally. Tim, is there already such a paragraph somewhere?

> I think that it is the nature of Axiom that it's development
> environment can not be made much simpler than this. There is
> the old saying: Everything should be as simple as possible,
> but not simpler. :)

As simple as possible: Axiom has a kernel in written in C or LISP and 
the rest is Aldor. Bill, would you consider that too simple?

OK, I agree, Axiom has many more languages:
(La)TeX
awk
sh
Makefile
noweb
perl(?)
awk
C
python(?)
etc...

But these languages are existing languages in their own right. Axiom 
adds BOOT, SPAD, the Interpreter language, Aldor. Aldor could even be 
seen to belong to the list above.

So what Axiom should add is a nice mathematical interface. Type 
mathematics, get out mathematics. Bnatural would be a first approach to 
this. Hopefully there will be some timeslot in the near future where 
someone could implement this.

Best regards
Ralf