Re: [Axiom-developer] AldorForAxiom

[Ralf Hemmecke]

On 03/16/2006 04:38 PM, Bill Page wrote:
> Yes, that is the main step. There is a nice little topological
> sort of the dependencies done in Java that I think should really
> be done in Aldor. In fact, Tim Daly **insists** that we should
> not add Java to the list of prerequisits for building Axiom, so
> when we begin distributing Aldor with the Axiom distribution we
> will have some work to do. :)

I thought that it is quite a strange to install Java just for the sake 
of sorting, but now I looked to the java programm, well, it's a bit long 
and not too well documented. But perhaps I could try to port it to Aldor 
on an un-busy weekend. ;-)

> > But I still wonder why all these .o files are needed for the 
> > execution inside Axiom. I could understand .ao files, but .o
> > files?
>
> These are the compiled algebra files just like the ones
> generated from SPAD.

Yes, clear, but if I just stay in Axiom, I don't need .o files. Am I wrong?

> > Interestingly, when I say ")co fact.as" Axiom prints:
> >
> >     Compiling AXIOM source code from file /home/hemmecke/fact.as
> >     using AXIOM-XL compiler and options
> > -O -Fasy -Fao -Flsp -laxiom -Mno-AXL_W_WillObsolete -DAxiom -Y 
> > $AXIOM/algebra
> >        Use the system command )set compiler args to change these
> >        options.
> >
> > And if I then look into my directory I see.
> >
> >  >ls fact*
> > fact.ao  fact.as  fact.asy  fact.fn  fact.lsp  fact.o
> >
> > Where did the .o file come from?
>
> It was compiled by GCL from the .lsp source which was generated
> by Aldor. (Notice option '-Flsp' above.

Notice that there is no "-fo" on the command line. And if there were, I 
guess the aldor compiler would call a sequence .as -> .ao -> (hidden .c) 
-> .o.

> > So ")co fact.as" is not the same as calling
> > "aldor ABOVE_COMPILER_OPTIONS fact.as".

> Almost, except you need to compile the generated lisp.

Aha. And that is also .o? Is this the same object format that comes out 
if I compile a .c file?

> > Where should I look to make ")compile" more transparent to me.

> Unfortunately I can only suggests the Axiom source code.

I would not have expected anything else, but if someone could make my 
search space a bit smaller that would help.

> > And someone tell what happens when I execute "fact 4" in
> > Axiom after compiling ")co fact.as"? I always believed that
> > Axiom executes LISP code. So why is there any need for a .o
> > file?

> Axiom executes *compiled* lisp code. The .o files are machine
> code object files that are loaded dynamically by GCL.

Well, I simply removed fact.o and started a new Axiom. I said:

)lib fact
fact 3

and got 6 as result. Which code did I execute?

> > That should in priciple be possible once I have libaxiom.al,
> > I just haven't tried it yet.

> Let me know if you succeed.

I'll post my experience with that.

Ralf