[Gm2] Re: Modula-2 remarks

[Gaius Mulley]

Hi Max,

the equivalent of the inline statement in m2f is the asm
(same syntax as gcc, only the asm keyword is in capitals).
So we can write:

IMPLEMENTATION MODULE libtest ;

PROCEDURE sqrt (x: LONGREAL) : LONGREAL ;
BEGIN
   ASM("   fldl %1 ; fsqrt" : : "g" (x));
END sqrt ;

END libtest.

and

DEFINITION MODULE libtest ;

EXPORT QUALIFIED sqrt ;

   PROCEDURE sqrt (x:LONGREAL) : LONGREAL ;

END libtest.

The same is with gm2, for convenience I've copied a snippet of the gm2.texi
documentation at the bottom of the mail message.
(The documentation needs a lot of work).

I agree with your rationale of library modules. My vote would be for adding
new directories, pim-libs, iso-libs which contain the different dialect
libraries. Both pim-libs and iso-libs may need to import from FIO though
otherwise there would be alot of code duplication. We might also need
different architecture directories (for GNU Modula-2). Though we might
be able to tap into the builtins of the gcc backend for some of the functions.
Maybe we can introduce a Builtin.def to incorporate this cleanly.

> My opinion on GNU Modula-2 :
> The language should be the ISO defined language.

I'm not sure that this dialect is used much, (language not libraries).