primitive inclusion

primitive inclusion

[Peter S. Christopher]

Hi listers,

	I'm having a problem finding a primitive function in one of my
modules. Here is the setup. I have a C program which (after
booting guile) defines a routine a-la:

scm_c_define_gsubr("dvect?", 1, 0 ,0, dvect_p);

Then the program loads a file using

scm_c_primitive_load("root.scm");


At the top of root.scm I include a module via: (use-modules (slab)) Now
within the text of the slab.scm I expect to use dcect?. However, the guile
cannot reference dvect? -- it can't find it. The problem is that dvect? is
being put in the (guile-user) module. And the slab module can't see it.
I've tried to (use-modules (guile-user)) and this seems to have no effect.

Does anyone have any ideas how I can get it so that the slab module can
see dvect? ?  

Thanks for any help,

Pete



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: primitive inclusion

[Steve Tell]

On Fri, 25 Jul 2003, Peter S. Christopher wrote:

> Hi listers,
> 
> 	I'm having a problem finding a primitive function in one of my
> modules. Here is the setup. I have a C program which (after
> booting guile) defines a routine a-la:
> 
> scm_c_define_gsubr("dvect?", 1, 0 ,0, dvect_p);
> 
> Then the program loads a file using
> 
> scm_c_primitive_load("root.scm");
> 
> 
> At the top of root.scm I include a module via: (use-modules (slab)) Now
> within the text of the slab.scm I expect to use dcect?. However, the guile
> cannot reference dvect? -- it can't find it. The problem is that dvect? is
> being put in the (guile-user) module. And the slab module can't see it.
> I've tried to (use-modules (guile-user)) and this seems to have no effect.
> 
> Does anyone have any ideas how I can get it so that the slab module can
> see dvect? ?  

I had similar problems trying to make a program compatible with both
guile-1.5/1.6 and guile-1.4 - one of the differences between the two is
which module C primitives end up in vs. the default module that scheme
code runs in.

I came up with this hack, placed immediately after initializing guile and
before defining primitives to make guile-1.6 behave enough like 1.4:

#ifdef HAVE_SCM_C_READ_STRING
        {       SCM exp = scm_c_read_string("(define-module (guile))");
                scm_primitive_eval_x(exp);     }
#endif

While this works, making guile 1.6 behave like 1.4 isn't really the way to
go any longer.

I see that the guile-1.6.4 reference manual has a nice section "Accessing
Modules from C."  (thanks authors!)

When writing a program for guile >= 1.6 only, it appears that one should
use scm_c_define_module() to place their C primitives into the desired
application-specfic module.

Steve


> Thanks for any help,
> 
> Pete
> 
> _______________________________________________
> Guile-user mailing list
> Guile-user@gnu.org
> http://mail.gnu.org/mailman/listinfo/guile-user
> 

-- 
--
Steve Tell  tell@telltronics.org 




_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user

Re: primitive inclusion

[Peter S. Christopher]

Hi there,

	Thanks for the help Steve, worked like a charm. For the benefit
of anyone who is interested: here is how one defines primitives in the
(guile) module (rather than the (guile-user) module which seems to be
standard).

<CODE>

SCM slab_write_sloppy();
SCM slab_read_sloppy();
SCM map_globals();
SCM dvect_p();

void init_my_proc(void *data){
  scm_c_define_gsubr("slab-write-sloppy", 5, 0 ,0, slab_write_sloppy);
  scm_c_define_gsubr("slab-read-sloppy", 4, 0 ,0, slab_read_sloppy);
  scm_c_define_gsubr("map-globals", 1, 0 ,0, map_globals);
  scm_c_define_gsubr("dvect?", 1, 0 ,0, dvect_p);
  return;
}

static void
inner_main (void *closure, int argc, char **argv){
  /*VARIABLES AND WHAT NO*/

  // This call makes the "guile" module the current-module for the
  // duration of the call to init_my_proc
  scm_c_define_module("guile", init_my_proc, NULL);

  /*Etcetera*/

} 
</CODE>

Thanks again.
 
Pete

On Sat, 26 Jul 2003, Steve Tell wrote:

> On Fri, 25 Jul 2003, Peter S. Christopher wrote:
> 
> > Hi listers,
> > 
> > 	I'm having a problem finding a primitive function in one of my
> > modules. Here is the setup. I have a C program which (after
> > booting guile) defines a routine a-la:
> > 
> > scm_c_define_gsubr("dvect?", 1, 0 ,0, dvect_p);
> > 
> > Then the program loads a file using
> > 
> > scm_c_primitive_load("root.scm");
> > 
> > 
> > At the top of root.scm I include a module via: (use-modules (slab)) Now
> > within the text of the slab.scm I expect to use dcect?. However, the guile
> > cannot reference dvect? -- it can't find it. The problem is that dvect? is
> > being put in the (guile-user) module. And the slab module can't see it.
> > I've tried to (use-modules (guile-user)) and this seems to have no effect.
> > 
> > Does anyone have any ideas how I can get it so that the slab module can
> > see dvect? ?  
> 
> I had similar problems trying to make a program compatible with both
> guile-1.5/1.6 and guile-1.4 - one of the differences between the two is
> which module C primitives end up in vs. the default module that scheme
> code runs in.
> 
> I came up with this hack, placed immediately after initializing guile and
> before defining primitives to make guile-1.6 behave enough like 1.4:
> 
> #ifdef HAVE_SCM_C_READ_STRING
>         {       SCM exp = scm_c_read_string("(define-module (guile))");
>                 scm_primitive_eval_x(exp);     }
> #endif
> 
> While this works, making guile 1.6 behave like 1.4 isn't really the way to
> go any longer.
> 
> I see that the guile-1.6.4 reference manual has a nice section "Accessing
> Modules from C."  (thanks authors!)
> 
> When writing a program for guile >= 1.6 only, it appears that one should
> use scm_c_define_module() to place their C primitives into the desired
> application-specfic module.
> 
> Steve
> 
> 
> > Thanks for any help,
> > 
> > Pete
> > 
> > _______________________________________________
> > Guile-user mailing list
> > Guile-user@gnu.org
> > http://mail.gnu.org/mailman/listinfo/guile-user
> > 
> 
> -- 
> --
> Steve Tell  tell@telltronics.org 
> 
> 



_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://mail.gnu.org/mailman/listinfo/guile-user