Re: "load" on Windows

[Eli Zaretskii]

> From: "Gisle Vanem" <address@hidden>
> Date: Sat, 5 Oct 2013 14:33:26 +0200
> 
> "Eli Zaretskii" <address@hidden> wrote:
> 
> > Well, the tests in the test suite that test this feature did work for
> > me at some point, so you may wish first to verify they do for you, and
> > then compare your extension with the ones used by the test suite, to
> > see what's different.
> 
> Well this is just silly. I've added tracing code a lot of places (enabled
> by option --trace) . I do see my mk_test_gmk_setup() gets called, but not 
> the new function added by gmk_add_function(). 
> 
> gmk_add_function() calls define_new_function(), but therein I see a:
>   ent->fptr.alloc_func_ptr = func;
> 
> What about setting:
>   ent->fptr.func_ptr = func;
> 
> too?

They are one and the same, since they are members of a union:

  struct function_table_entry
    {
      union {
        char *(*func_ptr) (char *output, char **argv, const char *fname);
        char *(*alloc_func_ptr) (const char *fname, int argc, char **argv);
      } fptr;

> How else is the new function supposed to be called? I don't understand
> this contorted logic.

I don't understand the question, sorry.  The logic looks quite simple
to me: In expand_builtin_function we have:

  if (!entry_p->alloc_fn)
    return entry_p->fptr.func_ptr (o, argv, entry_p->name);

  /* This function allocates memory and returns it to us.
     Write it to the variable buffer, then free it.  */

  p = entry_p->fptr.alloc_func_ptr (entry_p->name, argc, argv);
  if (p)
    {
      o = variable_buffer_output (o, p, strlen (p));
      free (p);
    }

For loaded functions, define_new_function has this:

  ent->alloc_fn = 1;
  ent->fptr.alloc_func_ptr = func;

So expand_builtin_function should see that alloc_fn is non-zero, and
invoke your function via the alloc_func_ptr pointer.

> For reference, here is my complete mk_test.c:

Thanks.  Your problem is here:

>   EXPORT int mk_test_gmk_setup (const gmk_floc *flocp)
>   {
>     gmk_add_function ("hello_world", hello_world, 0, 255, 0);
                         ^^^^^^^^^^^

Make functions cannot have the '_' character in their names, so it
seems.  Here's why:

  /* Look up a function by name.  */

  static const struct function_table_entry *
  lookup_function (const char *s)
  {
    const char *e = s;

    while (*e && ( (*e >= 'a' && *e <= 'z') || *e == '-'))  <<<<<<<<<<<<
      e++;

So if you name your function hello-world instead, it will work.

Paul, if this limitation is deliberate, I suggest to document it where
we explain the arguments of gmk_add_function.

Btw, Gisle, I don't understand why you needed to use the EXPORT
thingy, the DLL is compiled just fine without it, and only exports
non-static variables and functions anyway.  Doing away with it makes
the code portable to platforms other than Windows.