[Patch, fortran] PR24518 and PR24520 - Improvements to MOD an d DOT_PRODUCT

[Tobias.Schlueter@Physik.Uni-Muenchen.DE]

Quoting THOMAS Paul Richard 169137 <prthomas@drfccad.cea.fr>:
> As far as we (or, sorry Uros, I should speak for myself; I) understand it,
> the above should work.  However, the call to mathfn_built_in returns
> NULL_TREE, although it does find other functions such as BUILT_IN_TRUNC.
>
> You will find below a list of pointers and names, obtained by inserting in
> the above:
>
> for (i = 0; i < END_BUILTINS; i++)
>   gfc_warning_now (" %d  %d  %s", i, (int)built_in_decls[i],
> built_in_names[i]);
>
> Ignoring the segfault at the end, it is evident that a good proportion of
> the builtins are missing.  It was my understanding that library functions
> would be used if an honest-to-goodness builtin is not available.  Can this
> be done using tree expressions, ie. have I failed to initialize
> built_in_decls sufficiently, or do I have to call the library explicitly?

I think you have failed to init them sufficiently.  Have a look at
gfc_init_builtin_functions in f95-lang.c.  From a brief comparison it looks
like the functions that are non-NULL in your list are exactly the ones
initialized there.

- Tobi