Re: [RFC] Fortran search path for intrinsic includes and modules

[Richard E Maine <Richard dot Maine at nasa dot gov>]

On Nov 24, 2005, at 10:19 AM, Jakub Jelinek wrote:

> On Thu, Nov 24, 2005 at 06:46:33PM +0100, Steven Bosscher wrote:
> > Right. Other compilers give their modules the .mod extension too, and
> > those .mods are incompatble with gfortran's. So we should put gfortran
> > specific things in a gfortran specific place.
>
> Here is another version of the patch, that appends unless -nostdinc
> (like cpplib) -I /usr/lib/gcc/<target>/<version>/<multilibdir>/finclude
> (if it exists), or some other path based on the default search 
> algorithm
> in the driver....

It is no big deal, but if you are defining a place for .mod files and 
it is gfortran-specific anyway, can I suggest a more Fortran-friendly 
naming convention? In my personal coding, I use a directory name of 
"mod" instead of "include" for module files. One could spell out 
"module" if that seemed preferable. To me as a Fortran user, mod files 
don't have much to do with "include".

As I said, no big deal. If the "include" were an existing convention 
for .mod files, I'd see just going along. But if you are defining a new 
convention from scratch anyway, why not make the name more Fortran 
oriented? If I had code with both modules and include files (either C 
or Fortran ones), I typically wouldn't want them in the same directory 
anyway. Having separate paths for module and include would still allow 
someone who did want to use the same directory to do so.

--
Richard Maine                |  Good judgment comes from experience;
Richard.Maine@nasa.gov       |  experience comes from bad judgment.
                            |        -- Mark Twain