https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82215
Thomas Koenig <tkoenig at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2017-09-15
CC| |tkoenig at gcc dot gnu.org
Ever confirmed|0 |1
Severity|normal |enhancement
--- Comment #3 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
I think this is the PR that comes as a PDF report :-)
I think it would actually be enough to have a flag
to suppress the output of the *.mod file.
You could then do
foo.mod: foo.f90
gfortran -fsyntax-only foo.f90
foo.o: foo.mod foo.f90
gfortran -fmodule-files=no foo.f90
This would first generate the module file, and then compile it.
The problem with this (which is like your suggestion in the footnote,
but I don't think it is necessary to store the source path in the module)
is the potential to generate inconsistent *.o and *.mod files if
the user is not careful.
We could just say "Use at your own risk" in the docs and give
an exaple of correct use. I am not sure this is a good
solution.
Or we could decide to read in the module files, check for
consistency, and abort if something inconsistent is found.
This would be a major rewrite.
And all of this to address a problem that, as far as I can
tell, is already addressed using F2008's submodules.
A better solution would be a utility (or compiler flag)
that parses a *.mod file and outputs source code suitable
for calling the original module as a submodule. This could ease
the transition a lot. (A bit like the automatic output
of C interop prototypes, which I wrote recently). It would
just be syntactic sugar around the Fortran tree dumper.
So, undecided on the best way to proceed, but confirming.
