[netCDF #KAS-529028]: configure of netcdf-fortran incorrectly detects capitalized module names

Hi Takeshi,

> I maintain the netcdf-fortran port in MacPorts.

Thank you for doing this!  I have mistakenly assumed the MacPorts
maintenance was done by Takeshi Horinouchi from the University of 
Kyoto, but I'm happy to give you credit from now on, and your name 
has been added to 
  http://www.unidata.ucar.edu/software/netcdf/credits.html

> A user reported that f90 modules with capitalized file names are
> installed, which do not work.
> 
> I found two problems in configure script of netcdf-fortran (version 4.2).
> 
> 1. conftest.f90 with module definition is compiled without -c to fail.
> 2. Mac usually uses a case-insensitive file system.
> test -f CONFTEST.mod is true for both conftest.mod and CONFTEST.mod.
> 
> Please find a patch for configure. It would be nice if these problems
> are fixed in the future versions.

Unfortunately, the "configure" file is generated automatically by running
"autoreconf -i -f" from the "configure.ac" source file in the distribution.
If I just patch the "configure" file, it will be overwritten by any later
change to the configure.ac source.

I can see the configure.ac source responsible for generating the erroneous
code is:

# Test whether F90 module names get capitalized.
if test "x$nc_build_f90" = xyes; then
   AC_PROG_FC_UPPERCASE_MOD
fi
if test "x$ac_cv_prog_f90_uppercase_mod" = xyes ; then
AC_MSG_FAILURE([Upper Case Modules names not supported])
fi
AM_CONDITIONAL(UPPER_CASE_MOD, [test "x$ac_cv_prog_f90_uppercase_mod" = xyes])

but I see nothing there that can be modified to fix the problem,
which makes me wonder if it's actually a bug in the autoconf m4
macros AC_PROG_FC_UPPERCASE_MOD where the problem should be fixed.

The original developer of this macro for autoconf was apparently here:
http://lists.gnu.org/archive/html/automake/2004-01/msg00245.html
but I can't find the source definition for AC_PROG_FC_UPPERCASE_MOD
anywhere in the installed autoconf or automake files, so I'm not sure
how to fix this in a way that's maintainable.

If you have any ideas about the best way to handle this for MacPorts,
I'd be interested.

--Russ

> Regards,
> 
> Takeshi
> 
> 
> 
Russ Rew                                         UCAR Unidata Program
address@hidden                      http://www.unidata.ucar.edu



Ticket Details
===================
Ticket ID: KAS-529028
Department: Support netCDF
Priority: Normal
Status: Closed