Compile MF-NWT on debian jessie about pymake

[langevin-usgs]

@rabbl, there is some non-standard Fortran in MODFLOW-NWT at the moment. You should be able to just fix those issues in the source code, which is what I had to do. I don't remember the details, but I don't think they were difficult to fix.

[rabbl]

@langevin-usgs, sorry, but I'm not very familiar with fortran.
Compiling on debian with pymake comes the following error-message:

gfortran -O2 -fbacktrace -ffpe-summary=overflow -c ./src_temp/gsol7.f -o ./obj_temp/gsol7.o -I./obj_temp -J./mod_temp
./src_temp/gsol7.f:326.25:

        USE, INTRINSIC :: IEEE_ARITHMETIC
                         1
Fatal Error: Can't find an intrinsic module named 'ieee_arithmetic' at (1)
Traceback (most recent call last):
  File "make_mfnwt.py", line 42, in <module>
    make_mfnwt()
  File "make_mfnwt.py", line 32, in make_mfnwt
    expedite=False, dryrun=False, double=False, debug=False)
  File "/usr/local/lib/python2.7/dist-packages/pymake/pymake.py", line 633, in main
    expedite, dryrun, double, debug, fflags)
  File "/usr/local/lib/python2.7/dist-packages/pymake/pymake.py", line 357, in compile_with_gnu
    subprocess.check_call(cmdlist, shell=shellflg)
  File "/usr/lib/python2.7/subprocess.py", line 540, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['gfortran', '-O2', '-fbacktrace', '-ffpe-summary=overflow', '-c', './src_temp/gsol7.f', '-o', './obj_temp/gsol7.o', '-I./obj_temp', '-J./mod_temp']' returned non-zero exit status 1

In the sourcecode on ./src_temp/gsol7.f:326.25 there is the following:

C-------CONJUGATE GRADIENT ITERATIVE SOLVER
      SUBROUTINE GSOL_CGAP(NCORESM,NCORESV,
     2                     IPC,ICNVG,ISOLN,NOUTER,NINNER,
     3                     HCLOSE,RCLOSE,ETA,NR,NNZ,DAMP,IA,JA,A,
     4                     NRLU,NNZLU,NJLU,NJW,NWLU,JLU,IU,JW,WLU,Mi,
     5                     X,B,X0,DSCALE1,DSCALE2,
     6                     D,P,Q,Z,T)
        USE, INTRINSIC :: IEEE_ARITHMETIC
        IMPLICIT NONE

What could I change to fix this?
Thanks for helping

[mjigmond]

@rabbl I'm not sure if it's an option for you, but IEEE_ARITHMETIC was added at GCC version 5 (https://gcc.gnu.org/gcc-5/changes.html). Even so, you'll run into a crash that I fixed by changing a data type in gwf2swr7_NWT.f (that itself will generate some warnings but make will succeed).

14351: DOUBLEPRECISION, INTENT(INOUT) :: T1
change to
14351: REAL, INTENT(INOUT) :: T1

because https://gcc.gnu.org/onlinedocs/gfortran/SECNDS.html.

[rabbl]

Thanks @mjigmond for your suggestions. Debian brings gcc 4.9 with it and I had some problems installing the 5.0 version.
I have solved the issue with compiling and using modflow-owhm for my use case and its compiling and running as expected.

[jdhughes-usgs]

The latest release of MODFLOW-NWT (v 1.1.1) available on the web includes changes so that it will compile with gfortran. I have updated the make_mfnwt.py script in the examples directory. I have also added functionality to create a makefile (-mf or --makefile option) with pymake. It works for gfortran on windows or mac (and presumably linux) and for ifort on mac (and presumably linux). It may be a little fragile but I will be doing more testing to clean up some deficiencies.

[langevin-usgs]

The make_mfnwt.py script seems to be working now (at least on my mac), due to the use of all standard Fortran in the latest NWT release. Readme.md has been updated to reflect this.

Still need to add a NWT build to the autotesting, but that is a separate issue.