[MITgcm-devel] MNC genmake2 and Mac

[Baylor Fox-Kemper]

Hi David,
   I had a problem with this a while back, and Ed helped me fix it.   
Check out the following from my optfile:

Cpp='/Usr/Bin/Cpp -Traditional -P -Xassembler-With-Cpp'
Fc=/Opt/Ibmcmp/Xlf/8.1/Bin/F77
Link=/Opt/Ibmcmp/Xlf/8.1/Bin/F77
Fflags='-Qsuffix=F=For'
Foptim='-O3'
#Skip_Netcdf_Check=T
#Have_Netcdf=T
Libs='-L/Users/Baylor/Research/Netcdf-3.6.0-Xlf/Lib -Lnetcdf -L/Usr/ 
Lib -Lsystemstubs'
Includes='-I/Users/Baylor/Research/Netcdf-3.6.0-Xlf/Include'
Fc_Namemangle="#Define Fc_Namemangle(X) X"

You definitely the need the systemstubs linking, which is a new weird  
'feature' of tiger.  You may be able to compile with the skip_netcdf  
flag and run.  I remember that for a while I was not able to run the  
mnc tests, but the mnc from the model worked fine...
   Cheers,
     -Baylor



On Oct 4, 2006, at 3:14 PM, David Ferreira wrote:

> Hi,
>
> I'm trying to get the MNC package to work on my Mac, and obviously
> it doesn't go smoothly.
> Although I have netcdf installed, genmake2 runs some test that  
> fails and then
> disables MNC. The first complain from the genmake2 test is a  
> warning about
> failing to preprocess the test.
> Trying to solve it, I realized that this message is also issued on
> Aces and Columbia where MNC works fine. So it seems it has
> nothing to do with my Mac and/or netcdf library
>
> The problem may be in those lines :
>
>    COMM="cat genmake_tnc.F | $CPP $DEFINES $INCLUDES"
>    echo "$COMM" >> genmake_tnc.log
>    $COMM > genmake_tnc.$FS 2>/dev/null  &&  RET_CPP=t
>    if test "x$RET_CPP" = xf ; then
>       echo "  WARNING: CPP failed to pre-process the netcdf test." \
> 	    >> genmake_tnc.log
> 	echo "    Please check that \$INCLUDES is properly set." \
> 	    >> genmake_tnc.log         .....
>
> if "/dev/null" is replaced by "std_err_tnc", I get that:
>
> sdt_err:tnc:
> cat: |: No such file or directory
> cat: -traditional: No such file or directory
> cat: -P: No such file or directory
> cat: -xassembler-with-cpp: No such file or directory
> cat: -D_BYTESWAPIO: No such file or directory
> cat: -DWORDLENGTH=4: No such file or directory
> cat: -DNML_TERMINATOR: No such file or directory
> cat: -DHAVE_SYSTEM: No such file or directory
> cat: -DHAVE_FDATE: No such file or directory
> cat: -DHAVE_ETIME: No such file or directory
> cat: -DHAVE_CLOC: No such file or directory
> cat: -DHAVE_SETRLSTK: No such file or directory
> cat: -DHAVE_SIGREG: No such file or directory
> cat: -DHAVE_STAT: No such file or directory
> cat: -I/sw/include: No such file or directory
>
>
> If typed explicitly, the command $COMM seems to work OK,
> (which is actually what is done in the second attempt to
> test netcdf with -lnetcdf added)
>
> Looks fishy to me, this warning about preprocessing
> seems to be broken (and misleading), no ?
> Could it be fixed ?
>
> Anyway my problem has probably something to do
> with  the "name-mangling" discussed on the support list a while ago.
> Ed suggests to install netcdf by hand and points to some  
> documentation:
>
> http://mitgcm.org/mwiki/index.php/NetCDF_Local_Install
>  [ ae ehnd http://mitgcm.org/mwiki/index.php/MNC_Package ]
>
> but I can't access these pages: have they disappeared ?
>
> Thanks,
> david
>
> -- 
>
> Ferreira David               Tel : 617 253 7967 EAPS Room 54-1515,
> Massachusetts Institute of Technology,
> 77 Massachusetts Avenue
> Cambridge, MA, 02139
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel