[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 941223: netCDF fortran to C

>Organization: University of Oklahoma
>Keywords: 199412231532.AA15319 netCDF CRAY Unicos m4

Hi Adwait,

> > >   I was trying to link libnetcdf.a with my fortran source when I 
> > > got some errors. A little digging turned up that that Fortran to C calling
> > > library had not been compiled properly. I tried stepping through the code
> > > but I could not find the source of the error. I am sending you part of 
> > > the 
> > > make.log which shows the error Please let me know how I can fix it. 
> > > Thanks,
> > 
> >  ...
> > >         ./fortc -L . -O unicos common.inc > netcdf.inc
> > >         ./fortc -L . -O unicos jackets.src > jackets.c
> > > 
> > > m4:-:529 can't open file
> > 
> > This worked fine on our tests in building under UNICOS.
> > 
> > To diagnose this problem, I will need a little more information.  First,
> > what version of UNICOS are you using, and which version of "m4" are you
> > using.  To get this information, it may suffice to run the following two
> > commands:
> > 
> >  % uname -a
> >  % which m4
> > 
> > The diagnostic above is emitted by the m4 macroprocessor, invoked from the
> > shell script netcdf/fortran/fortc.  In the netcdf/fortran directory, please
> > run the following commands and mail me back the output produced in files
> > "tmp1" and "tmp2".
> > 
> >  % sed -f fortc1.sed jackets.src > tmp1
> >  % m4 common.m4 unicos.m4 tmp1 > tmp2
> > 
> 
> Russ,
> 
>       This is what I got...
> 
> mario% uname -a
> mario mario 8.0.2 mor.130 CRAY C90
> mario% which m4
> /usr/bin/m4
> mario% sed -f fortc1.sed jackets.src > tmp1
> mario% m4 common.m4 unicos.m4 tmp1 > tmp2
> 
> m4:tmp1:529 can't open file
> divert(1)
> 
>       tmp1 is quite big, so I will send it to you in another message. 

This is apparently a bug in the /usr/bin/m4 utility on your Cray, since I
just tried 

    m4 common.m4 unicos.m4 tmp1 > tmp2

with the version of the `tmp1' file you sent and two other versions of m4,
and they both ran to completion with no errors, producing the expected
output in `tmp2'.

Here are three possible alternative workarounds:

  1.  You could run the following command from the netcdf/fortran/ directory
      on a different UNIX platform than UNICOS to produce the `jackets.c'
      file that is needed to complete the building of the Fortran netCDF
      interface for UNICOS:

      % m4 common.m4 unicos.m4 tmp1 | sed -f fortc2.sed > jackets.c

      Then copy the resulting `jackets.c' file to your UNICOS
      netcdf/fortran/ directory and proceed with the "make all test" from
      the top-level netcdf/ directory, which should now not try to use m4 to
      generate the jackets.c, since it has a later modification time than
      the sources from which it is derived.

or  2. I could send you the `jackets.c' file that results from my running the
      above command here.  I'm not positive such a large file will make it
      through the mail successfully to your mailbox, but the tmp1 file you
      sent made it here OK.  Please let me know if you would prefer to try
      this.

or  3.  Build the GNU m4 on UNICOS and use that.  It is available from
      ftp://prep.ai.mit.edu/pub/gnu/m4-1.4.tar.gz.

Before the next release, we'll verify that the distribution builds OK on the
current version of UNICOS, providing some other workaround if necessary.

______________________________________________________________________________

Russ Rew                                                UCAR Unidata Program
address@hidden                                          P.O. Box 3000
http://www.unidata.ucar.edu/                          Boulder, CO 80307-3000
______________________________________________________________________________