Re: 20041226: re 20041215 email from delire

>To: address@hidden
>From: Christine Delire <address@hidden>
>Organization: UCAR/Unidata
>Keywords: 200412151620.iBFGK0lI018167

Hi Christine,

> I sent an email yesterday about a problem installing netcdf on a G5
> (see below). Our email server is having problems. I haven't received
> any message for the last 24 hours. If you sent something, could you
> send it again at this address:
> 
> address@hidden

Sorry to take so long replying.  We're shorthanded this week and
trying to get out the final release of 3.6.0 (maybe later today).

> I am trying to install netcdf on a G5 running OS 10.3.6 (Panther).
> I did install it first with fink but a fortran 77 code that was working 
> fine on a G4 now crashes with a 'bus error' when calling a netcdf 
> function for the first time. So I am trying to re-install it.
> 
> The 1st error appears when testing the installation. There seems to be 
> a conflict of precision (which is coherent with the error I had with 
> the version installed with fink).

What you are seeing in the output from "make test" when it is testing
the old netCDF version 2 Fortran interface, are just warnings, not
errors:

  Making `test' in directory /Users/delire/Desktop/netcdf-3.5.1/src/fortran

  + g77 -c ftest.F
  ftest.F: In subroutine `tncacpy':
  ftest.F:187: warning:
                 call ncagt (outcdf, NCBYTE, attnam, bytval, iret)
                      1
  ftest.F:196: (continued):
                 call ncagt (outcdf, NCGLOBAL, attnam, shval, iret) 
                      2
  Argument #4 of `ncagt' is one precision at (2) but is some other precision at 
(1) [info -f g77 M GLOBALS]
  ftest.F:187: warning:
                 call ncagt (outcdf, NCBYTE, attnam, bytval, iret)
                      1
  ftest.F:198: (continued):
                 call ncagt (outcdf, NCGLOBAL, attnam, lngval, iret)         
                      2
  Argument #4 of `ncagt' is one precision at (2) but is some other precision at 
(1) [info -f g77 M GLOBALS]
  ...

These are expected, because the old version 2 interface used type
punning to return values of multiple types in the same argument.  All
the tests work after these compiler warnings, so you can ignore them.
They aren't the result of something wrong with g77, but are to be
expected.

Since the Fortran tests ran fine using g77, I don't think the cause
of your "bus error" is the netCDF library.  If you compile and link
your Fortran program the same way they are compiled and linked in the
test:

  g77 -o PROG -I/Users/delire/Desktop/netcdf-3.5.1/include PROG.f 
-L/Users/delire/Desktop/netcdf-3.5.1/lib -lnetcdf

(where PROG is the real name of your file/program), it should work
fine, just as the tests did.  If you can isolate a small example that
fails and send it to us, we could try to reproduce it here.  The fact
that it worked on the G4 but not on the G5 may mean you have to
recompile everything that's used in the Fortran program, not just the
netCDF library.

--Russ