20030717: netcdf make test failures using Intel compilers on Linux

Curt,

>Date: Thu, 17 Jul 2003 12:23:38 -0400
>From: Curt Hillegas <address@hidden>
>Organization: OIT - Princeton University
>To: Steve Emmerson <address@hidden>
>Subject: Re: 20030717: netcdf make test failures using Intel compilers on Linux

The above message contained the following:

> I had been pointing at some old instructions which are similar, but not
> identical (they give an extra option to try) -
> http://www.unidata.ucar.edu/packages/netcdf/other-builds.html#intel-ifc
> 
> But using the instructions you pointed me to does not work either.  I
> have attached the relevant files.  You may notice in the test.log that
> the first error is actually very early on (see line 22 of that file). 

That line is

    fill_seq indices         1 2 3  75.000000 != 2.718282

and isn't an error.  The first error is in the "nc_test" subdirectory:

    *** Testing nc_put_var_float ...
            FAILURE at line 2831 of test_put.c: Numeric conversion not 
representable

This might have more to due with a bug in our testing code than with any
problem of the netCDF package.  We've seen this before.  The testing
code uses what should be extreme values (e.g., what should be the largest,
representable floating-point value) to test the interface.

I suggest that you try the package on some netCDF files and in your
programs.  If it seems to work, then it's probably OK.

Regards,
Steve Emmerson