20000505: utdec() error reading NCEP reanalysis data

Eunjoo,

>Date: Fri, 05 May 2000 12:00:24 +1000
>From: Eunjoo Jung <address@hidden>
>Organization: School of Mathematics UNSW 
>To: Steve Emmerson <address@hidden>
>Subject: utdec() error reading NCEP reanalysis data
>Keywords: 200005030232.e432WhG20618

In the above message, you wrote:

> Thank you for prompt reply again.
> I have two "udunits.inc" files.
> 
> 1)one is from the ftp.cdc.noaa.gov/Software/udunits.inc
> 2)the other is from the udunits-1.11.7/src/lib
> 
> when I use the first one, there is no problem compiling and linking.
> but I've got running error.

The file <ftp://ftp.cdc.noaa.gov/Software/udunits.inc> declares that the
"utmake()" function returns an "integer*4".  This is incorrect for a Dec
Alpha.  On that system, the "utmake()" function returns an "integer*8".

> otherwise I use the second one, I couldn't compile the fortran program.
> here are the compile error message I've got.
> c-----------------------------------------------------------------
> fort: Error: ./udunits-1.11.7/include/udunits.inc, line 27: Unsupported or
> malformed preprocessor directive.
> #ifndef PTR
> --^

Because the offending line above strictly complies with the C standard,
it would appear that your "fort" compiler isn't preprocessing the file
using a Standard C preprocessor.

It appears that NOAA hard-coded a pointer to be equivalent to a Fortan
"integer*4" -- which is wrong for your system.  I suggest the following:

    1.  Edit NOAA's "udunits.inc".  Change the "integer*4" return value
        of "utmake()" to "integer*8".

    2.  Edit NOAA's "read.f" file.  Change the type of the variable
        "unitptr" from "integer*4" to "integer*8".

Regards,
Steve Emmerson   <http://www.unidata.ucar.edu>