Installation Problems. Please help.

Dear netCDF Developers

I would like to report an installation problem. Please help, since I 
would really like to use netCDF as soon as possible. I am working on 
a Decstation 5000 running Ultrix 4.2a. I am trying to install netCDF
2.02. I removed xdr_float.o from my libc.a and replaced it with xdrfloat.a 
from the netCDF distribution. I also set:

       LIBTYPE= nonsharable
       OS     = ultrix_4.2

in the top Makefile. BTW, there is no file fortran/NOTES as referred
to in the INSTALL file.

1. If I set XDR_INC = all in the top Makefile as recommended in the
INSTALL text I get the error:

cpp: error xdrtest.c:29: Can't find include file types.h
cpp: error xdrtest.c:30: Can't find include file xdr.h
*** Error code 1

I fixed this by changing <types.h> and <xdr.h> in xdrtest.c to "types.h" 
and "xdr.h". Everything compiles OK but the system fails the fortran
tests thus:

 --- Testing ncredf, ncdren, ncvren, ncaren, ncendf ...
ncendef: rename failed: Cross-device link
 error in ncdinq or ncddef, dsize=           0
 error in ncdinq or ncddef, dname=lat                            
 error in ncvinq or ncvdef
 error in ncanam  or ncapt
 error in ncanam        or ncapt
 --- Testing ncacpy ...
 error in ncagt G
 error in  ncagt CHAR N
 --- Testing ncadel ...
ncendef: rename failed: Cross-device link
*** Termination code 11

2. If I don't define XDR_INC in the top Makefile then everything compiles
OK with no changes to xdrtest.c required. However the system again fails 
the Fortran tests exactly as before.

3. Exactly same problem when I tried CC=gcc etc..

What should I do to get rid of these errors ? Is it essential to get
rid of these errors if I want to use the Fortran interface ?

       best regards
                Ata <(|)>
| Mail          Dr Ata Etemadi, Blackett Laboratory,                          |
|               Space and Atmospheric Physics Group,                          |
|               Imperial College of Science, Technology, and Medicine,        |
|               Prince Consort Road, London SW7 2BZ, ENGLAND                  |
| Internet/Arpanet/Earn/Bitnet atae@xxxxxxxxxxxxxxxx or ata@xxxxxxxxxxxxxxxx  |
| Span                              SPVA::atae       or     MSSLC:atae        |


>From owner-netcdfgroup@xxxxxxxxxxxxxxxx    18 Wed, Sep
Date:    Wed, 18 Sep 1991 15:01:28 -0400 (EDT)
From: REZA@xxxxxxxxxxxxxxx
To: netcdfgroup@xxxxxxxxxxxxxxxx
Subject: Fill problem
Received: by unidata.ucar.edu id AA09789
  (5.65c/IDA-1.4.4 for netcdfgroup-send); Wed, 18 Sep 1991 13:03:42 -0600
Received: from uri.gso.uri.edu by unidata.ucar.edu with SMTP id AA09785
  (5.65c/IDA-1.4.4 for <netcdfgroup@xxxxxxxxxxxxxxxx>); Wed, 18 Sep 1991 
13:03:38 -0600
Message-Id: <910918150128.202020aa@xxxxxxxxxxxxxxx>
X-Vmsmail-To: SMTP%"netcdfgroup@xxxxxxxxxxxxxxxx"


        I just installed Ver 1.11 of netCDF, and have a problem with
setting fill values for short integers. It fills only every other value with
my fill value (0) and the rest with the default fill value (-32767). Does
anybody know what is the problem? 

Here is the NCDUMP output:

netcdf fcoads {
dimensions:
        lat = 2 ;
        lon = 1 ;
        time = 2 ;
        stat = 8 ;

variables:
        short A(time, lat, lon, stat) ;
                A:units = "Celsius" ;
                A:long_name = "air temperature" ;
                A:_FillValue = 0s ;
                A:missing_value = 0s ;

data:

 A
  -32767, 0, -32767, 0, -32767, 0, -32767, 0,
  -32767, 0, -32767, 0, -32767, 0, -32767, 0,
  -32767, 0, -32767, 0, -32767, 0, -32767, 0,
  -32767, 0, -32767, 0, -32767, 0, -32767, 0 ;


}

Thanks, Reza
REZA@xxxxxxxxxxxxxxx