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

Re: 970212: netcdf2.4.3 build errors on SGI Challenge

This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.

  • Subject: Re: 970212: netcdf2.4.3 build errors on SGI Challenge
  • Date: Thu, 13 Feb 1997 08:53:49 -0700
>To: address@hidden
>From: "Neil R. Smith" <address@hidden>
>Subject: netcdf2.4.3 build errors on SGI Challenge
>Organization: Dept. Meteorology/Texas A&M University
>Keywords: 199702130028.RAA07186

Hi Neil,

> I'm trying to build netcdf 2.4.3 in a user's directory on a 24-cpu SGI
> POWER CHALLENGE 10000 XL running IRIX 6.2.  I attach the following
> logged output: config.log, make log, make test log.  
> 
> If you have the time or patience, I would greatly appreciate your help
> in determining the problem.

From the output you've sent, it looks the library tests of the C and
Fortran interfaces worked perfectly, but the test of the ncdump utility
failed:

 ...
> making `test' in directory /lite/bowman/lib/netcdf/netcdf-2.4.3/src/ncdump
> 
>       ../ncgen/ncgen -o test0.nc -n test0.cdl
>       ./ncdump test0.nc > test1.cdl
>       ../ncgen/ncgen -o test1.nc -n test1.cdl
>       ./ncdump -n test0 test1.nc > test2.cdl
> test1.cdl test2.cdl differ: char 987, line 69
> *** Error code 1 (bu21)

The message is saying that test1.cdl and test2.cdl, which should be the
same if the ncdump and ncgen utilities are working, differed on line
69.  The differences:

    % diff test1.cdl test2.cdl
    69c69
    <  dscalar = _ ;
    ---
    >  dscalar = 1.32640798021345e-315 ;
    75c75
    <  snodata = _, _ ;
    ---
    >  snodata = 0, 0 ;
    77c77
    <  lnodata = _, _ ;
    ---
    >  lnodata = 0, 0 ;
    79c79
    <  fnodata = _, _ ;
    ---
    >  fnodata = 0, 0 ;
    81c81
    <  dnodata = _, _ ;
    ---
    >  dnodata = 1.32640798021345e-315, 1.32640798021345e-315 ;

indicate a problem with a delicate test for the default special values
used to indicate missing data.  We did not see this problem with IRIX
6.1, but I managed to duplicate it on our Power Indigo2 R8000 running
IRIX64 6.2.

For now, you can assume the C and Fortran libraries work fine, and you
can "make install" to finish the installation.

The bug you have found with ncdump will only cause problems for
displaying missing values.  We'll let you know when we find a fix or
workaround.

Thanks for reporting the problem!

_____________________________________________________________________

Russ Rew                                         UCAR Unidata Program
address@hidden                     http://www.unidata.ucar.edu