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

Re: 19990913: netCDF, Win32, suggestion (for readme 32/64 Bit)

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 Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated.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: 19990913: netCDF, Win32, suggestion (for readme 32/64 Bit)
  • Date: Wed, 15 Sep 1999 15:55:40 -0600
>To: address@hidden
>From: Peter Paul Smolka <address@hidden>
>Subject: netCDF, Win32, suggestion (for readme 32/64 Bit)
>Organization: .
>Keywords: 199909131433.IAA24351

Hi Peter,

> 1) It is EXTREMELY (italic) helpful for ALL users of the precompiled
> versions of netCDF (3.5 or so) the readme file says whether
> the real numbers (floating point) are compiled for REAL*4 (32 Bit)
> or REAL*8 (64 Bit).

Yes, sorry, we'll try to include something with the binary
distributions in the future that lists the compilers and flags used to
build them.  On a Unix system, this would just be the output of a
script like this:

 #!/bin/ksh
 uname -a
 for i in CC CPPFLAGS CFLAGS COMPILE_F FC FFLAGS CXX CXXFLAGS LIBS
     do
         rhs=$(eval print -- '$'$i)
         case "$rhs" in
         *" "*) print $i=\"$rhs\" ;;
         "")    ;;
         *)     print $i=$rhs
         esac
     done

but I'm not sure how to do it on systems like Windows NT, where the
information is scattered around in Makefiles.

> I use ccm3.6 under Windows NT (compiled for 64 Bit) and
> making a blind test with a new netcdf (3.5) to check whether
> garbage results or not is not so nice.
>
> If they are NOT compiled for 64 Bit (REAL*8), I suggest to do so.
>
> In this context however Characters should be 8 bit and Integers
> 32 Bit (4 Byte, INTEGER*4)

We have to worry about backward compatibility and providing only one
or two binary distributions for each platform.  This often means
creating a 32-bit version since it works on the 64-bit platforms but
not vice-versa.  However, we'll consider your suggestion for the next
round of binary distributions.

> 2) ccm2nc
>
> There is a very nice utility concerting the ccm history file to
> netcdf format.
>
> It might also be very practical to put a respective 64 Binary
> in this directory (OR on the NCAR page under utilities where
> the source comes from).
>
> I DO know that it is not your task to do this.
>
> I am only speaking about places in the hall of fame....

OK, sign me up, I just added an entry about ccm2nc to our
user-contributed software catalog at

  http://www.unidata.ucar.edu/packages/netcdf/contrib.html

where search engines and users may be able to find it.  Thanks for the
information!

--Russ

_____________________________________________________________________

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