20040213: installation problem with netCDF

Mohamed,

>Date: Mon, 16 Feb 2004 15:56:58 +0200
>From: "Mohamed A. Elzeir" <address@hidden>
>Organization: El-Minia University, Egypt
>To: "Steve Emmerson" <address@hidden>
>Subject: RE: 20040213: installation problem with netCDF 
>Keywords: 200402061657.i16GvAp1007360 netCDF install

The above message contained the following:

> I followed the steps. Results are similar to those obtained when I added
> the line
> CC =
> before line 1021 in configure file, i.e.; the C-compiler was successful
> while the C++ compiler was not. The error is reported in config3.log
> (attached).

To explain to others who might read this, the error reported in the 
file "configur3.log":

    checking C++ compiler "/gad_opt/SUNWspro/bin/CC"... configure: warning: 
/gad_opt/SUNWspro/bin/CC failed on standard C++ test program
    configure: warning: Could not find working C++ compiler
    configure: warning: Setting CXX to the empty string
    configure: warning: The C++ interface will not be built

is explained in the file "config3.log":

    configure:1225: checking C++ compiler "/gad_opt/SUNWspro/bin/CC"
    configure:1243: /gad_opt/SUNWspro/bin/CC -o conftest  -DNDEBUG  conftest.C  
1>&5
    Undefined                       first referenced
     symbol                             in file
    std::bad_alloc::__vtbl          
/gad_opt/SUNWspro/SC5.0/lib/libCstd.a(zOtiZS4hQ7
    CvMDXtQoDv.o)
    [Hint: try checking whether the first non-inlined, non-pure virtual 
function of 
    class std is defined]

    void __Cimpl::cplus_init()        (command line)
    ld: fatal: Symbol referencing errors. No output written to conftest
    configure: failed program was:
    #line 1230 "configure"
    #include "confdefs.h"
    #ifdef __cplusplus
    //extern "C" void exit(int);
    #endif

                                    #include <iostream>
                                    int main() {
                                        std::cout << "";
                                        return 0;
                                    }

Your C++ compiler should have been able to build this test program.  The 
failure indicates a prblem with the C++ installation.

> I do not know how to fix that error which is related to
> undefined symbol.

It looks like you have subdirectories under "/gad_opt/SUNWspro/" that
contain version-specific compiler installations.  It is possible that 
the "/gad_opt/SUNWspro/bin/CC" compiler is actually a link to a version-
specific compiler in another subdirectory and that that installation
isn't quite complete.

Can you ask advice from a system administrator on this problem.  You might
be able to experiment with the compiler installations (copying libraries,
creating symbolic links, etc.) using the above test program until it works.

> I used the commands:
> echo $CC, and echo $CXX after executing CONFIGURE, I found that values
> are not changed:

Those values would not be changed because the configure script executes in
a child process and a child process can't affect the environment variables
of its parent process.

> CC=/gad_opt/SUNWspro/bin/c90, and
> CXX=/gad_opt/SUNWspro/bin/CC
> although the report in configure.log is:
> setting CXX to the empty string
> The C++ interface will not be built
> 
> I ran the other commands:
> make >& make.log,
> make test >& test.log
> make install >& install.log
> 
> They ran smoothly with only one error:
> Not making 'cxx/all, test, install' because no C++ compiler.

Well,  you know have C and Fortran versions of the netCDF library.  Do you
need the C++ version?

> Executable files ncdump and ncgen were generated.
> 
> I will try to use netCDF in a FORTRAN interface and let you know if it
> works.
> 
> Below is my reply to your question why I came from hot dry area to
> completely different environment:
> 
> I was moving around the world between hot and cold and dry and wet
> places since 1989 when I was granted a M.S. scholarship from the Belgian
> minisry of foreign affairs, then extended for a ph.d., then got a
> post-doc fellowship in Japan (in Yokosuka). Now, I have an initial
> two-year contract in Aberdeen, the UK. Two essential motives make me
> move around the world. Both of the motives are enhanced by the situation
> in Egypt (and common to all developing countries). The first motive is
> the scientific one; as a university staff, I should conduct research to
> be promoted. Egypt's (as a developing country) research facilities do
> not give any opportunity for the native people to conduct any serious
> research. Although some university staff manage to submit
> less-than-minimum-standard research for promotion, I preferred the harsh
> route of competing for international opportunities. I should admit
> competition is fierce and chance of people like me is extremely slim.
> However, I manage to hunt a chance every couple of years where I publish
> as many articles as I can then return back to my Egyptian university to
> resume my duties.
> The second motive is financial. There is no fund for research in Egypt
> and industry is not yet up to the level of funding research projects.
> Hence, I should seek the research fund from outside the country. The
> opportunities to get a complete project fund are not available for
> developing countries. Hence, the alternative is to seek working as part
> of an existing funded project (like EUROGEL in which I am a researcher
> now).
> I am sorry to disturb you with all these problems. However, I assume you
> are aware of them because of your stay in our area for sometime.

I understand completely.

> Best regards
> Mohamed
> http://www.geocities.com/melzeir/

Regards,
Steve Emmerson