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

Dirk Slawinski: Re: 20001106 NetCDF and Lahey

>From: Dirk Slawinski <address@hidden>
>Subject: NetCDF and Lahey
>Organization: University of Western Australia
>Keywords: 200011060521.eA65Lf419935 netCDF Linux Lahey Fortran

------- Forwarded Message

Date:    Fri, 10 Nov 2000 16:26:55 +0800
From:    Dirk Slawinski <address@hidden>
To:      Russ Rew <address@hidden>
Subject: Re: 20001106 NetCDF and Lahey 


>I typed in the keyword "Lahey" in the search in the "netCDF support
>email archive" at the bottom of the netCDF home page
><http://www.unidata.ucar.edu/packages/netcdf/>, and it pointed to the
>following:
>
>   http://www.unidata.ucar.edu/glimpse/netcdf/3903
>
>which reports on successfully compiling netCDF for Lahey f95 using the
>following environment variable settings:
>
> >     setenv CXX /usr/bin/g++
> >     setenv CC /usr/bin/gcc
> >     setenv CFLAGS '-march=pentiumpro -O2'
> >     setenv FC /usr/local/lf95/bin/lf95
> >     setenv CPPFLAGS -DNAGf90Fortran
> >
> > The library compiles, but "make test" stops here:
> >
> > ...
> > make[2]: Entering directory `/home/hnv/netcdf/netcdf-3.4/src/fortran'
> > + /usr/local/lf95/bin/lf95 -c -O ftest.F
> > "ftest.F", line 7: error: #include: cannot find file "nfconfig.inc"
>
>and then suggests more about how to get around the missing
>"nfconfig.inc" problem.  This sounds like it got further in building
>with Lahey on a Unix platform, so maybe that will help for your
>attempt to build from source on Linux.  We still don't have access to
>Lahey compilers here, which makes it difficult for us to try to debug
>problems with Lahey compilers.

   Yes it is a UNIX/Linux based one and that seems to be ok according to 
folks at Columbia.  Unfortunately even though the Win32 Lahey compiler "is 
the same" as the Linux one according to Lahey there are still issues.  At 
present I am rather close though.  I had to modify configure to take into 
account that Lahey uses .obj files under Win32 and that -o <filename> is 
-out <filename>.  I then had to change the fortran/*.c files so to account 
for the extra "_"s due to the nf/nc naming convention:

   #define nf_open_ nf_open
   FCALLSCFUN3(NG_INT....

the internal libname is mangled to that nf_open does not become _nf_open_ 
but rather _nf_open__.  Now it is a matter of getting lahey to recognize 
the embedded libs in libnetcdf.a (C routines).

   If you have any more suggestions that would be great.  If not then oh 
well.  Either way once I have this all figured out I'll email you guys the 
bits that you need to add or should add to your distribution.


>I saw your posting on Nov 2 to the netcdfgroup mailing list asking
>whether anyone else had had any success compiling netCDF on Win32
>platforms with Lahey compilers.  Did anyone respond?
>
>There was another note to the netcdfgroup in February that
>reported some success with netCDF using a Lahey compilers on a
>Fujitsu/Linux platform that may be of some help:
>
>   http://www.unidata.ucar.edu/glimpse/netcdfgroup-list/1297
>
>--Russ
>
>_____________________________________________________________________
>
>Russ Rew                                         UCAR Unidata Program
>address@hidden                     http://www.unidata.ucar.edu

---
                                                      TTFN  Dirk  =:>

/--------------------------------------------------------------------\
|    Dirk Slawinski      |      The Centre for Water Research        |
|  Research Associate    |     University of Western Australia       |
|                        |                                           |
| PHONE:  +61 8 9380 2410  (CWR)                      ,-_|\          |
|   FAX:  +61 8 9380 1015  (CWR)                     /     \    ]]   |
| EMAIL:  address@hidden           (Perth) *_,-._/   d")   |
|   WEB:  http://www.cwr.uwa.edu.au/~slawinski/           v    _(@)_ |
\--------------------------------------------------------------------/

------- End of Forwarded Message