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

Re: netcdf on NeXT

> Organization: .
> Keywords: 199402071747.AA24476 NeXT

Hi Kelley,

> someone in my group, doug cripe, sent me some email conversions
> he had with you and others in unidata about getting ncdinq
> to work.  you had suggested making sure we had the latest
> version of netcdf.  well, we have an older version so i
> ftp'd down to unidata and grabbed 2.3.2.  (we had 2.02)
> 
> i'm running w/ NeXTSTEP 3.2 on an 040 cube with the latest
> version of the absoft 77 compiler.

I don't know of anyone who has ever successfully gotten the netCDF Fortran
interface built for NeXTSTEP using the absoft Fortran compiler.  Craig
Mattocks (address@hidden) got it built under NeXTSTEP 3.0
using the f2c version of f77.

> i ran configure and it was unable to figure out my OS,
> MACHINE nor FC.  well, our f77 compiler is called af77
> and NeXT does not recognize uname.  

The configure script depends on the C precompiler to predefine a suitable
macro for the operating system.  In the case of NeXTSTEP, it uses

    #ifdef NeXT
    OS_nextos
    #endif

so it assumes "NeXT" is predefined, to get OS defined as "OS_nextos".  This
used to work.  If you know of or can determine a different constant that is
now used, e.g.  "__NeXT", we'd like to know about it so we can fix the
configure test.

To account for a missing "uname", you could just create a shell script
called "uname" and put it in a directory that will be found by the configure
script when it runs.  The only uname arguments the configure script uses are
"-m" for machine architecture and "-r" for operating system revision, so
your uname should output "NeXT" when invoked with a "-m" argument and "3.2"
when invoked with a "-r" argument.

> so in CUSTOMIZE script, i added the following lines:
> 
> # fortran call for us  -- kwitt
> FC=/usr/local/bin/af77
> #
> # os for us, configure isnt working for this  -- kwitt
> OS=OS_nextos
> #
> # machine for us, uname is non-existent for configure  -- kwitt
> MACHINE=NeXT

That should have worked OK too.

> i must have done something wrong here because i ran configure
> again and my machine crashed.  well, it locked up completely.
> i thought it was something else so i ran it again later
> on in the afternoon.  boom.  unfortunately, i did not
> see what the last message from configure was and i dont want
> to crash the thing again as kiwi is our server and netmaster.
> 
> can you spot anything off the top of your head?  if you guys
> have NeXT libraries already compiled, would we be able to
> grab them?

I've never heard of a machine crashing when the configure script was run,
and I don't have any ideas about what might cause this.  All the configure
script does is compile and run some tiny programs to determine properties of
the environment by compile-time or run-time tests.  If the OS or MACHINE
environment variables have some special meaning for SeXTSTEP, that could
cause problems.

We don't have any precompiled versions of the libraries for NeXT here at
Unidata, though it's possible Craig Mattocks or Glenn Davis
(address@hidden) might have something.  To see what is causing the
crash, you will probably have to run the configure script in verbose mode
with output to a file, e.g

  % sh -x configure > configure.out 2>&1

--Russ