960102: Problems running under OSF1 version 4.0 on an Alpha station

Alison,

> To: address@hidden
> cc: Alison Macdonald <address@hidden>
> From: Alison Macdonald <address@hidden>
> Subject: Problems running under OSF1 version 4.0 on an Alpha station
> Organization: Oregon State University
> Keywords: 199701010014.RAA03943

In the above message you wrote:

>       I have been running the GMT (Generic Mapping Tools) software package 
> out of the University of Hawaii which uses NetCdf. I have an Alpha 
> workstation previously running version 3.2b of the operating system, now
> running 4.0. Since the upgrade I have not been able to create netcdf files
> through the GMT calls to netcdf routines. I have re-installed both
> NetCdf and GMT. 
>
>       The test case I am using is a call to the gmt program xyz2grd. 
> It opens the file using ncopen, calls ncvarid, ncattget, ncvarget
> and ncendef.  The file exists after the open, but after the call to
> ncendef, the variable ncerr has been set = 32.

Hmm...  A error code of `32' indicates a failure to XDR something.

> It then proceeds to
> call ncvarput and ncclose. The file disappears at ncclose and the
> program produces an error when it checks the ncerr value.
>
> GMT and NetCdf did not have this problem under the previous version
> of the operating system and the test routines supplied with netcdf,
> do not produce any errors (other than a warning in the test programs
> themselves about strings being continued over lines). I wondered if
> you had any experience with the software being run under this new
> operating system that might give me clue as to the source of the
> problem.

My workstation is OSF1 V4.0B (it used to be V3.2 and V4.0).  As far as
I can tell, the netCDF 2.4.3 package works correctly on this system.
Therefore, I suspect that the problem lies with the GMT package (which
we don't have here).  The only realistic hope in fixing a hypothetical
problem in the netCDF package would seem to lie with either you or the
GMT people getting me a small piece of C code that illustrates the
problem.

> 
> I am also writing to the GMT people because although the error occurs in 
> the NetCdf routines, I don't know how to reproduce the problem outside of
> GMT.

If you converse with the GMT people, please tell them that I'll be happy
to assist in any way that I can.

>               I would be grateful for any assistance you can offer
> 
>                       Alison Macdonald 
> 
> 
> ps. Some of the things I have checked include:
>       removing the optimization  (same error occurs)
>       the write access on the file  (I appear to have write access)
>       going back to the previous version of the  operating system (still 
> works)

This last one is odd because, if the low-level error is in the XDR
layer, then I would think that the problem would manifest itself under
both versions of the OS.

--------
Steve Emmerson   <address@hidden>