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

Re: 19990623: netCDF problems on Linux RH 6.0



>To: address@hidden
>cc: FABART Veronique <address@hidden>
>From: POLCHER Jan <address@hidden>
>Subject: netCDF problems on Linux RH 6.0
>Organization: Laboratoire de Meteorologie Dynamique du CNRS
>Keywords: 199906231322.HAA20363

Hi Jan,

> I have compiled netCDF on a new Linux RH 6.0 workstation and get some
> strange errors when I apply it. I guess it is linked to the new glibc as
> this is one of the major changes between the new system and the old one
> where everything works fine.  First let me explain how I go there.
> 
> I have compiled the netCDF lib and performed the tests without any
> problems on the RH 6.0 system. I did this first with pgf77 and then with
> g77. To this point everything was OK. By the way, using the RH 5.4 library
> in RH 6.0 gives undefined externals at link time.
> 
> The I applied the library in a real life problem and I noted that when the
> definition of variables goes beyond a given size it starts to corrupt the
> data which is already in the file. Very strange problem which happens
> with the g77 and pgf77 compilers !
>
> To check that the problem is not with my code I have compiled it with the
> same compilers and the netCDF library for RH 5.4 on a RH 5.4
> system. It ran without any problems. The same executable also works
> perfectly on a RH 6.0 system !
> 
> This is point at which I am stuck now. My questions are simple :
> - Have you already heard from such a problem.
> - Do you know why netCDF would not work with the new glibc
> - Do you have any ideas on how to resolve the problem or what can I
> do to help you find a solution.

Oops, I think this might be a rarely encountered (and hard to
reproduce) netCDF bug that is fixed in netCDF version 3.5, (a test
version is available from
<ftp://ftp.unidata.ucar.edu/pub/netcdf/netcdf-3.5-alpha.tar.Z>).

I cannot explain why the same code would work on a RH 6.0 system that
was compiled on a RH 5.4 system, but the description of the problem
otherwise sounds like the bug that we finally found a fix for.  If you
don't want to try the 3.5 alpha release referred to above, you can
also try the patch to netCDF 3.4 contained in the email support
message:

  http://www.unidata.ucar.edu/glimpse/netcdf/3379

If either of these fixes the problem, I'd like to know about it.
Thanks for reporting it ...

--Russ

_____________________________________________________________________

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