>From: Alain Mazurier <mazurier@xxxxxxx>
>Keywords: 200104041159.f34BxML07668
Hi Alain,
> I am trying to append a Netcdf file ,that it is already exist ,with a new
> variable and new data:
> After appending, when i am looking to the file, it didn't
> append it but it has overwritten data ?
> I have the same size between the file before and after appending
>
> this is my code:
>
> NcFile* lp_fichier;
> lp_fichier = new NcFile( aiv_file, NcFile::Write);
> NcDim* lp_dim[2];
> lp_dim[0] = lp_fichier->get_dim("mbCycleNbr") ;
> lp_dim[1] = lp_fichier->get_dim("mbBeamNbr") ;
>
> NcVar* lp_var = lp_fichier->add_var((NcToken )"mbAzimutBeamAngle",
> ncShort,(const NcDim*)lp_dim[0], (const NcDim*)lp_dim[1]);
> lp_var->add_att("type", "real");
> lp_var->add_att("long_name", "Beam azimut");
> lp_var->add_att("name_code", "MB_AZIMUT_BEAM_ANGLE");
> lp_var->add_att("units", "degree");
> lp_var->add_att("unit_code", "MB_DEGREE");
> lp_var->add_att("add_offset", 0.0);
> lp_var->add_att("scale_factor", 0.01);
> lp_var->add_att("valid_minimum", 0);
> lp_var->add_att("valid_maximum", 35999);
> lp_var->add_att("missing_value", 32767);
> lp_var->add_att("format_C", "%.2f");
> lp_var->add_att("orientation", "direct");
>
> long lv_nbData = lp_dim[0]->size()*lp_dim[1]->size();
> short* lp_tabData = new short[lv_nbData];
> for (long lv_i=0; lv_i< lv_nbData;lv_i++)
> lp_tabData[lv_i]=32767;
> lp_var->put(lp_tabData, lp_var->edges());
>
> lp_fichier->close();
> // liberation des ressources
> delete [] lp_tabData;
>
> Have you an idea or better a solution
I can't duplicate the problem without an example input file, but I
suspect this is an example of the "redef bug" in netCDF 3.4 that
occurred under certain circumstances when adding a new variable. In
terms of the underlying C interface, the bug occurred when
... nc_enddef() is called after nc_redef(), the file is growing such
that the new beginning of a record variable is in the next "chunk",
and the size of at least one record variable exceeds the chunk size
(see netcdf.3 man page for a description of this tuning parameter
and how to set it). This bug resulted in corruption of some values
in other variables than the one being added.
This bug is fixed in the just-released netCDF 3.5.0, so you could
verify whether this was the problem by trying your program with the
new release. See the next message to the netcdfgroup mailing list for
an announcement of netCDF 3.5.0 and how to get it.
If this doesn't fix the problem, please let us know and provide access
to an example file that demonstrates the problem.
--Russ
_____________________________________________________________________
Russ Rew UCAR Unidata Program
russ@xxxxxxxxxxxxxxxx http://www.unidata.ucar.edu
