[petsc-dev] status of Netcdf in PETSc

[Barry Smith]

Ok so Satish was making it work in 2005. 

29bfbbabe8cc0fcc7eee07383f3fad5d2b37f8ec

But then Shri was removing it in 2010. (Probably was never used in the intervening 5 years).


fde6f1ec0c1dfc2ab7388d2373e012536937051f  
Removing support for NetCDF from new vecload interface because NetCDF requires HDF5 for parallel I/0.
(Seems a weak reason to remove, but ...)

8bc7493c1bc1e61bf7b6e5f49460319d964a0b54


Anyways I'm going to remove the rest of the crumbs now because they are confusing in the source.

Barry



> On Aug 27, 2016, at 8:54 AM, Barry Smith <bsmith at mcs.anl.gov> wrote:
> 
> 
> #define PETSCVIEWERNETCDF 'netcdf'
> 
> #define PETSCVIEWERNETCDF       "netcdf"
> 
> PETSC_EXTERN PetscErrorCode PetscViewerNetcdfOpen(MPI_Comm,const char[],PetscFileMode,PetscViewer*);
> PETSC_EXTERN PetscErrorCode PetscViewerNetcdfGetID(PetscViewer, int *);
> 
>     <ul>
>        <li>Added new viewer_type PETSC_VIEWER_NETCDF, PETSC_VIEWER_HDF4</li>
>        <li>PETSc error handlers now trap many more signals.</li>
>      </ul>
> 
> 
> .  viewer - the PetscViewer; must be a binary, MATLAB, hdf, or netcdf PetscViewer
> 
> PETSC_EXTERN PetscErrorCode PetscViewerCreate_Netcdf(PetscViewer);
> 
> PETSC_INTERN PetscErrorCode VecView_MPI_Netcdf(Vec,PetscViewer);
> 
> But that is all. So someone must have removed Netcdf support be left a few crumbs around?
> 
>  Barry
>