[petsc-dev] plexexodusii.c incompatible with complex - not covered?

Matthew Knepley knepley at gmail.com
Fri Apr 13 15:58:03 CDT 2018


On Thu, Apr 12, 2018 at 12:26 PM, Vaclav Hapla <vaclav.hapla at erdw.ethz.ch>
wrote:

> At master, configured with
>   --download-exodusii --download-hdf5 --download-netcdf --download-pnetcdf
> --with-scalar-type=complex
> I get bunch of compiler warnings
>   warning: passing argument 2 of ‘VecGetArrayRead’ from incompatible
> pointer type
>   warning: passing argument 2 of ‘VecRestoreArrayRead’ from incompatible
> pointer type
> in
>   src/dm/impls/plex/plexexodusii.c.
>
> Is this configuration not covered by nightly tests?
>
>
> I want to fix it anyway. Do you agree with the following scheme?
>

I think we can do it inline. Can you send the compile error?

  Thanks,

    Matt


>   Vec v;
>   const PetscScalar *arr;
>   PetscReal *arr_real;
> #if defined(PETSC_USE_COMPLEX)
>   PetscInt i;
> #endif
>
> ...
>   ierr = VecGetArrayRead(v, &arr);CHKERRQ(ierr);
> #if defined(PETSC_USE_COMPLEX)
>   ierr = PetscMalloc1(size, &arr_real);CHKERRQ(ierr);
>   for (i = 0; i < size; ++i) {
>     arr_real[i] = PetscRealPart(arr[i]);
> #if defined(PETSC_USE_DEBUG)
>     if (PetscImaginaryPart(arr[i])) {
>       SETERRQ(PETSC_COMM_SELF, PETSC_ERR_SUP, "Vector contains complex
> numbers but only real vectors are currently supported.");
>     }
> #endif
>   }
> #else
>   arr_real = (PetscReal*)arr;
> #endif
>
> ...
>   ierr = VecRestoreArrayRead(v, &arr);CHKERRQ(ierr);
> #if defined(PETSC_USE_COMPLEX)
>   ierr = PetscFree(arr_real);CHKERRQ(ierr);
> #endif
>
>
> Thanks,
> Vaclav




-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20180413/e00cc2c8/attachment.html>


More information about the petsc-dev mailing list