[petsc-users] Why PETSc needs PetscScalar datatype?

[Satish Balay]

PETSc requires fortran preprocessing [or one could say C preprocessing
if preprocessing is not part of any fortran standard - but most
compilers support it] - and preprocessing is case sensitive.

Satish

On Sat, 4 Aug 2018, Youjun Hu wrote:

> Thank Satish.
> PetscFortranComplex works.
> I just want to distinguish between real and complex variables in my Fotran
> code, so I prefer to use PetscFortranComplex when a variables is complex.
> I began to learn petsc just several weeks ago.
> Another unusual thing I noticed is that petsc keywords are case-sensitive
> in Fotran codes, which breaks the convention that Fortran is
> case-insensitive.
> Correct me if I am wrong.
> 
> Youjun
> 
> On Sat, Aug 4, 2018 at 3:24 PM Satish Balay <balay at mcs.anl.gov> wrote:
> 
> > For some reason it is PetscFortranComplex [on fortran side]
> >
> > I'm not sure I understand your logic for using PetscComplex and not
> > PetscScalar.
> >
> > We recommend using PetscScalar. You can choose not to follow our
> > recommendation.
> >
> > But we don't say 'PetscScalar' is the wrong thing for users [as you
> > appear to infer]
> >
> > Satish
> >
> > On Sat, 4 Aug 2018, Youjun Hu wrote:
> >
> > > I just tried to change PetscScalar to the specific datatype PetscComplex
> > in
> > > my Fortran code, but the compiler raised the following error:
> > > PetscComplex   my_coeff
> > > Error: Unclassifiable statement at (1)
> > >
> > > It seems that the compiler does not know this datatype.
> > >
> > > My Petsc was configured with ./configure   --with-scalar-type=complex
> > > --with-precision=double
> > >
> > > Before the modification, the code was complied successfully.
> > > Could someone help?
> > >
> > > Thanks,
> > > Youjun
> > >
> > >
> >
> >
>