[petsc-dev] Preprocessor hell: #define VecType

[Matthew Knepley]

On Fri, Sep 28, 2012 at 12:25 PM, Karl Rupp <rupp at mcs.anl.gov> wrote:

> Hi guys,
>
> a long time ago, I stumbled upon one thing that caused a lot of headaches
> when interfacing my code with PETSc. Stripping away all the details, my
> solver interface looked about like this:
>
> template <typename MatrixType, typename VecType>
> VecType solve(MatrixType const & A, VecType const & b);
>
> Everything went well if PETSc was not in use. However, I got really weird
> compiler errors as soon as I've included petscksp.h.
>
> Now, fast-forward to today. While browsing through the existing
> CUDA-enabled implementations, I found the following in src/ksp/pc/impls/
> sacusp.cu:
>   #undef VecType
>   #include <cusp/precond/smoothed_**aggregation.h>
>   #define VecType char*
> While this preprocessor define is inside an implementation file, not in a
> header file, one may consider that acceptable.
>
> However, a quick grep on the main include/ folder reveals
>  petscvec.h:#define VecType char*
> which is ultimately the reason why my template code went nuts. While I do
> understand that name clashes can appear, I tried to find the deeper meaning
> of VecType. The manual lists only one reference on page 170:
>  VecLoad(PetscViewer viewer,VecType outtype,Vec *newvec);
> and the associated hyperlink
>
> http://www.mcs.anl.gov/petsc/**petsc-dev/docs/manualpages/**
> Vec/VecLoad.html#VecLoad<http://www.mcs.anl.gov/petsc/petsc-dev/docs/manualpages/Vec/VecLoad.html#VecLoad>
> points to an HTML page defining the prototype
>  #include "petscvec.h"
>  PetscErrorCode  VecLoad(Vec newvec, PetscViewer viewer)
>
> This suggests that VecType is not in use here, and it may not be in
> serious use at all. I can find about five uses of VecType in the whole
> include/-folder, so it shouldn't be too hard to get rid of that (e.g. by
> renaming to PetscVecType or any other more descriptive name), should it? Do
> I miss something?
>

You are missing the fact that all PetscObjects have an implementation type
and it is a string, and there is
one of these #defines for every class.

However, I would support namespacing all these Types since they are not in
heavy use, e.g. PetscVecType, PetscMatType.

   Matt


> Best regards,
> Karli
>
>
> PS: For reference, the following code compiles and prints "Hello World!":
>
> #include <iostream>
> #include "petscksp.h"
>
> int main(int argc, char **args)
> {
>   VecType test = "Hello world!";
>
>   std::cout << test << std::endl;
>   return EXIT_SUCCESS;
> }
>
> To eliminate any warnings about deprecated conversion of strings, just add
> 'const' in front of 'VecType'.
>



-- 
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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20120928/caccc179/attachment.html>