[MOAB-dev] (compiler ?) errors with xlc

[Jason Kraftcheck]

acaceres at mcs.anl.gov wrote:
> sorry if you get this twice, not clear if it went through...
> 

Its already fixed (r3679).

Thanks,

- jason

> ---
> Hi,
> 
> Are both of these methods needed?
> 
> inline ErrorCode ParallelComm::get_owner(EntityHandle entity, int &owner)
> inline ErrorCode ParallelComm::get_owner(EntityHandle entity, EntityID &owner)
> 
> IBM's XL compiler can't seem to distinguish them, at least on one particular system, and gives the following error:
> 
> "moab/ParallelComm.hpp", line 288.13: 1540-0401 (S) The member "moab::ParallelComm::get_owner(EntityHandle, EntityID &)" is already declared.
> "moab/ParallelComm.hpp", line 287.13: 1540-0424 (I) "get_owner" is declared on line 287 of "moab/ParallelComm.hpp".
> 
> 
> Commenting out one of the methods (the int& one) seems to fix things.
> 
> 
> I also had to make the following change to get things to compile:
> --- src/Core.cpp        (revision 3670)
> +++ src/Core.cpp        (working copy)
> @@ -1197,8 +1197,8 @@
>                               Range& adj_entities )
>  {
>    std::vector<EntityHandle> results;
> -  ErrorCode rval = get_adjacencies_intersection( mb, begin, end, to_dimension,
> -                                                   create_if_missing, results );
> +  ErrorCode rval = moab::get_adjacencies_intersection( mb, begin, end, to_dimension,
> +                                                       create_if_missing, results );
>    if (MB_SUCCESS != rval)
>      return rval;
> 
> 
> 
> Finally, --enable-parallel doesn't imply the -DUSE_MPI switch, don't know if that's intentional.


-- 
"A foolish consistency is the hobgoblin of little minds" - Ralph Waldo Emerson