[mpich-discuss] MPICH2 v.1.2.1 build error; -Switch case constants must be unique

[Dave Goodell]

Hi Ray,

It sounds like you have found a bug in our optional type handling  
code, specifically with respect to the new C99 complex types that were  
added for MPI-2.2 support.  What version of the PGI compilers are you  
using?

In order to figure out what's really going on here, I'll need to take  
a look at your config.log.  Would you please file a bug report for  
this at:

http://trac.mcs.anl.gov/projects/mpich2/newticket

and attach your config.log to the new ticket?

In the very short term, you should hopefully be able to get around  
this problem by deleting lines 200-211 from src/mpi/coll/oputil.h  
(reproduced below for clarity).

Thanks,
-Dave

--------8<--------
    200	#if defined(HAVE_FLOAT__COMPLEX)
    201	#  undef MPIR_OP_TYPE_MACRO_HAVE_C_FLOAT_COMPLEX
    202	#  define  
MPIR_OP_TYPE_MACRO_HAVE_C_FLOAT_COMPLEX(mpi_type_,c_type_)  
MPIR_OP_C_COMPLEX_TYPE_MACRO(mpi_type_,c_type_)
    203	#endif
    204	#if defined(HAVE_DOUBLE__COMPLEX)
    205	#  undef MPIR_OP_TYPE_MACRO_HAVE_C_DOUBLE_COMPLEX
    206	#  define  
MPIR_OP_TYPE_MACRO_HAVE_C_DOUBLE_COMPLEX(mpi_type_,c_type_)  
MPIR_OP_C_COMPLEX_TYPE_MACRO(mpi_type_,c_type_)
    207	#endif
    208	#if defined(HAVE_LONG_DOUBLE__COMPLEX)
    209	#  undef MPIR_OP_TYPE_MACRO_HAVE_C_LONG_DOUBLE_COMPLEX
    210	#  define  
MPIR_OP_TYPE_MACRO_HAVE_C_LONG_DOUBLE_COMPLEX(mpi_type_,c_type_)  
MPIR_OP_C_COMPLEX_TYPE_MACRO(mpi_type_,c_type_)
    211	#endif
--------8<--------

On Dec 11, 2009, at 3:09 PM, Ray Sheppard wrote:

> Hi,
>  I have run into some trouble.  Orginally, I hit this error using  
> pgf90
> and pgcc to build MPICH on an Intel based cluster:
>
> PGC-S-0083-Switch case constants must be unique (opsum.c: 57)
> PGC-S-0083-Switch case constants must be unique (opsum.c: 94)
>
> This turned out to be using: MPIR_OP_TYPE_GROUP() for both (COMPLEX)  
> and
> (COMPLEX_EXTRA).  So, I just thought I would comment out the
> (COMPLEX_EXTRA) line and see what happens. What happened is that the
> package stumbles again in opprod.c.  I have not yet found an
> explanation or a fix in your documentation.  So, I am writing to ask
> how I can make it work.  Thanks.
>                                        Ray
> -- 
>                         Respectfully,
>                           Ray Sheppard
>                           RSheppar at indiana.edu
>                           http://rac.uits.iu.edu/hpc/
>                           317-274-0016
>
>                   Principal Analyst
>                   High Performance Applications
>                   Research Technologies
>                   University Information Technological Services
>                   IUPUI campus
>                   Indiana University
>
>         My "pithy" saying since everyone has one these days:
>             5 tons phytoplankton = 1 pound fish.
>             -- Please help protect the environment.
>
> _______________________________________________
> mpich-discuss mailing list
> mpich-discuss at mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss