[mpich-discuss] Fwd: MPI_T_ERR_ ... error codes for FORTRAN
Junchao Zhang
jczhang at mcs.anl.gov
Tue Sep 30 10:56:46 CDT 2014
Hi, Mike,
From MPI-3.0 A.1.1., I can not see it requires all constants are valid in
C and FORTRAN. Instead, in 14.3, it says "Since the MPI tool information
interface primarily focuses on tools and support libraries, MPI
implementations are only required to provide C bindings for functions
introduced in this section." And in MPICH's f08 bindings, it is the same
(i.e. no support of MPI-T).
--Junchao Zhang
On Tue, Sep 30, 2014 at 10:39 AM, Michael L. Stokes <Michael.Stokes at uah.edu>
wrote:
> On 09/30/2014 10:36 AM, discuss-owner at mpich.org wrote:
>
>> According to the MPI-3.0 spec (A.1.1), C and FORTRAN constants are
>> supposed to be shared (with noted exceptions).
>> These include the MPI_T_ERR_ ... error codes. But from what I can
>> determine, none of the MPI_T_ERR_ error codes
>> seem to compile under the mpif77 compiler (intel) using MPICH 3.1.2. So I
>> post the following questions:
>>
>> 1) Am i interpreting the specs correctly in that all constants (error
>> codes in this instance) are valid in C and FORTRAN?
>>
>> 2) Is the behavior I'm seeing correct given the current intel compiler
>> and version of MPICH? The snippet of code I'm
>> using is the following. Here I'm using a perl script hereto document to
>> write FORTRAN code where $target is any
>> of the MPI error codes, and $ext is 'f' in this case.
>>
>> PROGRAM ERRORCODE
>> IMPLICIT NONE
>> INCLUDE "mpif.h"
>>
>> IF( $target.ge.MPI_SUCCESS .and.
>> + $target.le.MPI_ERR_LASTCODE) THEN
>> WRITE(6,1)
>> 1 FORMAT('$ext "$target"
>> +is verified ')
>> ELSE
>> WRITE(6,2)
>> 2 FORMAT('$ext "$target"
>> +is not verified ')
>> ENDIF
>>
>> CALL EXIT(0)
>> END PROGRAM ERRORCODE
>>
>> 3) Will this behavior change under the f08 FORTRAN bindings?
>>
>> Thanks
>> Mike
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mpich.org/pipermail/discuss/attachments/20140930/2e39c916/attachment.html>
-------------- next part --------------
_______________________________________________
discuss mailing list discuss at mpich.org
To manage subscription options or unsubscribe:
https://lists.mpich.org/mailman/listinfo/discuss
More information about the discuss
mailing list