Perfect; fixed in: https://svn.open-mpi.org/trac/ompi/changeset/26971 https://svn.open-mpi.org/trac/ompi/changeset/26972
It'll be in the 1.6 nightly tarballs starting tonight. Many thanks for the patch. On Aug 8, 2012, at 9:36 AM, ESCOBAR Juan wrote: > On 08/08/2012 03:32 PM, Jeff Squyres wrote: > >> Ah -- interesting. Thank you for the patch! I'm unfamiliar with this >> Fortran style of splitting strings. Would it be better to split this up >> into multiple print * statements? >> >> You sent: >> >> print *, "Open MPI WARNING: You are calling MPI_SCATTERV with incorrect >> sendcounts and displs arguments! & >> Your code may crash or produce incorrect results. & >> ***Your code will fail to compile in future versions of Open >> MPI*** & >> because this old/incorrect Fortran subroutine binding will be >> removed. & >> Please update the type of your sendcounts and displs parameters >> & >> to make this warning go away (and have correct code!)." >> >> Should we instead do something like this: > > :-) It's OK to . >> >> print *, "Open MPI WARNING: You are calling MPI_SCATTERV with incorrect >> sendcounts and displs arguments!" >> print *, "Open MPI WARNING: Your code may crash or produce incorrect >> results." >> print *, "Open MPI WARNING: ***Your code will fail to compile in future >> versions of Open MPI***" >> print *, "Open MPI WARNING: because this old/incorrect Fortran subroutine >> binding will be removed." >> print *, "Open MPI WARNING: Please update the type of your sendcounts and >> dispels parameters" >> print *, "Open MPI WARNING: to make this warning go away (and have correct >> code!)." >> >> This makes the warning a bit more intrusive (since it's multiple lines), but >> that's good. :-) >> >> >> On Aug 8, 2012, at 9:19 AM, ESCOBAR Juan wrote: >> >>> Hello . >>> >>> The 'warning print' statement introduced in the correction made since >>> version 1.5.4 and 1.6.X on this script : >>> openmpi-1.6.1rc2/ompi/mpi/f90/scripts/mpi_scatterv_f90.f90.sh : >>> >>>> print *, "Open MPI WARNING: You are calling MPI_SCATTERV with incorrect >>>> sendcounts and displs arguments! >>> >>> generate a compilation error when compiling with PGI compiler ( last >>> version 12.6 tested ) >>> because the 'print' line if far to long for a 'standard/normed ' fortran >>> compiler = 132 characters max by lines . >>> >>> => splitting the 'print' line with "&" solve the problem . >>> >>> Find the suggested script ( version 1.6.1rc2 ) in the mail >>> >>> A+ >>> >>> Juan >>> >>> -- >>> Juan ESCOBAR >>> Support MESONH >>> Laboratoire d'Aérologie, O.M.P. >>> 14 avenue Edouard Belin >>> 31400 - TOULOUSE >>> Tel : +33 (0)5-61-33-27-49 >>> Fax : +33 (0)5-61-33-27-90 >>> >>> <mpi_scatterv_f90.f90.sh>_______________________________________________ >>> users mailing list >>> us...@open-mpi.org >>> http://www.open-mpi.org/mailman/listinfo.cgi/users >> > > > -- > Juan ESCOBAR > Support MESONH > Laboratoire d'Aérologie, O.M.P. > 14 avenue Edouard Belin > 31400 - TOULOUSE > Tel : +33 (0)5-61-33-27-49 > Fax : +33 (0)5-61-33-27-90 > > _______________________________________________ > users mailing list > us...@open-mpi.org > http://www.open-mpi.org/mailman/listinfo.cgi/users -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
