compile SMPI as C++ source Agreed, that's not real C++ code. We would need to objectify everything, and kill our XBT containers, at the very least. But this will allow to further objectify the kernel. I'll deal with the user interfaces later.
sed -i -e 's/\t/ /g' [sources] Please people, stop using tabs
Make all predefined MPI_Datatype const We encountered one MPI application that was using them as such and thus failing to build with SMPI.
Update copyright notices Only files with an existing SimGrid copyright header and that were already changed in 2015 were modified.
Use xbt_strdup instead of strdup
Added xbt_assert statement to smpi_mpi_dt.c * This will help prevent a value <= 0
Fixed bug in SMPI datatypes, causing SG to segfault. - For pre-defined datatypes such as MPI_INT, the 'in_use' counter of that datatype was able to obtain negative values in certain cases. This caused SimGrid to crash with a segfault, for instance when running fupermod. This bug was reported by Tania Malik on the 24th of February on the SG Mailing list.
Simplify use of dict for smpi attr handling. Switch from char keys to direct int keys, as in MPI, because we can, actually (with dict_ext functions)
Add preliminary support for MPI_Pack, MPI_Pack_size, and MPI_Unpack. Activate tests for these functions. Somes cases with too complex imbricated datatypes are not supported (yet).
somehow we forgot MPI_REAL datatype C-fortran compatibility
Fix bug in keyval delete
give correct pointer when calling delete function
duplicated lines
store the int in a long-enough char array
cleanup, remove some useless functions
Add MPI_Type*keyval and MPI_Type*attr functions
Change way replay is handled, to allow cohabitation between replay and "classic" SMPI
resolve conflict
[smpi] Use SMPI namespace for some public variables/functions
Avoid tautological-compare Clang complains about tautological compare: it seems an unsigned integer is always non-negative.