[mpiwg-ft] New draft that separates ERR_PENDING from ERR_PROC_FAILED_PENDING

[Aurélien Bouteiller]

Please review the draft that modifies the errors ERR_PENDING into ERR_PROC_FAILED_PENDING

The name of the error code is rather long. Shorter error name propositions are welcome. 
MPI_ERR_WOULDBLOCK ??? 

-------------- next part --------------
A non-text attachment was scrubbed...
Name: ft-newpending.pdf
Type: application/pdf
Size: 213558 bytes
Desc: not available
URL: <http://lists.mpi-forum.org/pipermail/mpiwg-ft/attachments/20140124/f0434b6e/attachment.pdf>
-------------- next part --------------


For references, and this was the only though worthy issue for this change, the document intentionally doesn?t specify what the implementation behavior must be for ERR_PENDING when ERR_PROC_FAILED_PENDING could be raised (or the contrary). 

MPI_Irecv(ANY_SOURCE, comm1, &r1) 
MPI_Irecv(ANY_SOURCE, comm1, &r2)
MPI_WaitAll(r1,r2) can legitimately result in 

err_status(r1): PROC_FAILED_PENDING
err_status(r2): PROC_FAILED_PENDING

err_status(r1): PENDING
err_status(r2): PROC_FAILED_PENDING

err_status(r1): PROC_FAILED_PENDING
err_status(r1): PENDING