Thanks Josh.
george.
On Mar 22, 2012, at 10:09 , Josh Hursey wrote:
> Should be fixed in r26177.
>
> On Thu, Mar 22, 2012 at 7:51 AM, Josh Hursey wrote:
>> Fair enough. Upon further inspection of the request_invoke() handler,
>> you are correct that it is not required here if we do not modif
Should be fixed in r26177.
On Thu, Mar 22, 2012 at 7:51 AM, Josh Hursey wrote:
> Fair enough. Upon further inspection of the request_invoke() handler,
> you are correct that it is not required here if we do not modify the
> default value for req_status.MPI_ERROR.
>
> I'll work on a revised patch
Fair enough. Upon further inspection of the request_invoke() handler,
you are correct that it is not required here if we do not modify the
default value for req_status.MPI_ERROR.
I'll work on a revised patch this morning and commit. One that does
not use this field.
Per your comment from your fir
On Mar 21, 2012, at 15:13 , Josh Hursey wrote:
> I see your point about setting MPI_ERR_PENDING on the internal status
> versus the status returned by MPI_Waitall. As I mentioned, the reason
> I choose to do that is to support the ompi_errhandler_request_invoke()
> function. I could not think of
I see your point about setting MPI_ERR_PENDING on the internal status
versus the status returned by MPI_Waitall. As I mentioned, the reason
I choose to do that is to support the ompi_errhandler_request_invoke()
function. I could not think of an better way to fix this, so I'm open
to ideas.
MPI_Wai
My point was that MPI_ERR_PENDING should never be set on a specific request.
MPI_ERR_PENDING should only be returned in the array of statuses attached to
MPI_Waitall. Thus, there is no need to remove it from any request.
In addition, there is another reason why this is unnecessary (and I was too
In the patch for errhandler_invoke.c, you can see that we need to
check for MPI_ERR_PENDING to make sure that we do not free the request
when we are trying to decide if we should invoke the error handler. So
setting the internal req->req_status.MPI_ERROR to MPI_ERR_PENDING made
it possible to check
Josh,
I don't agree that these changes are required. In the current standard (2.2),
MPI_ERR_PENDING is only allowed to be returned by MPI_WAITALL, in some very
specific conditions. Here is the snippet from the MPI standard clarifying this
behavior.
> When one or more of the communications comp
