Sorry I'm having a hard joining this meeting (conflict with two other meetings). Regarding #520: while implementing our PoC, we found that we probably need a new combiner type (see 5.1.13, MPI_TYPE_GET_ENVELOPE). The proposed pair types are neither named nor any of the general datatypes (they are special in that they are allowed for min/maxloc). I actually got the automatic generation of the min/maxloc code done but I am somewhat stuck in the nitty-griddy of OMPI's internal datatype handling. I will have to take another look at that at some point before the forum meeting. And as I said on the ticket, not having a guarantee that they resulting pair type can actually be used with min/maxloc makes this interface useless, IMHO. If we don't want to go as far as allowing arbitrary non-complex types for the index, I am fine with allowing a subset of index types. Cheers Joseph On 3/10/22 10:13, Marc-André Hermanns via mpiwg-tools wrote:

Hi all,

This is a reminder of our regularly scheduled call today:

    March 10, 2022 at 8 am Pacific / 11 am Eastern / 5pm CEST.

Meeting Details (private repo): https://github.com/mpiwg-tools/mpi-standard/wiki/Meeting-Information

**NOTE**: I created a new Zoom room for 2022. The info is updated on the wiki page referenced above. If you need access, let me know.

Agenda: ------- - Tickets   - https://github.com/mpi-forum/mpi-issues/issues/542   - https://github.com/mpi-forum/mpi-issues/issues/520   - further tickets (if time permits)

Cheers, Marc-Andre

_______________________________________________ mpiwg-tools mailing list [email protected] https://lists.mpi-forum.org/mailman/listinfo/mpiwg-tools