Open MPI logo

Hardware Locality Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Hardware Locality Development mailing list

Subject: Re: [hwloc-devel] [hwloc-svn] svn:hwloc r2339
From: Samuel Thibault (samuel.thibault_at_[hidden])
Date: 2010-08-19 09:06:08


Brice Goglin, le Sat 17 Jul 2010 12:34:02 +0200, a écrit :
> Le 17/07/2010 11:47, Samuel Thibault a écrit :
> >> +HWLOC_DECLSPEC int hwloc_get_membind(hwloc_topology_t topology, hwloc_cpuset_t set, int * policy);
> >>
> > Mmm, I wouldn't have thought of providing the places by giving a cpuset.
> > Why not using nodesets?
>
> If we really start using nodesets intensively, we shouldn't use
> hwloc_cpuset_t for their type, it's confusing. Maybe #define
> hwloc_nodeset_t hwloc_cpuset_t and add a big comment saying that
> nodesets may be manipulated with cpuset functions?

Actually, I'm realizing that I should have realized before with the
radset interface of various OSes that it may be better to just have a
hwloc_set_t type, and hwloc_cpuset_t/hwloc_nodeset_t types typedefed to
it, and tell people to use hwloc_set_* functions with both (but keep
#defines for hwloc_cpuset_* API compatibility for some time).

> Then, we'd need some conversion routines such as
> hwloc_cpuset_from/to_nodeset(topology, cpuset, nodeset).

This is done.

> And a big comment saying that we don't duplicate all helpers from
> nodesets and that people should convert to cpusets before using
> helpers.

Sure.

Samuel