I’m not sure you are correct in stating that the simplest case “obviously” still needs to work. In fact, there were several proposals on the last telecon to the contrary as it was unclear that the community would be able to agree on defaults. So people suggested either requiring configuring with a default, or using MCA params for a sys admin to set the defaults and exiting with error if they aren’t provided.
So I guess I’m asking: before expending a lot of effort worrying about syntax, do we actually have agreement that users can run without any additional stuff? With what defaults, and how will those defaults be made clear to them? If not, then I have some concerned opinions on what you are proposing. If yes, then I’m less concerned as the 10% who care are advanced users and willing to figure it out. > On Oct 21, 2015, at 11:07 AM, Jeff Squyres (jsquyres) <jsquy...@cisco.com> > wrote: > > On Oct 21, 2015, at 12:05 PM, Ralph Castain <r...@open-mpi.org> wrote: >> >> It seemed like this topic was straying, so I’m glad to hear that specifying >> nothing means we still execute. > > Yes, I'm not trying to change the simple/easiest case. That obviously still > needs to work. > >> My question remains, though: what is the default? What are the default >> values of the networks/qualifiers? > > This proposal is not addressing that issue. Today, I think we (developers / > OMPI experts) mostly understand what it does. And for most users, I think > OMPI does the Right Thing. When UCX comes in, it will get a bit muddier (but > this is exacerbating an existing problem -- not causing it). We might want > to dovetail a conversation about the defaults on to this proposal, but it's > still (logically) a separate -- albeit related -- issue. > > -- > Jeff Squyres > jsquy...@cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > > _______________________________________________ > devel mailing list > de...@open-mpi.org > Subscription: http://www.open-mpi.org/mailman/listinfo.cgi/devel > Link to this post: > http://www.open-mpi.org/community/lists/devel/2015/10/18236.php
