On Jan 5, 2011, at 11:21 AM, Samuel Thibault wrote: >> Maybe return EINVAL when passing REPLICATE to alloc_membind? > > Well, if we go that way we should add tests in mostly all the src/bind.c > functions, then.
I just said in a prior email that I didn't have much of an opinion here, but this corner case might convince me otherwise: it's definitely *erroneous* to pass REPLICATE to alloc_membind, so it's worth reporting the error. If that means we report errors in the other functions when non-expected flags are passed for the sake of consistency, then so be it. I have a bunch of edits to documentation comments hwloc.h; please don't edit until I commit (still need some further clarification before I can commit...). -- Jeff Squyres jsquy...@cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/
