I read here in the GNU Hurd docs that Mach ports are "globally unique." I know that one of the design goals for Mach was for it to be easily used in distributed computing environments, so does this mean Mach ports are globally unique across all Mach systems at any given point in time? This would mean that Mach ports can be used as references to objects that live on remote Mach systems, which is convenient. But how is this implemented?
Or does this statement mean Mach ports are only guaranteed to be globally unique on a single Mach system at any given point in time?