[Sis-ams] a design question for us to think about
Scott Burleigh
Scott.Burleigh at jpl.nasa.gov
Fri Feb 9 19:59:38 EST 2007
Krupiarz, Christopher wrote:
> Scott,
>
> Thanks that helps. Are messages that are published instead of announced
> handled in a different fashion between continuia?
Yes, slightly, but this difference is invisible to the recipient except
for this phenomenon that we're discussing here -- that a node can
receive remotely-announced messages that it would not receive if those
messages were announced locally. A node never receives published
messages that it didn't subscribe to, regardless of whether the messages
are published locally or remotely.
> Are published
> messages not put in the envelope data structure?
Yes, they are. But the RAMS gateway knows which nodes in the local
continuum subscribed to which published messages, and it only forwards
those messages (via the envelope structure) to the nodes that subscribed
to them.
> I guess I'm confused
> about the differences of how messages which arrive from a remote
> continuum to a node depended on how they are sent (either published or
> announced).
>
It is kind of confusing. The spec tries to explain the differences as
clearly and unambiguously as possible, but nonetheless this stuff not at
all obvious. The Remote AMS section is easily the densest and most
difficult part of the book, because what we're trying to do is
surprisingly difficult to do in a general and efficient manner. It
would have been a lot simpler if we didn't need to support remote-Send
and remote-Announce in addition to remote-Publish. But I think we've
got it worked out now and I believe the investment will prove to be
worthwhile.
Scott
More information about the Sis-ams
mailing list