RE: [bt-spec] BTP Issue 68 : (sub)coordinators,(sub)composers with no inferiors

[Peter Furniss <peter.furniss@choreology.com>]

 

BTP Issue 68 : (sub)coordinators, (sub)composers with no inferiors

Category: minor technical
Submitter: Pyounguk Cho(Pyounguk.Cho@iona.com)
Detail: Is it possible to have (sub)coordinators or (sub)composers as inferiors when they have no inferiors?
Proposed Remedy:
(sub)coordinators and (sub)composers should be delisted automatically when all of their inferiors have resigned.

---

  Assuming the btp interoperable control protocol is used to create the sub-coordinator, sub-composer, then they will be initially be things enrolled with the superior (making them sub-*) but of course no inferiors. They won't get any inferiors until they get an enrol.  It is already defined whether they are atomic or cohesive (at BEGIN time).

 

If we aren't using the interoperable control protocol, all that is externally visible is the enrollment of the new actor as Inferior to the existing superior, followed, at some point, by the presence of the new CONTEXT (marked as atomic or cohesive) in relation to some outbound application message.  In fact, it is impossible by external observation (i.e. line tapping) alone to determine that this new CONTEXT is from a sub-co* .

 

If all the inferiors resign during the active phase, then we definitely hang around as there may be new enrollments. If all inferiors return RESIGN in response to PREPARE (or when it is known by other means that there will be no more enrollments), then the sub-co* should itself issue RESIGN to its superior.  (That could be regarded as how "delisted automatically" is implemented in the protocol perhaps.)

 

Peter