Below is the spec text capturing my proposed
resolution for Issue 218. Please let me know if you have any questions.
Proposed
Spec Text Changes to resolve Issue 218
In Section 12.6, after
paragraph 1:
Add: “The same isolation semantics apply to properties,
being merely projections of variables and thus always coupled with them; access
to properties is identical to access to variables, which is controlled by the
enclosing isolated scope.”
In Section 12.6, after
paragraph 3:
Add: “Any
partnerLinks declared within an isolated scope have their access protected by
that enclosing scope; the protection applies specifically to the endpointReference
part, and not the messageExchange parts, of the partnerLink state. Any
messageExchange declared in a scope serves only to provide a handle to access a
facet of the state of its associated partnerLink; it is intrinsically
stateless. The control afforded by the enclosing isolated scope does not apply
to messageExchange handles. Rather, partnerLink states are protected based on
the endpointReference.”
“By definition,
correlation sets are only mutable at initialization; they are immutable
throughout the remainder of their lifecycle. Any correlation sets declared
within an isolated scope do not have their access controlled by the enclosing
scope. However, the initialization of a correlation set declared within an
isolated scope is performed in an atomic fashion – in the same sense as
that of an <assign> operation – ensuring that the correlation set
will not be initialized partially or wholly by multiple concurrent inbound
message activities.”
Best,
-Charlton.
--
|
