if you want the spec to be implementable, you should make this explicit
as subbu points out. we had considerable confusion over this. the
statement in section 8 is not obvious nor is it clear.
Rich Thompson wrote:
The statement "Any Producer that
supports
cloning portlets on performBlockingInteraction() invocations
MUST
support the releasePortlet() operation" from section 8 was
meant to capture this requirement. We have already noted an errata item
for this occurrence missing the rename of releasePortlet() from early
drafts
to destroyPortlets() in the final version.
Rich
Following up on Subbu's original response. The behavior still seems
inconsistent to me because destroyPortlets is in the optional management
interface... Subbu says it in his "c." but then seems to recant
later on by
saying "it's not an error" (see below)
My point is this: if implicit cloning is allowed, then the management
interface MUST be supported... The spec. seems to stop short of saying
that. Or maybe I just missed it :(
Subbu wrote:
a. POPs are offered by the producer, while CCPs are created subsequently
either due to an implicit cloning during performBlockingInteraction or
an explicit cloning via the portlet management interface.
c. CCPs cannot exist if the Producer does not offer the portlet
management interface.
Yes, it is not an error for the consumer to send cloneBeforeWrite even
when the Producer does not offer the portlet management interface.
To unsubscribe from this mailing list (and be removed from the roster
of
the OASIS TC), go to
http://www.oasis-open.org/apps/org/workgroup/wsrp/members/leave_workgroup.php.
|