extended issues homework

["Mark Little" <mark.little@arjuna.com>]

Kids attract ailments like magnets! My 10 year old son has come down with
tonsillitis so I'm not sure how long I'll be able to stay on todays telecon.
(Luckily ou won't need me to be present in order to keep the line active.)
In case there isn't time for me, I did look at the deferred issues that Jim
and I had raised prior to the 1.0 vote, and here's my conclusion:

Issue 102: Is entity identity modification to be allowed within BTP.

I'm no longer sure that this adds sufficient utility for the added
complexity it will involve. So I'm happy to withdraw the issue.

Issue 103: CONTEXT used only with application msgs, not with BEGIN, BEGUN

I still agree with the initial assessment that Jim made when registering
this issue: it makes context use throughout the model more consistent,i.e.,
CONTEXT only ever appears as an application-level message. However, it would
cause problems for interoperability with existing implementations.

Issue 105: Qualifier to allow automatic completion of all-cancelled
cohesions

The intent here is to allow a cohesion to be terminated as soon as possible
to release any coordinator-related resources. The example given is when an
application knows that all participants have been registered with a cohesion
prior to invoking prepare_inferiors. If that information isn't available to
the coordinator, then even if all inferiors cancel or indicate "read-only"
semantics, the cohesion coordinator must continue to exist until an explicit
cancel_transaction message is sent. Adding a qualifier that basically says
"terminate if your participant set becomes zero for the next round" seems
like a nice optimization to me.

Mark.