Re: [ws-tx-implement] BA scenario 12

[Peter Furniss <peter@furniss.co.uk>]

Having one message causing creation of two participants would be a good test of MixedOutcome (although it can be done with the PA1, PA2 mechanism as I sent in an earlier message, assuming the contexts sent to PA1, PA2 are identical).  With one PA with PS1, PS2 there should be no sequencing assumption between the registrations (as indeed there would not be if we had PA1, PA2).

It would of course be a failure of the test if PS1 gets told to compensate, PS2 gets told to close.

I do not see how this scenario can be achieved without reconsidering issue 070.

Peter




Doug Davis wrote:

BA Scenario 12 shows the IA sendsing an application message to two participants.  This is a change from all other tests where there is just one message sent to a single PA.  If the test needs to simulate more than one participant then the "single PA" should register twice.  I would like to suggest that we change BA1.12 to match what we've done in previous test (see AT 3.1 for an example).
-Doug