[bpel4people] NEW ISSUE: Title: Declaration of imported LPG not mentioned in B4P spec

["Mark Ford" <mark.ford@activevos.com>]

TARGET: B4P, Section 3.2.1 Using Logical People Groups

DESCRIPTION: Constellation 3 provides the ability to import a human
interaction into a bpel process. In this scenario, the imported human
interaction may contain one or more logicalPeopleGroup definitions. Does the
import of a human interaction constitute an implicit declaration of a
logicalPeopleGroup within the bpel process or is an explicit declaration
required? Some type of declaration is necessary since these LPG's are
stateful. Consider the following example consisting of a standalone human
interaction file imported into a bpel process.

<htd:humanInteractions targetNamespace="urn:example" ...>
   ...
   <htd:logicalPeopleGroups>
      <htd:logicalPeopleGroup name="LPG-1">
         ...
      </htd:logicalPeopleGroup>
   </htd:logicalPeopleGroups>
   <htd:tasks>
      <htd:task name="T-1">
         ...
         <htd:potentialOwners>
            <htd:from logicalPeopleGroup="LPG-1">
               ...
            </htd:from>
         </htd:potentialOwners>
         ...
      </htd:task>
      ...
    </htd:tasks>
</htd:humanInteractions>

<bpel:process xmlns:ex="urn:example" ...>
   ...
   <bpel:import location="example.htd" namespace="urn:example" .../>
   ...

   <b4p:humanInteractions>
      <htd:logicalPeopleGroups>
         <htd:logicalPeopleGroup name="LPG-1" reference="ex:LPG-1"/>      
      </htd:logicalPeopleGroups>
   </b4p:humanInteractions>

   ...

   <b4p:peopleActivity ...>
      <b4p:localTask reference="ex:T-1"/>
   </b4p:peopleActivity>

   ...

</bpel:process>

The declaration seems redundant but in the case where the process is only
utilizing a subset of the imported logicalPeopleGroups it better conveys to
the deployer which logicalPeopleGroups need to be bound to people queries.

PROPOSAL: none