IBM WebSphere Adapters User Manual page 84

CCI clients of resource adapters that support batch results must be capable of
recognizing a top-level container and iterating through the child objects that
represent the results of the query. The client can then extract any individual
business object in the container and deliver it to the rest of the system as with any
single business object. This obviously requires the creation of additional business
object container structure definitions by the user or during metadata import for
each business object type that the user intends to query in batch. The business
object container is shown in the illustration.
Customer container BO
Customer
Business Object
Address
Business Object
Customer
Business Object
Address
Business Object
Business object container
This approach is favored over that of the JDBC-style ResultSet support as described
in the JCA specification. However, if an EIS provides native ResultSet support or if
it makes sense then adapter developers are encouraged to implement the CCI
ResultSet interfaces to provide customers with a high-performance alternative to
the generic batch retrieval approach described in this document.
The container object is produced by the data binding for use in the Websphere
Process Server or Websphere Enterprise Service Bus environment. Internally, the
adapter will model the multiple retrieved records as a list of top-level records that
can be iterated over using the getNext() method of the output Record.
Note: The use of the operation name RetrieveAll rather than RetrieveByContent
(as used by the WebSphere Business Integration adapters) distinguishes this
operation as a new and clear standard. Support for RetrieveByContent was
inconsistent across previous adapters: some would retrieve a single object and
return special error codes if there were more objects that matched while other
adapters would create special containers to return multiple values.
78
WebSphere Adapters: WebSphere Adapter Toolkit User Guide