Example

A PARC based application will look something like this:

Summary:

- InputAdapter : This interface is responsible for producing Records to be processed by the Pipeline. Typically it will extract records from a data store, instantiate the corresponding Record type and pass the resulting object into its target Channel.



- Channel : A thread safe abstraction for passing Records between Parc components.



- Filter : Filters store the business logic for a certain step in the pipeline process. Data validations and transformations are typically accomplished by a Filter, but they may be used for any type of processing.



- OutputAdapter : This interface is the counter-part to the InputAdapter. It is responsible for taking errored records and valid records and persisting them to their respective data stores. Typically errors go into an error log (db error table, log file, other) and the valid records are sent to the destination system.



- ExecutionStrategy : The ExecutionStrategy wraps the Pipeline and can be used to run the system in different ways. For example, a ConditionalLoopStrategy can be used to run the pipeline over and over until the InputAdapter finds no more work. Or a SingleCallStrategy may be used to run the pipeline only once. The ExecutionStrategy adds a additional layer of flexibility around the process.



- Record : The Record interface is implemented by objects passed through the system. It's mostly a marker interface but requires a few error related methods. An abstract class is provided that implements those in a basic, but reasonable way.



- Pipeline : The Pipeline manages the filter processing section of the system. It creates all the Filter threads and notifies the Strategy when all records have been pushed to the last Channel.