UML 2 Activity and Action Models, Part 3: Control Nodes

This is the third in a series introducing the activity model in the Unified Modeling Language, version 2 (UML 2), and how it integrates with the action model [1]. The previous article addressed the execution characteristics of actions in general, and additional functionality of actions that invoke behaviors [2]. The first article gave an overview of activities and actions that is assumed here [3]. The remainder of the series elaborates other specific elements. This article covers control nodes, which route control and data through the flow model. It also points out the differences in concurrency support between UML 2 and UML 1.x activities. To recap, UML 2 activities contain nodes connected by edges to form a complete flow graph. Control and data values flow along the edges and are operated on by the nodes, routed to other nodes, or stored temporarily. More specifically, action nodes operate on control and data they receive via edges of the graph, and provide control and data to other actions; control nodes route control and data through the graph; and object nodes hold data temporarily as they wait to move through the graph. Data and object are unified in UML under the notion of classifier, so they are used interchangeably. The term "token" is shorthand for control and data values that flow through an activity. Decision and Merge Fork and Join Final nodes Initial node Activity final Flow final Decision and Merge Fork and Join Final nodes Initial node Activity final Flow final