Sequence Diagram (SD, etd)

Purpose

The sequence diagram (SD) specifies the time and control aspects of a system. Typically, you use the SD to analyze only the more complex business events. Simple business events rarely require an SD.

Compared with a collaboration diagram

A collaboration diagram shows the objects and relationships involved in an interaction. It shows the messages exchanged among the objects during the interaction, numbering them to show the sequence of messages. It does not show the timing of the messages.

File type

In most cases, a diagram's file type matches its abbreviation. This is not true for the SD, which has a file type of etd. In OMT, the SD is an event trace diagram (ETD).

Definitions

Events are actions between the objects in your project. They can also transmit data.

A scenario is the sequence of events during one execution of a program. A scenario can include all the events or only the events sent to or received by certain objects in the system. An SD can contain more than one scenario. For example, you might create an SD that provides two scenarios for placing a customer order: one for an existing customer and the other for a new customer.

Example SD

This SD shows a simple scenario for the use of an automatic teller machine. In this scenario, the system rejects the card. For the customer this transaction is simple: he inserts the card and the card is rejected. For the system, several actions are necessary. A constraint specifies 10 seconds as the maximum acceptable time interval for the transaction.

This sequence is fairly general; you can work out the Insert Card event more completely by including details such as entering the personal identification number and requesting the type of transaction. The analyst must decide what level of detail to use in the SDs.

Control panel

The control panel for the sequence diagram.

Initiators and Objects

Initiator

The initiator starts the sequence. Each sequence can only have one initiator, but you can draw more than one sequence in a diagram. The initiator represents an external agent, such as a user, that interacts with the system. It can be a class, but does not have to be.

Label

name

Object

An object is an instance of a class. The object line represents a portion of the lifetime of an object in the system. Time flows from top to bottom on the symbol. The spacing between event symbols is not important. It does not indicate the length of time between events.

Label

name[:type]

The type must be a class defined in a class diagram. If you omit type, the UML suite assumes the class name is the same as the object name.

Active object

An active object, also known as a concurrent object, owns a thread of control and can initiate control activity. For example, processes and tasks are active objects.

Label

name[:type]

The type of an active object does not have to be defined in a class diagram. If you omit type, the UML suite assumes type is the same as name.

In Scope region

The In Scope region is a white rectangle that can be placed on top of an initiator, object, or active object to show that the object is in scope. You can move the In Scope region up and down the axis of the object, but not away from it. When you move the In Scope region, nearby messages move with it. When you move messages, the In Scope region does not move. This allows you to move messages from one In Scope region to another.

Label

No label.

Object terminator

The object termination symbol represents the destruction of the object.

Label

No label.

Messages

Messages

A message represents an individual stimulus from one object to another. A message can signal an event or transmit data. It occurs at a point in time.

Multiple messages can be sent from the same point on the initiator or object line; that is, sent at the same point in time. An object can send a message to itself. Such a message must arrive at the object at another point in time.

Label

Each message has three labels:

• A message name with this syntax:
  [condition]message-name[(parameter[,parameter,...])]

  • condition is an optional boolean expression that must be enclosed in square brackets
  • message-name is the message name.
  • parameter is an optional argument.

  Two timing marks, one at either end of the message. Timing marks are optional. You generally use them in combination with timing constraints to document the time interval between two or more messages, as shown in Example SD.

Flat message

The Flat Message symbol is a stick arrowhead. It shows the progression to the next step in a sequence.

Nested message

The Nested Message symbol is a filled solid arrowhead. It represents a procedure call or other nested flow of control. The nested sequence is completed before the outer level sequence resumes.

Asynchronous message

The Asynchronous Message symbol is a half stick arrowhead. It shows an asynchronous message between two objects.

Return message

A return message is a message that an object returns as a result of a process initiated by a message flow.

Create / Destroy messages

A create message is a message drawn to the name of an object.

A destroy message is a message drawn to an object terminator.

Timing Constraints

Timing constraint

Timing constraints are text blocks used to document the time interval between two timing marks, as shown in Example SD.

How to insert a timing mark

Timing marks are labels at the beginning and end of each message.

ÿ To insert a timing mark:

1. Move your cursor to the end of the message, and to the opposite side of the connecting object line.
2. When the text cursor appears, enter the timing mark.

How to insert a timing constraint

ÿ To insert a timing constraint:

1. In the control panel, click the text block tool.

   

2. Click in the drawing area where you want to place the text block. A small square appears indicating the location of the timing constraint.
3. Move the cursor just below and to the right of the square. An I-bar appears indicating that you can enter text.

   

4. Enter the text of the timing constraint. The text replaces the small square.

How to move a timing constraint

ÿ To move a timing constraint, drag it from one location to another using either both mouse buttons (Windows) or the middle mouse button (UNIX).