How to represent an event based architecture in a static UML model?

Question

I have a fairly basic C# event based system but I'm not sure how I model it in UML. I obviosuly want to show the event publisher, subscriber, handlers and EventArgs classes .. I think you use 'signals' but I can't find any examples. Can anyone point me to an example or shed any light?

Thanks

Edit: I am creating a static model, I don't need to represent state or paths through the process. Sorry If i didn't make that clear in the initial question..

Answer 1

The "Publisher-Subscriber" pair pattern (a.k.a "Observer"), may be implemented different in each programming (language) framework, therefore, designed different, in U.M.L.

Any way, conceptually, when an event ("signal" or "message") is sent, from a publisher (a.k.a "server") to any subscriber ("client"), sometimes, an "id" to identify a particular event, from other events, its provided, and some additional parameters or data its also sent.

As other answers already mention, you may require a (class) diagram to describe the static model. (Note that there is a "aggregation", not "composition", "association" can be used):

..............................
+--------------------------+..
|      <<Publisher>>       |..
|      VectorDrawApp       |..
+--------------------------+..
| [+] create()             |..
+--------------------------+..
| [+] send(EventArgs e)    |..
+------------+-------------+..
............/ \...............
............\ /...............
.............|................
.............|................
+------------+-------------+..
|      <<Subscriber>>      |..
|          Figure          |..
+--------------------------+..
| [+] create()             |..
+--------------------------+..
| [+] receive(EventArgs e) |..
+--------------------------+..
..............................
+--------------------------+..
|        <<Event>>         |..
|        EventArgs         |..
+--------------------------+..
| [+] Sender: TObject      |..
+--------------------------+..
| [+] receive(EventArgs e) |..
+------------+-------------+..
.............|................
.............+................
............/ \...............
...........+---+..............
.............|................
+------------+-------------+..
|        <<Event>>         |..
|  FillEventArgs: EventArgs|..
+--------------------------+..
| [+] ForeColor            |..
| [+] BackColor            |..
| [+] FillStyle            |..
+--------------------------+..
..............................

And also, you may require a diagram to describe the dynamic model:

.........................................
+----------------+..+----------------+...
| <<Publisher>>  |..| <<Subscriber>> |...
|  VectorDrawApp |..|      Figure    |...
+--------+-------+..+--------+-------+...
.........|...................|...........
.......+-+-+...............+-+-+.........
.......|   |...send(fill)..|   |..Fill().
.......|   +==============>+   +---+.....
.......|   |...............|   |...|.....
.......|   |...<<return>>..|   |...|.....
.......|   |<--------------+   +<--+.....
.......|   |...............|   |.........
.......+-+-+...............+-+-+.........
.........|...................|...........
.........X...................X...........
.........................................

Stereotypes, in U.M.L., are your "drinking buddies", and allow you to describe or restrict what an actor, object, class, trait, or interface does.

When you use them, highlight when an object or class, are subclasses of a class, or implement, an interface that is relevant to the activities, that are been model, even if there are other parent classes, or interfaces.

Cheers.

Answer 2

As indicated by lexu and John, you can use statecharts and activity diagrams to model some of the dynamic aspects of your system.

For your static model, you can model the events a class can handle as operations. You can use a stereotype (<<event>>) to differentiate these operations from others (e.g. synchronously called methods).