Relation between Use Cases and Activity

[sargasso]

Another option is to create Use case elements for each scenario and use the realize association between the primary Use case and the scenario.  You could then make each element a composite element and navigate to the diagram that goes with each.   (Not sure if this is really UML compliant though).

Neither do I - but when it's use is justifiable it really works well!

One time I felt it was justified was when the outcomes of the scenarios were so wildy different that they really needed to be separate use cases.  However from the users point of view they were the same use case. Using the <<scenario>> realises use case technique allowed us to progress the analysis phase with a good agreement and communication between the designers and the users.


bruce

[TrtnJohn]

Another point to consider is that if your Use Case is getting too complex maybe you aren't factoring them correctly.  I start to get nervous if my scenarios get too complex.  This is usually an indicator to me that I am starting to become less user/actor centric and more system oriented when defining my Use Cases.  In these cases I usually started with something too broad.  To fix this I try and rethink the user point of view and break the cases into more simple tasks. And then I use the <<include>> association to tie the Use Cases together.  I really don't like having huge scenarios that need an Activity diagram in order to understand them.

[thomaskilian]

I find Use Case modelling always a bit touchy. The simple reason is that you interface between REAL world and ABSTRACT model. This abstraction process has been described in many informal ways so far. Unfortunately you will never reach a stadium where you can have a complete formal way to transform REAL to ABSTRACT (as far as I understood the main thesis of Gödel). So you will always have to use your stomach to decide about the process and how you use it.

[mikewhit]

I thought Gödel said that you couldn't have a complete AND self-consistent system ?

As I have said, I did like the Shlaer-Mellor OOA approach.

[thomaskilian]

No. It's simpler. You just can't have a complete axiomatic system. This BECAUSE you can put an axiom inside which will crash your system. So my intention is to say: you will never have a formal way to come from reality to abstraction. (I'm no mathematictian though, only an old computer scientist. Read: old 'computer scientist', not 'old computer' scientist.)

[mikewhit]

I thought (old (computer scientists) ) were just practical mathematicians anyhow !

PS. Did you get to the end of G.E.B. http://www.amazon.co.uk/exec/obidos/ASIN/0140179976/

[thomaskilian]

I got by without any mathematics except for parts of my school knowledge (btw. my wife has a doctors degree in Applied Numerical Mathematics and I can ask her whenever it's neccessary).

And yes: I read it (understandig parts of it) in the year G.E.B. was published. Some 2 or 3 years ago I read it again and was even more impressed than the first time. I like Eschers mind puzzling pictures. I have about a hundred CD with Bachs music which I listen regularly. And maybe I'll *really* understand Gödels thesis when I'll going to read G.E.B. the third time