Question

I have a question regarding sequence diagrams.

When drawing sequence diagrams,

Is it enough to draw one diagram per user case ?

or do we need to draw a set of sequence diagrams to cover each user case instance ? (is a sequence of actions a system performs that yields an observable result of value to a particular Use Case Actor)

Answer 1

What you need to do is describe the functionality of the system enough that you can begin development. That doesn’t necessarily mean 1 sequence diagram per use case; it doesn’t even mean you have to have any sequence diagrams at all. In fact, I would submit that sequence diagrams belong as definitions of individual method calls, not use cases, but they’re only necessary if you’re doing a really big design up front, which is usually not a wise thing to do, as you don’t always have enough information at design time to define the entire system.

I personally rarely use sequence diagrams; I stick with activity diagrams for use cases and only build a sequence diagram if I actually need to define the flow of a particular method of a particular class, and those only happen if circumstances make it necessary to define at design time.