Responding to Bestler... ... The first sentence is not necessarily true. I have already provided two examples of Moore equivalents of Mealy machines that did...
Responding to McNeile... Since Higham hasn't responded, I will take a shot... ... I think the intent was to convey the idea that the STT describes all the ...
Responding to H.S.... My point in item (1) is this: The definition of a state machine given by Paul Higham, that it is a "a function d:Q X A à Q where Q is...
... <snip> ... No I meant the event data. The decision is functionally dependent on object AND event data. Suppose an object takes one of two transitions when...
... What you seem to be consistently missing is that there is a class of applications where an object has states that MUST be visisble, because they are part...
... Academically I can concede that an Internet protocol is more mutable than a mathametical algorithm. Concrete examples from any of my recent dealings with...
... natural place to put an action? Reducing the balance is part of a Withdraw, and the action naturally goes (Mealy style) on the Withdraw transition. Putting...
... I think migratory "sub-classes" are often a very informative and necessary part of the class diagram. They allow you to model associations that only exist...
This month's Embedded Systems Programming contains a nice article by Stephen Mellor on the subject of xtUML. http://www.embedded.com/story/OEG20030115S0043...
... As far as I can determine there is no way with eUML to specify a reflexive transition on a parent state that performs certain actions (such as forwarding a...
Responding to McNeile... ... The ordered pairs are not transitions; they are {state, input}. That says nothing about about how many transitions there are in...
Responding to Bestler... ... Just to clarify, the behavior in a transition action does not have to depend on the event identity but it could, which would be...
Responding to Bestler... ... If those states must be public, then use something like the GoF State or Strategy patterns. But the context here is object state...
I'm afraid that my day job was getting in the way of responding to this, thanx to H.S. for filling in. My responses to items 2 and 3 would be more or less...
Responding to Paul Higham ... Yes -- I agree. If you change the state machine by adding these states you can make your defintion work. However, I didn't...
Responding to H.S.... On (1) I have responded to Paul. On (2) and (3). I still say that reducing the balance is naturally associated with the Withdraw...
... A polymorphic event should leave the super-class' state unchanged and propagate to the sub-class. ... I don't think that Executable UML considers the...
... Where you would have to redundantly specify the *same* behavior for each sub-class. I hope we all agree that redundant specification of behavior that is...
Responding to McNeile... ... I agree. Some object somewhere must encapsulate the rules and policies for updating accounts surrounding an actual account...
I agree -- validation and update are separable, and should be separated. I find that, when working with non-technical business people, the "natural placment"...
... ??? Would you? If the behavior is the same for each sub-class, then maybe the transition should exist on the super-class' state model. ... I have to...
Responding to McNeile... ... But then they are even less likely to understand state machines. B-) ... The key lies in whether one associates events with...
Responding to H. S. ... ... This isn't my experience. Business people like state charts -- as long are they logical from a business perspective. I've always...
Responding to McNeile... ... Fascinating. Are you using state charts for domain modeling or object modeling? One of the reasons translation has not moved out...
... Don't forget that the "user community" is not always Business People. Frequently it is composed of other software engineers. This is true when you do GUI...
Responding to H.S. ... I am doing domain modelling -- but producing executable models. Here is a little example that may illustrate my point more clearly: ...
... There are a wide range of applications where an event implies very specific behavior on the receiver's part. It is part of the domain policy, not...
Responding to McNeile... ... Then the state machines defined in the domain are not object state machines (i.e., the behavior will span multiple objects in the ...
... This doesn't match the definition I gave of a "toolkit", but it is another excellent example of a problem that is best defined in terms of explicit actions...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
