Re: Ambiguities of the Harel/UML FSM model

Responding to Huber...

The bottom line was that Pathfinder had to define its own rules for
resolving the action sequence/context.



Pretty much every standard in computer science leaves such leeway,
that's called unspecified behavior. The more interesting question is
whether a user employing two different implementations could run into a
problem because the behavior is unspecified. For example, when exiting a
state configuration with orthogonal regions, the UML standard does not
specify in what sequence the regions are exited. In practice this is not
a problem because:




a) The user knows (or should know) that the sequence is unspecified

As I recall that was pretty much the problem; the sequence can't be unspecified in the model because code generators are quite literal-minded and they do what you say, not what you meant. If a different code generator using different assumptions is used the resulting application could potentially yield different results without any change to the user's model. So the model specification /must/ be unambiguous about the functional behavior.

[BTW, as I indicated, this issue came up in a meeting I attended a few years ago. It is possible the spec has been fixed since then because of all the work done on an execution model starting in v1.5. (Lack of a proper execution model for UML represented a major hole in the UML spec until then.) Since I am on the model side, I wouldn't be aware of any such fixes on the transformation side. However, I am skeptical because it usually takes years to iron out inconsistencies with peripheral definitions in UML when something changes. The spec has become so big that the Left Hand/Right Hand problem is ubiquitous.]


*************
There is nothing wrong with me that could
not be cured by a capful of Drano.

H. S. Lahman
hsl@xxxxxxxxxxxxxxxxx
Pathfinder Solutions  -- Put MDA to Work
http://www.pathfindermda.com
blog: http://pathfinderpeople.blogs.com/hslahman
(888)OOA-PATH



.

Relevant Pages

  • Re: Interface usage best practices
    ... Rational's contribution was mainly in the original UML proposal that was ... > Why do you see OCL as a proprietary standard? ... doesn't affect the fact the semantics is well defined. ... The operation name implies a responsibility, ...
    (comp.object)
  • Re: Interface usage best practices
    ... Just "the Booch notation". ... It is far superior to UML in many respects. ... Why do you see OCL as a proprietary standard? ... In general I favour that approach, its how SDL tools work too in ...
    (comp.object)
  • Re: does a software architect need UML skills?
    ... It is called a defacto standard. ... a lot of companies will say they use UML for all ... And that is a pretty bad characteristic for an architect. ... and I'm perfectly capable of communicating with ...
    (microsoft.public.dotnet.languages.csharp)
  • Re: Any XML (or another ASCII-based file) standard UML?
    ... Such standard for portability between UML ... stereoRef ... create an XML schema for UML. ...
    (comp.object)
  • Re: Which are your favorite UML tools?
    ... I plan to add support of UML CASE ... :)) Interoperability via XMI is a nightmare I ... one tool which fully implements the corresponding standard specifications. ...
    (comp.lang.python)