RE: Help...Verify UML Design



RE: Help...Verify UML Design

From: Islam Essam <islm78_at_yahoo.com>
Date: Wed, 28 Jun 2006 05:01:07 -0700 (PDT)
Message-ID: <20060628120107.54095.qmail@web53506.mail.yahoo.com>
First of all I want to thank all people that
participated in this mail thread that I’ve created 4
days ago. This group contains really a very qualified
people.

Regarding the discussions that happened for making sw
designs repository, I totally agree with Tom for
dividing the discussion into two parts:
1-Repository for UML models of a pre-designed
software, for example (Hospital DB, POS, Accounting
System, …) and many other Case-Studies. Each model
must have different UML diagrams like Class Diagram,
Sequence Diagram, State Diagram, Activity Diagram, …
2-Repository for Metamodels, Actually I’m not expert
in this field so I’ll leave other people to discuss
this issue.

Regarding the first point (UML models) it could be
found in one of the following:
a-Enterprises: as designs for their Software
b-Technical Papers: as examples to illustrate their
work
c-Students: as small/medium projects

The problem with (a) is that Enterprises will never
let their designs go out of their organizations, and
it is logic.
The problem with (b) is that Technical Papers just put
a very small portion of the design to illustrate a
specific point. You will never find a complete UML
design in such papers.
The problem with (c) is that Students make their
projects by the book (very academic), so you cannot
test real life problems that will happen in real large
projects, just like the inconsistencies problem that
I’m trying to address. 

So, taking the initiative to make a repository of UML
designs is very effective, especially for those who
are trying to enhance UML designs using third-party
tools. 

From this point, I suggest to make a web site
dedicated for that purpose, this web site will be
acting as a common repository that will allow
volunteers to upload their UML designs. I suggest also
to agree on a standard format for the models (what
about using XMI as a standard representation for the
models and another PDF file that contains the diagrams
as images)

By doing such a project, many people will be able to
test their tools, and verify the existing models.
Then, they can report their comments on the web site
also. 
For example, If some one creates a tool that verify
the inconsistencies in the UML models, he will report
these kinds of inconsistencies that found in a
specific models. 

By that way researchers will be able to compare their
approaches and technique in such intangible problems.

If anyone could volunteer in making this web site, he
is very welcomed. Who says I??

@@@@@@@@@@@@@@@@

Regarding my tool that check for UML inconsistency,
the tool checks for the following sources of
inconsistency in any UML model represented in XMI
format:
1-Shortness in applying syntactic rules
2-Shortness in applying semantic rules
3-Undefined modeling process
4-Incapability of dynamic execution
5-Intra-model inconsistency 
6-Inter model inconsistency
7-Incompatability with the target language

The tool is part of my thesis that aims to cover all
these inconsistencies sources based on formal methods.
A paper that describes that technique in details will
be published after I find a way to test the tool,
which will be the UML designs repository (a dream till
now)

Sorry, for the very long email ,,, and thank you for
your time.

Best Regards,
Islam Essam. 


--- Chris Britton <chris.britton@blueyonder.co.uk>
wrote:

> I too have been looking at the problem of
> verification of design but very
> much from a non-theoretical point of view. I think
> the notion of
> testing/analysing/verifying is incredibly important.
> 
> But I have found UML is missing a lot of data that's
> needed to do a
> comprehensive job. In particular the high-level
> modelling is missing; it
> does not have the data to allow you to do a good job
> at modelling the
> relationships between activities, data, people and
> external
> actions/entities. Since an IT application (at least
> a business operational
> application) is a cog in the machine, not the whole
> machine (the whole
> machine is the overarching business process), you
> cannot verify the IT
> application is going its job unless you have
> modelled the context for that
> job.
> 
> OK I'll get off my soap box now. But if you are
> interest I wrote some papers
> on the subject which you can read at
> www.itmodeller.co.uk/papers.htm. Any
> comments gratefully received.
> 
> Best regards,
> Chris
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> To remove yourself from this list please mail
> puml-list-request@cs.york.ac.uk
> with a message containing the word "unsubscribe".
> 
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 
Received on Wed 28 Jun 2006 - 13:01:07 BST