On 14/01/07, Ashley at Metamaxim <[EMAIL PROTECTED]> wrote:
>
>
>
>
>
>
>
> Steve
>
> > I'm talking not just about IT enabled  processes but actual _business_
> > processes. Taking an extreme scenario  look at the Army. The person
> > in control gives a set of orders which the  sub-ordinates follow to the
> > letter. This is centralised control of  process, other examples would
> > be film directors, rescue services and  many others. Also included
> > would be any process where the process is  "when you finish that step
> > come and ask me what to do next".
> >  What I'm saying is that there are real business processes that are
> >  centralised via a controlling party.
>
> I agree. But I think that, even in these  cases, if "what needs to happen 
> next" is primarily determined by the states  of underlying objects, the type 
> of modelling in the paper is appropriate. If you  add a "business process" on 
> top, with its own sequencing of events, you can run  the risk of constraining 
> the flow of events in the business in an arbitrary  and unnecessary way.

What would you describe as the "objects" in a military scenario who
state determines the next step? I'd have a similar questions in other
scenarios (e.g. American Footbal coaching with its fixed plays called
from the side lines).

>
> Though I doubt whether this argument would be  enough to convince the Army to 
> change its ways :-)

Indeed, and I'd say that in fact this is a good reason why in fact the
object modelled approach isn't viable in this sort of scenario.
Controlled action is sometimes required, this requires very specific
modelling and messaging approaches and very strict compliance
procedures.  This isn't something that is needed all the time but my
point was that one size doesn't fit all.

>
> > Which for me is actually technology because  its assumed that the
> > solution is technology implementation. There needs  to be design
> > (hopefully!) but the assumption is that technology  implementation is
> > the end goal.
>
> I think you may be making the assumption  that the models are models of a 
> computerised solution. They are not -- they are  just models of the business, 
> and the rules and policies of the  business. The fact that they are 
> executable is a by-product of the fact  that they are formalised and have 
> behavioural semantics, but there is no  implication or requirement that the 
> actual business solution be  computerised.

Fair enough then I mis-read it, but the paper is extremely technical
in nature, and you did say its as technical as a "class diagram"
(which IMO is technical).  The key question therefore is what real
business scenario would this work for, and to answer this you need
something that works at a higher level of abstraction.

>
> I think that the key consideration before this kind  of modelling is 
> undertaken is whether the degree of formalisation of the  business entailed 
> is possible, appropriate and useful. I think that sometimes  you can only 
> find out by trying.

The problem with IT however is that we don't appear to have other
ways! :)  I like the models where the object are actually active
participants in the process, but I think its going to be relatively
limited in its application.
>
> > What I'd argue is that the paper you've  written needs
> > to be set in a wider context that selects the right  technology
> > implementation approach
>
> I don't have any problem with that.
>
> Rgds
> Ashley
>
>                   

Reply via email to