How deep do (we) analysts model ?

[sshearer]

Hi all,

I was reviewing discussions posted and started to read the one about "Sequence Diagrams and .Net" (http://www.sparxsystems.com.au/cgi-bin/yabb/YaBB.pl?board=UMLPRO;action=display;num=1116873337) when a thought occurred to me:

At what point DOES modeling stop and development begin ?

I realize that the answer depends on what your position in an organization is, and the size of the team you are in.  I know the range CAN go from 1 person who will just dive into development through to a team of analysts each doing high-level modeling a portion of a project that will be given to a development team to drive down further.  I know that in order to generate code, you need to drill down to the most detailed levels.

But my question is really more "What is being done in organizations today?"

Is there some way this posting could be converted to a Survey ?  Has a survey like this already been done ?

I'm interested in the answers because I'd like to know what tool-independent lessons (if any) people have learned from trying to generate source code from their models.

Thanks,
Steven.

[sargasso]

At what point DOES modeling stop

AFA I am concerned and anywhere I have control - never.  The model is an asset of the organisation and is of equal value to the delivered code/system.

and development begin ?

In my experience, about 2pm.    Seriously,  and in the light of the above, development per se "begins" IMO when the first coder starts the first prototyping exercise to solve the first realisation problem encountered.  AFA I am concerned and anywhere I have control even these go under config management. They have on many occasions proven their value later in the project or life of the system as prompters as to "what the hell were we thinking".

hth
bruce

[Paolo F Cantoni]

AFA I am concerned and anywhere I have control - never.  The model is an asset of the organisation and is of equal value to the delivered code/system.

Absolutely!!!  (but you already knew I would agree with you on this one bruce! )

In my experience, about 2pm.    Seriously,  and in the light of the above, development per se "begins" IMO when the first coder starts the first prototyping exercise to solve the first realisation problem encountered.  AFA I am concerned and anywhere I have control even these go under config management. They have on many occasions proven their value later in the project or life of the system as prompters as to "what the hell were we thinking".

hth
bruce

Again I can only agree...    The biggest missing factor in design documents and code is WHY???  Don't bother documenting what the code already says it does - anyone can read the code.  Say WHY it is doing it...

Paolo

[sargasso]

To be a bit more circumspect and to use a Grady Booch analogy the answer is "It depends".
If one is building a dog house, one may scratch a few preliminary sketches on the inside of a beer carton, grab a few bits of wood and start building.  If your building a private domicile you may progress through a series of conceptual sketches to a full set of architectural drawings and specifications before committing spade to dirt, saving the plans for posterity.  If you are building a public accomodation high-rise tower, you will prepare a full set of formal conceptual and physical specifications, drawings and models before anything.  After the tower is built those artifacts become an active asset used continuously in the proper maintenance of the facility.
Now you decide - are you building a doghouse or the Dorchester?
bruce