<<extends>>: from Use Case to Collaboration

Kirill

unread,

Mar 17, 1998, 3:00:00 AM3/17/98

to

Hi,
Let's say I have two use cases, one <<extends>> another. How do I reflect
this relationship when drawing Collaboration diagrams for these use cases?
Thanks for any thoughts you might have.
Kirill.
kirill...@asonet.com

Robert C. Martin

unread,

Mar 23, 1998, 3:00:00 AM3/23/98

to

Kirill wrote:
>
> Hi,
> Let's say I have two use cases, one <<extends>> another. How do I reflect
> this relationship when drawing Collaboration diagrams for these use cases?
> Thanks for any thoughts you might have.

There is no standard way. Indeed, there not necessarily a direct
mapping of a use case to a collaboration diagram.

Use cases do not discuss the OO structure of an application. Rather
they talk about how the system responds to user stimuli. They are a
way to rephrase the requirements in a behaviorally centered way.

Collaboration diagrams, on the other hand, are a way of describing
behavior as a set of messages flowing between objects. Before you
can draw a collaboration diagram, you have to have some idea what the
objects in the solution are likely to be.

This is the basic problem of software design. We have nice methods
for formalizing requirements, and nice ways of expressing designs,
but no automatic reliable way of converting those requirements into
those designs.

The process requires creativity and insight. Moreover the essential
insights often create object models that bear little resemblance to
the use cases. Indeed, the best designs are often those that are
executed at a level of abstraction far removed from the use cases.

--
**We are looking for good engineers, See our website
**for more information.

"One of the great commandments of science is:
'Mistrust arguments from authority.'" -- Carl Sagan

Stephan Meyn

unread,

Mar 25, 1998, 3:00:00 AM3/25/98

to

On Mon, 23 Mar 1998 11:23:43 -0600, "Robert C. Martin"
<rma...@oma.com> wrote:

>This is the basic problem of software design. We have nice methods
>for formalizing requirements, and nice ways of expressing designs,
>but no automatic reliable way of converting those requirements into
>those designs.

Yes, for all the preference I have for use cases, I find them a royal
pain, when I try to do something like requirements tracing into
designs. Anybody found a decent way of doing this?

Stephan

Stephan Meyn Remove nospam from address
Senior Consultant Tel Intl+61-414-599-624
Object Oriented P/L
North Sydney, NSW 2060 Australia

SoftDocWiz

unread,

Mar 26, 1998, 3:00:00 AM3/26/98

to

My current co-author, Doug Rosenberg, has some pretty deep expertise with
regard to the traceability of requirements through system design. I'd suggest
you check out his stuff at <http://www.iconixsw.com>, and look for our book,
about use-case-driven development using the UML, in the fall.

Kendall Scott
ken...@softdocwiz.com
http://softdocwiz.com

Ell

unread,

Mar 26, 1998, 3:00:00 AM3/26/98

to

Stephan Meyn (sm...@no-spam.nsw.bigpond.net.au) wrote:
: On Mon, 23 Mar 1998 11:23:43 -0600, "Robert C. Martin"

: <rma...@oma.com> wrote:
:
: >This is the basic problem of software design. We have nice methods
: >for formalizing requirements, and nice ways of expressing designs,
: >but no automatic reliable way of converting those requirements into
: >those designs.

: Yes, for all the preference I have for use cases, I find them a royal
: pain, when I try to do something like requirements tracing into
: designs. Anybody found a decent way of doing this?

Don't be fooled by the naysaysers and fibbers who refuse to acknowledge
what they have read. Read Jacobson's OOSE for a number of examples which
convert use cases requirements directly and automatically into designs.

Elliott
--
:=***=: Objective * Pre-code Modelling * Holistic :=***=:
Hallmarks of the best SW Engineering
"The domain object model is the foundation of OOD."
Check out SW Modeller vs SW Craftite Central : www.access.digex.net/~ell
Copyright 1998 Elliott. exclusive of others' writing. may be copied
without permission only in the comp.* usenet and bitnet groups.

Charles R. Martin

unread,

Mar 26, 1998, 3:00:00 AM3/26/98

to

e...@access4.digex.net (Ell) writes:

>
> Stephan Meyn (sm...@no-spam.nsw.bigpond.net.au) wrote:
> : On Mon, 23 Mar 1998 11:23:43 -0600, "Robert C. Martin"
> : <rma...@oma.com> wrote:
> :
> : >This is the basic problem of software design. We have nice methods
> : >for formalizing requirements, and nice ways of expressing designs,
> : >but no automatic reliable way of converting those requirements into
> : >those designs.
>
> : Yes, for all the preference I have for use cases, I find them a royal
> : pain, when I try to do something like requirements tracing into
> : designs. Anybody found a decent way of doing this?
>
> Don't be fooled by the naysaysers and fibbers who refuse to acknowledge
> what they have read. Read Jacobson's OOSE for a number of examples which
> convert use cases requirements directly and automatically into designs.

Sure, you can apply it mechanically to create *A* design.

But a *good* design? Only by chance.

Patrick Logan

unread,

Mar 26, 1998, 3:00:00 AM3/26/98

to

SoftDocWiz <softd...@aol.com> wrote:

: My current co-author, Doug Rosenberg, has some pretty deep expertise with

: regard to the traceability of requirements through system design. I'd suggest
: you check out his stuff at <http://www.iconixsw.com>, and look for our book,
: about use-case-driven development using the UML, in the fall.

Please provide a more specific URL than the one above. I've already gotten
bored looking through too many pages at that site.

--
Patrick Logan (H) mailto:plo...@teleport.com
(W) mailto:patr...@gemstone.com
http://www.gemstone.com

Robert C. Martin

unread,

Mar 27, 1998, 3:00:00 AM3/27/98

to

Stephan Meyn wrote:
>
> On Mon, 23 Mar 1998 11:23:43 -0600, "Robert C. Martin"
> <rma...@oma.com> wrote:
>
> >This is the basic problem of software design. We have nice methods
> >for formalizing requirements, and nice ways of expressing designs,
> >but no automatic reliable way of converting those requirements into
> >those designs.

> Yes, for all the preference I have for use cases, I find them a royal
> pain, when I try to do something like requirements tracing into
> designs. Anybody found a decent way of doing this?
>

There has been some discussion about traceability on the OTUG list
(www.rational.com, sign up for 'object technology user group').
The folks at ICONIX (Dan Rawsthorne, Doug Rosenberg) have described a
use case method that maintains traceability from use cases to
sequence diagrams. Sounds pretty interesting.

I remain somewhat skeptical of simplistic methods like this, since
I think they break down when designers abstract the design away from the

Ell

unread,

Mar 28, 1998, 3:00:00 AM3/28/98

to

For the benefit of Logan and CMartin who deny that RCM is saying he does
not like and does not practice employing analytical use cases as the basis
and skeleton of physical design.

[ Article reposted from comp.object,comp.software-eng ]
[ Author was Robert C. Martin ]
[ Posted on Fri, 27 Mar 1998 08:03:57 -0600 ]

--

Patrick Logan

unread,

Mar 28, 1998, 3:00:00 AM3/28/98

to

Ell <e...@access2.digex.net> wrote:

: For the benefit of Logan and CMartin who deny that RCM is saying he does

: not like and does not practice employing analytical use cases as the basis
: and skeleton of physical design.

What have you demonstrated? I think you are incorrectly reading between
the lines. You yourself have stated that you "generalize" from the domain
specifics. You seem to be looking for a fight. If you read his book, and
understand it, you would know better.

Charles R. Martin

unread,

Mar 28, 1998, 3:00:00 AM3/28/98

to

e...@access2.digex.net (Ell) writes:

>
> For the benefit of Logan and CMartin who deny that RCM is saying he does
> not like and does not practice employing analytical use cases as the basis
> and skeleton of physical design.

Ell, can you say "non sequitur"? I *knew* you could.

Robert C. Martin

unread,

Mar 28, 1998, 3:00:00 AM3/28/98

to

Ell wrote:
>
> For the benefit of Logan and CMartin who deny that RCM is saying he does
> not like and does not practice employing analytical use cases as the basis
> and skeleton of physical design.
>

Just in case anybody might care; I *do* happen to think that
use cases are powerful tools for describing what an application
is supposed to do. However, I do not accept that the best structure
for an application is derived from the structure of the use cases.
I think that many applications can be better served by designing
them at a level of abstraction that is above the use cases.

Phlip

unread,

Apr 10, 1998, 3:00:00 AM4/10/98

to

Ell wrote:

>For the benefit of Logan and CMartin who deny that RCM is saying he does
>not like and does not practice employing analytical use cases as the basis
>and skeleton of physical design.
>

>[ Article reposted from comp.object,comp.software-eng ]
>[ Author was Robert C. Martin ]
>[ Posted on Fri, 27 Mar 1998 08:03:57 -0600 ]
>
>Stephan Meyn wrote:
>>
>> On Mon, 23 Mar 1998 11:23:43 -0600, "Robert C. Martin"
>> <rma...@oma.com> wrote:
>>
>> >This is the basic problem of software design. We have nice methods
>> >for formalizing requirements, and nice ways of expressing designs,
>> >but no automatic reliable way of converting those requirements into
>> >those designs.
>
>> Yes, for all the preference I have for use cases, I find them a royal
>> pain, when I try to do something like requirements tracing into
>> designs. Anybody found a decent way of doing this?
>>
>
>There has been some discussion about traceability on the OTUG list
>(www.rational.com, sign up for 'object technology user group').
>The folks at ICONIX (Dan Rawsthorne, Doug Rosenberg) have described a
>use case method that maintains traceability from use cases to
>sequence diagrams. Sounds pretty interesting.
>
>I remain somewhat skeptical of simplistic methods like this, since
>I think they break down when designers abstract the design away from the
>use cases.

Of course I see the relationship. RCM states that he remains skeptical of a
(paperwork) method that permits traceability between the Use Case scenario
and the resulting design.

This clearly means he thinks there should be _no_ traceability between the
two! and that he wants to _throw_ Use Cases _away_!

He just wants to draw diagrams and see how many Design Patterns he can cram
into them!!

Elliott, he is obviously plotting to take over the world by writing crap
worse than Microsoft! THANK YOU for exposing this THREAT in our MIDST!

-- Phlip
======= http://users.deltanet.com/~tegan/home.html =======
-- Search Keywords: Software, C++, OO, Windows, Traci Lords --