Re: Offtopic: Reuse

[William W. Lee]

Please see my comments below:

At 03:20 PM 5/24/99 +0200, you wrote:
>On the topic of reuse, which has been discussed the past weeks. Slightly
>offtopic. Still here? Ok, read on :-)
>
>My experience is that whether or not a component is reusable or not is
>not something that just "happens"("oh, it was reusble"/"damn, not this
>time"), but rather a conscious decision as a parameter during the design
>and implementation. To really be able to reap the fruits of reusability
>care must be taken so that a component or suite of components are
>reusable. Reuse is not easy, and by adding it as a design parameter one
>must be aware of that the total cost will probably go up, as it should
>be designed to handle needs that are not yet there. There are certainly
>rules that helps this process, such as the use of design patterns etc.
>The basic reason for doing reusable work is that cost and effort will be
>lower in the long term. For me the basic meaning of the word reuse is
>"to not reinvent the wheel".
>

Couldn't agree with you more, Rickard. Reuse is an expensive process,
and requires additional upfront cost for any organization that wants
to _create_ application components that are reusable. I would think
with the current short deadline schedules, new business services offerings
have a higher priority over making applications reusable. One could
argue that if organizations invest in creating reusable applications,
then new business services can be rolled out quicker in the long run.

Let's say organizations can start their development effort with the
help of pre-built application components. These components take advantage
of reusability at different levels that you mentioned. This lowers
the upfront cost, because now organization can acquire the components
from different component vendors. This is the promise from the
component industry. It's not easy to get there.


<vendor>
We believe our JumpStart can take you there.
</vendor>

>One should also be aware of that reuse is possible at different levels:
>* Component reuse: the entire component can be reused in many different
>situations without modifications

<vendor>
This is easy, we provide common (horizontal) business objects that
can be reused across industries.
</vendor>

>* Customization: the component has some customizable properties that can
>be changed to handle different situations

<vendor>
We provide "pluggable" methods / workflows, runtime configurable properties.
</vendor>

>* Extensibility: the component can be *easily* subclassed to provide
>different behaviour. Note that this takes some effort as it requires
>that the component designer adds hooks and structured ways to accomplish
>this

<vendor>
The keyword here is "easily". We make it easy to extend components!
</vendor>

>* Implementation reuse: useful implementation patterns is broken out
>into reusable packages (such as generic longlived locking, dirty flag
>management, etc.)
>* Design reuse: aka Design Patterns, that is some design ideas that
>works well for a certain class of problems. For example:
>"a stateful session bean can be used to gather information through a
>series of wizard-like JSP pages which then can be used to modify the
>state of a set of entity beans by use of a shortlived transaction". This
>is a simple EJB-ish design pattern that will probably be reusable in
>many different applications.
>(+ probably some more that I can't think of right now)
>

In order to achieve the above reusability levels, many analysis
and design patterns must be used.

>When we say "the component is reusable" we (probably) intuitively mean
>the first meaning. But if the full range of reusability levels are
>considered far more work can be reused, and if done well, can
>substantially reduce application development time and effort.
>

Couldn't agree more.

Will
-------------------------------------
William Lee, Chief Technology Officer 1.617.573.5099 (Voice)
The Theory Center 1.617.262.0703 (Fax)
28 State Street, 11th Floor [EMAIL PROTECTED]
Boston, MA 02109 http://www.theorycenter.com
-------------------------------------

>Just my 2 cents... (FWIW etc.)
>
>/Rickard
>
>--
>Rickard �berg
>
>@home: +46 13 177937
>Email: [EMAIL PROTECTED]
>Homepage: http://www-und.ida.liu.se/~ricob684
>
�������������������������o unsubscribe, send email to [EMAIL PROTECTED] and include in the body of the message "signoff EJB-INTEREST". For general help, send email to [EMAIL PROTECTED] and include in the body of the message "help".