Re: Holy Smoke!

wizofoz2k_at_yahoo.com.au (Noons) wrote in message news:<73e20c6c.0309111346.1121bcce_at_posting.google.com>...

> technology. They are all born and bred from the OO movement of the
> early 90s, with NOT A SINGLE gram of prior IT experience anywhere.

That's a very good observation, though I would make one notable exception: Robert C. Martin from Object Mentor, who's been in this industry for decades. I tend to find his writings tend to be the some of the most pragmatic of the bunch (though lately he really has bought into XP quite thoroughly).

He was the first instructor that I've experienced that tried to pin down what *specfic* things that OO languages can do well, as opposed to hand-wavy bubbles and lines "everything is an object, love objects" stuff. Most of his concerns surround dependency management.

> anything. Unless you consider that "patterns" are design solutions.
> With which I disagree most strongly: they are possible solutions for
> *development*. In one narrow environment: J2EE. They never were intended
> to replace general IT design.

The original patterns were bastardized by industry. They were either specific solutions to a specific context of problem, or a language/vocabulary for discussing tradeoffs.

Unfortuantely, with J2EE, they turned into a consultant's wet dream of cookie-cutters.

> > There's no central arbitration of taste, so you might not like Ambler's
> > models or Cockburn's use cases, but they're OK with that.
>
> I'm quite sure they are. The whole show has gone on for so long very
> few around still remember we didn't need 5000 pages to define a
> design for a shopping cart 10 years ago. And it worked as well as any
> new design.

One note I will make is to not lump Cockburn in with Ambler. They have very different areas of expertise.

Cockburn's his speciality is to interview project teams and discover what made a team successful or fail, and he also is interested in requirements gathering. He's not a design or UML guy at all (his work on "use cases" has nothing do with UML) He is most certainly not a cookie-cutter man, and is not a one-size-fits-all process advocate. I attended a couple of seminars of his, tis why I say this.

Cheers
Stu Received on Fri Sep 12 2003 - 21:40:10 CDT