[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [aspectj-users] CNET News.com Article on AOP
|
Is this, the main claim of the article, true?
IBM and JBoss Group have teamed on a software
development technique they hope to add to Java
...
I saw no IBM sources in this article. Ron, do you know
of any official joint effort by IBM and JBoss Group?
Also, does anyone know if Forrester analyst John Meyer
was talking about AOP when they appended it to
his quote?
"It's going to be several years before the tool
industry can adopt the extensions that are approved
and ratified in the Java (standardization) space
(and) before people will be ready to adopt"
aspect-oriented programming.
I would think an analyst would distinguish between
tool/IDE productivity gains and language gains
(though the point about adoption could be true for
any JCP proposal, the article makes it sound like
he was considering AOP).
Also, I didn't realize that BEA itself did this:
BEA last month introduced the WebLogic Aspect Framework
I thought Sam Pullara (who happens to be a BEA employee)
posted it to a weblogic dev list. Does that mean BEA
introduced it? That makes it sound like a marketed and
supported feature of WebLogic.
Perhaps uncharitably, it seems like JBoss is trying to
enhance their stature by associating themselves with
(a) IBM and (b) AOP, even though they are late to the
game and haven't themselves contributed much to
the field. To make this sound plausible, you have to
use Microsoft C#; it's like saying that since
democrats and republicans were both against Russia
during the cold war, they're working together. You
hate to fight with allies that might help with AOP,
but profiteers make it harder to win the battle
for hearts and minds.
I guess we've been lucky in the past that articles
on AOP were mostly correct. If this is -- if
IBM is pushing Sun to standardize AOP, BEA
officially released support for AOP based on AspectJ,
and John Meyer is evaluating how long it will take
for developers to adopt AOP -- then I'll be psyched!
Wes
P.S. - Can we think of a better way to characterize
AspectJ than "experimental"? I'd think that a few
years of research, patents, and a thriving community
including many deployments in commercial contexts is
something more than that.
On Thu, 25 Sep 2003 08:37:51 -0700
"Ron Bodkin" <rbodkin@xxxxxxxxxxxxxx> wrote:
> Hi William,
>
> I believe AOP code is higher quality, because it is
> better modularized. You just can't rely on consistently
> implementing a policy for crosscutting concerns in
> scattered and tangled code. Errors always creep in. So if
> you care about implementing fine-grained security, error
> handling, performing statistics, or tracing effectively,
> AOP results in better quality through consistent
> application and update.
>
> I agree completely that AOP, like any powerful
> technology, can be misapplied. Just as OO was (and is).
> And that skillful engineers will improve the quality in
> any implementation.
>
> I've found it's sometimes necessary to refactor code to
> expose join points to AOP (better OO designs require
> little or no refactoring). And it's better to design with
> AOP in mind, though good OO designs are generally quite
> amenable to adding aspects (or refactoring to use
> aspects).
>
> I don't want to see AOP being overhyped either, but I
> don't see that happening (at least not yet). If anything,
> the PR it's getting is addressing a subset of what it can
> do (e.g., JBoss promoting it as a way of getting C#
> capabilities for Java).
>
> Thanks,
> Ron
>
> Ron Bodkin
> Chief Technology Officer
> New Aspects of Security
> m: (415) 509-2895
>
> > ------------Original Message-------------
> > From: "Louth, William" <william.louth@xxxxxx>
> > To: <aspectj-users@xxxxxxxxxxx>
> > Date: Thu, Sep-25-2003 8:05 AM
> > Subject: RE: [aspectj-users] CNET News.com Article on
> AOP
> >
> > Ron,
> >
> > I did not really like the reference to your view that
> AOP (AspectJ) code
> > tends to be better. This could be attributed to the
> fact that early
> > adopters tend to be skillful engineers who recognize a
> new tool and its
> > application to their problems, ;-).
> >
> > I am sure that as soon AOP becomes mainstream you will
> see poor designs
> > and code as people apply the technology wrongly - which
> is what happened
> > to EJB.
> >
> > I believe that AOP has a promising future for complex
> development
> > projects in the hands of experienced engineers but as I
> have stated
> > previously the more I apply AOP the more I am drawn
> into refactoring
> > existing designs and code. AOP will not solve a poor
> design and it
> > proves quite hard to introduce it into systems which
> have not been
> > designed with this in mind. There still exists problems
> with scaling AOP
> > to large projects. I really hope that JBoss's framework
> does not become
> > widely used until we solved how to handle the
> direct/indirect
> > interactions between different aspects on common
> object/component
> > models.
> >
> > Its great to see AOP having greater media coverage but
> I hope we don't
> > promise too much.
> >
> > All the best,
> >
> > William Louth
> > William.louth@xxxxxx
> >
> > -----Original Message-----
> > From: Ron Bodkin [mailto:rbodkin@xxxxxxxxxxxxxx]
> > Sent: Thursday, September 25, 2003 4:28 PM
> > To: aspectj-users@xxxxxxxxxxx
> > Subject: Re: [aspectj-users] CNET News.com Article on
> AOP
> >
> >
> > Indeed, the JBoss crew view aspect-oriented programming
> as a means of
> > implementing attribute-oriented programming, and that
> perspective is
> > clear in the article. And therefore it's also the
> emphasis of their
> > message, namely fighting C#. This is probably an
> attempt to pull
> > together with the rest of the Java world, by focusing
> on a common enemy
> > in Microsoft.
> >
> > However, the article does discuss defining rules and
> enforcing policies
> > for the entire system, which isn't something that
> attribute-oriented can
> > do (at least not completely). Given the audience is
> non-programmers, I
> > think the description is reasonable.
> >
> > Overall, having AOP hit the radar screen like this is
> helpful, even
> > though I'd like to see the title be "Growth in AOP for
> Enterprise Java."
> > Feel free to encourage Marc Fleury to change his
> message ...
> >
> > Ron
> >
> > Ron Bodkin
> > Chief Technology Officer
> > New Aspects of Security
> > m: (415) 509-2895
> >
> > > ------------Original Message-------------
> > > From: Carlos E Perez <ceperez@xxxxxxxxx>
> > > To: aspectj-users@xxxxxxxxxxx
> > > Date: Thu, Sep-25-2003 3:30 AM
> > > Subject: Re: [aspectj-users] CNET News.com Article on
> AOP
> > >
> > > It's good PR, however, the main theme of the article,
> > > that is "a way to fight the rising popularity of C#"
> completely
> > > distracts the audience from the real issue.
> > >
> > > The other problem is the association of Aspect
> > > programming with Attribute based programming. If you
> replaced
> > > "Aspect-oriented" with "Attribute-based" this article
> would have made
> > > more sense.
> > >
> > > Carlos
> > >
> > >
> > > --- Ron Bodkin <rbodkin@xxxxxxxxxxxxxx> wrote:
> > > > This article was published today as the lead story
> > > > on CNET News.com:
> http://news.com.com/2100-1007_3-5081831.html.
> > > > It's good to see the increase in interest and
> awareness
> > > > of AOP from industry news sources (as well as
> > > > industry analysts).
> > > >
> > > > Ron
> > > >
> > > > Ron Bodkin
> > > > Chief Technology Officer
> > > > New Aspects of Security
> > > > m: (415) 509-2895
> _______________________________________________
> > > > aspectj-users mailing list
> > > > aspectj-users@xxxxxxxxxxx
> > > >
> > > http://dev.eclipse.org/mailman/listinfo/aspectj-users
> > >
> > > _______________________________________________
> > > aspectj-users mailing list
> > > aspectj-users@xxxxxxxxxxx
> > > http://dev.eclipse.org/mailman/listinfo/aspectj-users
> > >
> > >
> > _______________________________________________
> > aspectj-users mailing list
> > aspectj-users@xxxxxxxxxxx
> > http://dev.eclipse.org/mailman/listinfo/aspectj-users
> > _______________________________________________
> > aspectj-users mailing list
> > aspectj-users@xxxxxxxxxxx
> > http://dev.eclipse.org/mailman/listinfo/aspectj-users
> >
> >
> _______________________________________________
> aspectj-users mailing list
> aspectj-users@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/aspectj-users
