Re: JBoss + EJB3 + Liferay

Hi Shawn, FWIW here's my opinion,

   I think its fairly unique. How is Velocity? Why did you choose this
stack? I know BEA was pushing jBPM a while back. Not sure what the queue
vendors (Vitria/Tibco/et al) push. There already were alot of BPM tools but
IMO BPM is a marketing buzzword as the majority of projects are exactly
business process implementations. I think the industry moved to portlets too
for new projects. Refactoring JSP was too much actual work and required
thinking about what you were doing rather than hacking away a million miles
an hour. As for AJAX, its almost a requirement to do operational systems in
a web browser without applets or ActiveX. No responsive operational system
is possible without some client-server interaction IME because even using js
with hiddens/iframe/cookies/other caching quickly blows up in IE on the
average worker's desktop hardware.

   Interviewing a few months back I noticed three distinct camps.
1) Large businesses are still using J2EE but have not moved to EJB3 etc
because it really offers nothing to them. A general float away from
proprietary frameworks (BEA) and such.
2) Medium businesses are using various open source frameworks.
Spring/iBatis/etc.
3) Smaller businesses are using specific technologies such as Eclipse
SWT/RCP, flash for a full website, and also need people for custom (old)
technologies like windows sockets programming and such.

   I think most of us are looking for a better stack. The traditional J2EE
process is not highly productive. Though alot better than the alternatives
I've seen so far. The benefit of web programming is it is very flexible and
incremental. The handful of large companies I've consulted for simply pick a
large J2EE vendor and throw the team into it. The problem with FOSS stacks
is there are so many. Call it 1000 blooming flowers but it really looks like
engineers plagued by NIH (not invented here) syndrome. Alot of the Apache
projects simply re-invent the wheel rather than "Build on the Work of
Others" (old Unix design guideline). Sun is at least as bad about this.
Consider Java logging for instance. When everyone already was using and
already does use Apache logging. NIH leads to re-evolution through the same
painful experiences everyone experienced before with the predecessors. E.g.
how many darn data controls do we have to have before someone decides to
actually re-use a technology!!!!!! At least with OpenSource one certainly
COULD reuse the technology.

  IMO even if you have the most productive framework in the world there are
so many frameworks that yours may be lost in the mix. Add in the monopoly
marketing engine plus the cost of labor overseas and things appear
differently. E.g. not too many people in Burundi know about Velocity and DWR
but AJAX (at least MSXMLHTTP and client-server interaction via
applets/ActiveX) is around a decade old and J2EE has been around a looong
time. At $90 per year a company can hire a thousand programmer in Burundi
for the cost of one expert in the USA. At least this is what managers often
reason.

  I'm being sarcastic of course. The $90/year is of course total BS because
most of these people are subsistence farmers (aka conserver culture) and
would require at least ten thousand dollars of infrastructure each and have
no programming experience. And I totally believe the best tools lead to
quality products and the best ROI. IMO, most of the stacks focus on
re-invention and not much on computer science. Most people in the industry
are well-educated and fairly cognizant of what is being done.

   What is weird/sad/funny to me is the invention of new programming
languages. Frustrating because the programming language has never been the
limiting factor in any project I've done in my career. GUI builders,
protocol libraries, format converters, integration support, and such are
always missing from a new language. I guess I'm talking about Python, Ruby,
PHP, and such. Even ANT and other psuedo-programming frameworks. I mean,
very complex builds were being done with shell for many decades. So what if
Windows didn't have a decent shell? Get an addon. I really believe the NIH
problem is a severe drag on our industry and makes us look unprofessional. I
mean, how many many hours were wasted studying XSLT and that is something
I've never seen used in all of its "glory". One reason I quit the Microsoft
world was the total adware frameworks they came out with every few years.
About the time you'd integrate these into your application they'd yank the
rug and move on to promoting something else. I try to stick to mainstream
technologies because who even knows how to implement OLE, behaviors, or that
distributed GUI framework thing Gates promoted in the mid-1990s? At least
with mainstream the customer has a chance of getting a few decades from
their investment. The framework proliferation is leading to something like
spaghetti code. Or call it a ball of wax/mud for the industry as a whole.

   In the end, none of the frameworks actually solve the customer's problem.
It takes custom programming for most large projects. The language and the
framework are just choices of how to skin the cat.

My $.02,
TimJowers


On 6/27/07, Shawn Hartsock <[email protected]> wrote:
>
> I'm currently working on a new development project and I'm curious to know
> how unique or working environment is to this group.
>
> Our application is on the JBoss Application Server version 4.0.5 with
> their implementation of EJB3. I've gotten EJB3 soap services and persistence
> running with the Liferay 4.2 portal in the same server instance.
>
> All our business logic is exposed as SOAP services to be used by our
> Portlets (or other applications). We write our portlet views in Velocity. We
> do our server-side Ajax calls in DWR 2.0 (the AOP version). We also do our
> business processes in jBPM.
>
> So how unique is this configuration in the triangle? In the world?
>
> --
> --Shawn Hartsock
> _______________________________________________
> Juglist mailing list
> [email protected]
> http://trijug.org/mailman/listinfo/juglist_trijug.org
>
>

_______________________________________________
Juglist mailing list
[email protected]
http://trijug.org/mailman/listinfo/juglist_trijug.org