I am standing and applauding Hu on his insightful perspective of EJB
development.  Finally, someone is also seeing that the emperor is not
wearing any clothes (well only some).

> -----Original Message-----
> From: A mailing list for Enterprise JavaBeans development [mailto:EJB-
> [EMAIL PROTECTED]] On Behalf Of Hu Shih
> Sent: Wednesday, January 30, 2002 3:49 PM
> To: [EMAIL PROTECTED]
> Subject: Re: EJB Development Cycle
>
> Perhaps you have lot of experience working with EJB on a large
project,
> but
> I beg to differ with much of what you said.
>
> I'm not a great fan of MS, but you can compare what's happening with
EJB
> with MFC.  At least with MFC, you were given a set of library classes
that
> performed crucial work for you and the code was available.  If
something
> goes wrong, you debug the framework code along with your code.
>
> Also, until you complete your test in your development environment,
you
> don't have to build an installation of your application.  EJB's
deployment
> requirement is like requiring people to build installation disk.  And,
you
> must install the system to have the other half the system be
generated.
> How
> can you call this productive or conducive to iterative development?
>
> My complaint with EJB is that much of "what the system does for you"
is
> generated during deployment time and is hidden away from the
developer.
>
> This is certainly the case with Sun's reference EJB server
implementation
> which I've been working with.  Perhaps it's a bit better with open
source
> stuff like JBOSS.  But the fact remains: deployment/installation must
be
> done before you can test your modules with the rest of the system.
The
> other half ("better half") doesn't even come into existence until you
> deploy.
>
>
> >compiling/deployment for you. And depending on how you design your
> >application you often time don't have to redeploy your beans at all
> unless
> >some major interface changes. It could also be that you are not unit
> >testing
> >your code enough. If you are waiting to discover bugs only when you
> deploy
> >your stuff I think it's a little too late to complain.
>
> How do I unit test a CMP bean before it's deployed?  Could you share
some
> concrete ideas based on your experience, instead of empty talk?
>
>
>
> >From: "Tinou Bao" <[EMAIL PROTECTED]>
> >To: "Hu Shih" <[EMAIL PROTECTED]>, <[EMAIL PROTECTED]>
> >Subject: Re:      [EJB-INT] EJB Development Cycle
> >Date: Wed, 30 Jan 2002 15:44:54 -0500
> >
> >I don't think you can simply blame EJBs for your development aches.
EJBs
> >provide you with several features, such as security, transaction,
> >persistence, distributed. Naturally the container needs to generate
> classes
> >to do this work for you. Otherwise you'll endup writing this code for
> >yourself. So, do you want to have a lot of classes generated by the
> >container or a lot of classes your developers code? If you don't need
> these
> >features then don't use EJBs. Regarding deployment time. Yes, EJBs
> require
> >an extra step but this should in general not cause you all the pains
you
> >speak of. You can write scripts that pretty much does all the
> >compiling/deployment for you. And depending on how you design your
> >application you often time don't have to redeploy your beans at all
> unless
> >some major interface changes. It could also be that you are not unit
> >testing
> >your code enough. If you are waiting to discover bugs only when you
> deploy
> >your stuff I think it's a little too late to complain.
> >
> >_________________________________________
> >
> >Tinou Bao
> >BAO Systems
> >Chairman of the Board and Chief Software Architect
> >www.baosys.com
> >
> >----- Original Message -----
> >From: "Hu Shih" <[EMAIL PROTECTED]>
> >To: <[EMAIL PROTECTED]>
> >Sent: Wednesday, January 30, 2002 10:08 AM
> >Subject: [EJB-INT] EJB Development Cycle
> >
> >
> > > Please forgive my frustrated tone, but
> > >
> > > I've been working with EJB (CMP beans with session facades) using
> Sun's
> > > J2SDKEE.  I find the development cycle extremely tiring and slow.
I
> >write
> > > bean code.  I go to their deploytool to deploy it.  I run the
thing.
> >Find
> > > problems and come back to my development environment.
> > >
> > > I understand some IDEs will make this cycle easier.  Still, I've
been
> >used
> > > to developing and testing code quickly in an iterative cycle.
After
> >many
> >of
> > > these iterations, I deploy to user environment--many
> >code/compile/test-runs,
> > > very few deployments.
> > >
> > > One of the big advantages of a quasi-intrepretive language like
Java
> is
> > > precisely that the repetitive code/compile/test-run cycle is
quick.
> > >
> > > EJB changed all this.  They stuck a rather long, tedious, and
often
> >painful
> > > deployment phase right between compile and test-run, often
breaking
> the
> > > development cycle with 10 to 20 minute breaks.
> > >
> > > Debugging such a deployed app is a nightmare.  Because EJB
(especially
> >CMP)
> > > generate so many classes and because so many of these are system
> >generated
> > > stuff, I have very little idea what's going on or what I have done
> >wrong.
> > >
> > > Also, so much information that are important to developers are
hidden
> >away
> > > from them in multi-level jars/ears/wars, etc. (and these things
are
> > > humongous).  And, why so many classes?  When running, I see over 4
> >classes
> > > generated for each EJB bean.  This is excess and debugging
nightmare!
> > >
> > > With Java, the trickiest configuration parameter was CLASSPATH.
With
> >EJB,
> >I
> > > have to know and worry about so many of these configurations, I
feel
> >like
> >I
> > > need a dictionary of them.
> > >
> > > What's going on?  It's almost as though EJB put Java back to the
level
> >of
> > > C++ and C++ templates.  I don't know about others, but I
generalize
> >dislike
> > > and dispise the condescending attitude of any system that tells
> >developers:
> > > "Don't worry.  We'll take care of you by generating lots of stuff
> under
> >the
> > > covers.  Why would you care about long breaks in the development
> cycle?
> >Go
> > > take a coffee break."
> > >
> > > Developers are an impatient and controlling bunch.  Java has been
good
> > > because it gives speed of development and gives enough control for
> > > developers.  In my opinion, EJB is going backwards (the wrong
> >direction).
> > > Is there an effort to address these?  Or, is it that to be an EJB
> >developer,
> > > you have to take all this willingly and gladly?
> > >
> > >
> > > _________________________________________________________________
> > > MSN Photos is the easiest way to share and print your photos:
> > > http://photos.msn.com/support/worldwide.aspx
> > >
> > >
>
>=======================================================================
==
> ==
> > > To 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".
> > >
> > >
>
>
>
>
> _________________________________________________________________
> MSN Photos is the easiest way to share and print your photos:
> http://photos.msn.com/support/worldwide.aspx
>
>
========================================================================
==
> =
> To 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".

===========================================================================
To 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".

Reply via email to