RE: [Springframework-developer] TSS Symposium

[<jue...@we...>]

Rod, everyone,

I'd also like to say that the TSS Symposium was a really great event. I =
enjoyed it lots more than JavaOne 2001, due to the focussed contents, =
and due to all the people that you could meet there in such a =
"comfortable" setting.

I especially enjoyed talking to other "open source guys" like Gavin, =
Vincent, Mike, Jason, etc. The outdoor discussion with John from Sun and =
Jim from IBM was very interesting too, lasting till after midnight, =
despite all those mosquitoes out there (I got more than 50 stitches...).

Of course, it was great to finally meet Rod in person! You can only talk =
so much via email, personal chats are a wholly different thing. We can =
really be glad that we have such a prominent and compentent spokesperson =
:-)

I agree with the main themes throughout the symposium: It's funny that =
they match our views and our mission so nicely! The trends are =
definitely going in our direction, in every respect.

There really seems to be a great Hibernate vibe. Everyone's mentioning =
it, numerous people seems to use it in production. Gavin King is a smart =
guy: He has achieved a really clean persistence tool that beats many =
other solutions on the market. In his own words, Hibernate has more =
users than all of the JDO vendors together, including Fortune 100 ones. =
Now consider that he works on Hibernate besides a full time job!

The PetStore-based MDA presentation with its "35% less time" result was =
pretty meaningless. ~530 man hours for the "conventional" J2EE team vs =
~350 for the OptimalJ team - that's too much for both ;-) If one would =
throw out EJB completely and built upon Spring and Hibernate, one would =
be able to implement a clear-layered version of that PetStore spec in a =
maximum of 200 hours, I guess. Let's not forget that the HTML templates =
and the data model are predetermined!

I haven't been convinced by JBoss 4's AOP support either. I really doubt =
that field interception via the classloader is the right way. For =
container internals like dirty checking of HTTP sessions, it might be =
useful though - but even to achieve just this, each and every class that =
might potentially be put into a session needs to be enhanced. I feel =
that this can't be the way, neither in terms of performance nor of =
performance.

I really like our AOP strategy: Just intercept methods of specific =
instances, not everything of any class loaded in the current =
classloader. Of course, this isn't as powerful as "full" AOP on fields =
etc - but the latter is power that I don't want anyway. We can achieve =
EJB-style functionality like declarative transactions and security =
easily, that seems more than enough to me for now.

I'll definitely go to another TSS Symposium if there will be one, =
possibly as speaker. In one year's time, there should be more than =
enough stuff to talk about :-)

Regards,
Juergen


-----Original Message-----
From: Rod Johnson [mailto:rod...@in...]
Sent: Monday, June 30, 2003 3:33 PM
To: spr...@li...
Subject: Re: [Springframework-developer] TSS Symposium


I'd just like to say that this was a great event. I really enjoyed it, I
felt people were really receptive to what I was saying--including Spring
JDBC, transactions, AOP CMT, bean factory concept.

I also got to meet and talk to some really interesting people like =
Cedric,
Floyd, Mike Cannon-Brooks and Gavin King. And even got to have a serious
conversation with John Crupi.

This was a great opportunity for the Spring community. Now my book is =
back
in circulation, we should also benefit from that. Sure, there are code
changes (and many enhancements) but explaining Spring over and over to
different people made both Juergen and myself aware just how simple and
consistent the overall concepts are--these haven't changed, which is one
reason that we now have a lot of code but not necessarily more =
complexity in
Spring apps.

Btw, it was great to meet Juergen! He did a lot of good work as an
ambassador for Spring, as well as picking up other talks. He even won a
group of people at a table, including me, a Tivo each in a trivia game!!
(Don't know if it will work in the UK yet.)

Maybe I should start a blog. Here's my brief summary of some of the =
content
I liked/didn't like:

- John Crupi's talk on next gen tools--not convinced Sun really "get" =
this
space
- Bill on JBoss AOP: not a great presentation (a bit flat), didn't =
really
justify their decision to use field interception or address performance =
and
other potential problems. Did show some nice examples of JBoss internals
implemented with AOP. I thought Bill made a good case for implementing a
J2EE server with AOP, rather than opening their AOP up to developers to =
lock
them into JBoss. Bill is charming and very normal, whatever impression =
JBoss
gives overall.
- Mike Cannon-Brooks: really good speaker (but then he would be, being a
fellow Aussie). XWork presentation was interesting--I like our way of =
doing
things, but XWork is elegant in its own way. (We have lots of stuff they
don't, though, such as JDBC, JTA etc.)
- MDA presentation based on Pet Store. I'm just going to say =
"unconvincing"
- Erik Hatcher: Excellent speaker (and nice guy), wish I could have =
caught
all his talks, but it was hard when I was presenting/networking. Very
pragmatic and wise regarding Struts limitations.
- Panel discussion: really interesting. Basically the "status quo" vs
"reduce complexity" schools coming into conflict. The audie- nce laughed
when Jim Knutson (from IBM) try to defend entity beans after I'd used =
them
as an example of design by committee before implementation not working.

I detected some themes throughout, such as:

- Entity beans suck (not just from yours truly). Vibe around Hibernate.
- Struts is sub-optimal. Even Erik Hatcher (Struts/XDoclet contributor =
and
guru) knows of the problems and is open to XWork/Spring. Maybe Struts =
really
is so bad that it can't even survive as a virtual monopoly.
- EJB is overused (my talks on this were very well received)
- AOP
- A real open source vibe. It was great to talk to Gavin, Mike & Jason
(XWork etc.). We're all doing great stuff and we're all pushing in the =
same
direction, if with different code bases.

It would be good to have Juergen's take, too.

Regards,
Rod

P.S. If they do another TSS symposium, try to go! It was good--way =
better
than JavaOne.




-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100006ave/direct;at.asp_061203_01/01
_______________________________________________
Springframework-developer mailing list
Spr...@li...
https://lists.sourceforge.net/lists/listinfo/springframework-developer