299 and JHoller presentation

[asookazian]

I posted here:

http://www.infoq.com/presentations/Spring-and-Java-EE-6-Jurgen-Holler

I'm wondering if PMuir and/or GKing will go hostile about some of his comments regarding JCDI toward the end of the presentation.

He made it sound like 299 will possibly not make it into EE 6 spec or it may merge into EJB 3.1 spec (or there's such discussion perhaps).

I though GKing stated that JCDI is applicable in both SE and EE envmts so why would it merge into EJB 3.1?

[arshadm]

Couldn't hear the talk because of rubbish css and web page design (firefox and ie8), so can't comment directly.

I suppose it makes sense to merge JCDI into EJB3.1, I mean there are only a couple of things that JCDI adds which are of real value (ability to access ejb session beans from the el and conversation semantics), all the rest of the stuff I can't see anybody using regularly.

The bigger question is what is the point of Spring? I mean it was useful when EJB 2.0 made a complete cockup of providing a useable framework for Enterprise development, but who wants to use JdbcTemplate when they can using JPA 2.0, or use FreeMarker when JSF 2.0 is available.

Regards.

[asookazian]

I mean it was useful when EJB 2.0 made a complete cockup of providing a useable framework for Enterprise development, but who wants to use JdbcTemplate when they can using JPA 2.0, or use FreeMarker when JSF 2.0 is available.

Lots of shops still use stored procs.  So I can see JdbcTemplate and related interfaces/classes being useful in those scenarios.  But you'd hope they'd start using JPA 1.0/2.0 with Hibernate as persistence provider to use manual flush and 2nd level cache, for example.

Spring may not seem as relevant technologically speaking as they were in 2004 with Spring/Hibernate apps taking over the J2EE space.  But the fact is that there are lots of Spring deployments, lots of Spring/Hibernate/Tomcat shops, and lots of momentum behing their technology, services, and stack.  AOP seems like overkill to me with the terminology and advanced concepts that seemed to be solved using JEE interceptors.

Even the Spring Tools Suite IDE is free whereas JBoss is charging $99 for JBDS (which you can get for free if you download Eclipse and JBoss Tools separately).  And I'm also interested in the Spring dm server and seeing how well it does in terms of adoption.

Unfortunately, from what I've read and experienced, I'd much rather stick with Seam 2/3 but there are so few shops out there using Seam that it's like you have to learn Spring eventually if you want to continue doing JEE development.

Hopefully, 299 and JEE 6 will change things...  But one of the main problems is the JCP ecosystem itself and the lack of leadership and coordination amongst the disparate EGs.  Just think about JSF 1.0 and EJB 3.0 in EE 5 and how the hole or disconnect there should have been properly plugged and thus obviating the need for an integration framework like Seam...

[nickarls]

To summarize my comment on InfoQ: FUD ;-)

As I see it, the component model is in a transition state that will be merged into the Managed Bean concept that will be fully integrated (EJB:s, JSF managed beans, 299 beans etc) in Java EE 7

[nickarls]

I'd like to add that I think that competition is good and the protest movement of Spring did wonders for the usability/simplicity of EE5+.

I also understand that they are holding on to the FUD of the heavyweight-container etc. After all, they have a commercial interest in keeping it so.

It is also unfortunate (from our point of view) that SpringSource has the budget to market themselves much more heavily than Seam.

[asookazian]

Nicklas Karlsson wrote on Aug 06, 2009 08:58:

It is also unfortunate (from our point of view) that SpringSource has the budget to market themselves much more heavily than Seam.

Very good point.  And that is a major disadvantage (in addition to the fact that SpringSource - Interface21 at the time - already had a much larger market share in JEE than when Seam entered the Java space in 2006).

Root cause: most likely Redhat.  I read that Marc Fleury resigned as CTO a couple years ago partly because he was unhappy with the amount of funding that was being directed from Redhat to JBoss for R&D.

I mentioned earlier that Im a little bit disappointed that there is still no significant investment in the R&D division. We invested in sales, support and marketing, but this is really the operational side. The R&D really hasnt benefited from a huge investment for which I was hoping and was the main reason I went to Red Hat.

http://www.eweek.com/c/a/Application-Development/Red-Hat-From-Cuddly-Penguin-to-Public-Enemy-No-1/

Not sure if that's still true or not...  And money for R&D is different from money for marketing/sales/training, etc.

[pmuir]

I posted here:

http://www.infoq.com/presentations/Spring-and-Java-EE-6-Jurgen-Holler

I'm wondering if PMuir and/or GKing will go hostile about some of his comments regarding JCDI toward the end of the presentation.

I'll try to find time to listen to it this - shame there isn't a transcript ;-) I talked to Juergen about where they were going with CDI and Spring back in the spring (no pun intended). It sounds like their position hasn't changed - watch from the sidelines.

He made it sound like 299 will possibly not make it into EE 6 spec or it may merge into EJB 3.1 spec (or there's such discussion perhaps).

I though GKing stated that JCDI is applicable in both SE and EE envmts so why would it merge into EJB 3.1?

Probably this is a publishing timelag - parts of CDI and EJB 3.1 were refactored to depend on two common specs - the Interceptors spec and the Managed Beans spec (see the latest drafts of EJB 3.1 and EE 6 on the jcp site, respectively). Anyway, it looks pretty definite now that CDI will be in EE6 - I haven't seen anyone dispute this on the EE expert group for at least 6 months now.