IBM has announced the expected availability in May 2006 of Websphere 6.1, their flagship J2EE container, supporting J2EE 1.4 and J2SE 5.0 and focusing on development support, providing easier installation, integration with IBM's Rational suite of tools, and integration with IBM's portal server.

In addition, IBM includes a set of JSF widgets and management services, along with various security enhancements.

The timing indicates that Websphere will be released around JavaOne, along with many other expected product releases around that time.

Good news for the J2EE community. Now, every server supports J2SE 1.5.

Wei Jiang
Acelet
Perfecting Java EE!

Good news for the J2EE community. Now, every server supports J2SE 1.5.Wei JiangAceletPerfecting Java EE!

As a duty bound J2EE community member, I must say, we the members of the J2EE community: don't care!

Look it is official. I took a vote. We just don't care.

Oh were they quick to support the spec! Again...

regards,

Messi

or do we use tiles here?
.V

IBM was one of slowest company to adhere to the J2EE specs...

Evidently something is changed and now WebSphere has some interesting features such as Scheduler and Warm Deploying to deploy (in cluster fashion) only the diff of the application without stop anything ;-)

bye,
Luca Garulli
Blogging on: http://zion-city.blogspot.com
http://RomaFramework.xwiki.com - The new way to build applications
http://www.Pro-Netics.com (member of Orixo.com - The XML business alliance)
http://www.OrienTechnologies.com - Light ODBMS, All in one JDO solution

IBM was one of slowest company to adhere to the J2EE specs...

Sorry but I don't understand your quote: WebSphere supports J2EE 1.4 since version 6.0 (released during December 2004).
This new version adds support to J2SE 1.5 but it's not required by J2EE 1.4 specs. New J2EE 1.5 specs are again under development.

Evidently something is changed and now WebSphere has some interesting features such as Scheduler and Warm Deploying to deploy (in cluster fashion only the diff of the application without stop anything ;-)

Again, you talk about features already available in previous versions.
In WAS 6.1 there are instead many welcomed features like SPNEGO support, SIP servlet and JSR168 support.

Ciao, Diego

I noticed the announcement says:
<quote>
JSR 168 — enables interoperability among portlets and portals. By allowing portlets to be accessed like servlets, via a URL, support for the JSR 168 programming model enables you to run portlets without a full Portal Server.
</quote>

Will there be a portal page that the WebSphere 6.1 portlet sits in or will you need to include all the HTML page markup in a portlet? Will portlet modes and window states be supported? How will Action and Render URLs be handled?

1. What about J2SE 5.0 support for WebSphere 5.1?
2. What about EJB3 support in WebSphere?
3. Will be WebSphere 7 Geronimo?

Is it 1 so difficult? with 1 the customers can upgrade to EJB3 without an exp€n$iv€ (in licence terms) migration.

Javier Paniza
OpenXava: Application Framework with WebSphere support

I hope IBM hired some user interface guys and fixed the console.

WAS 5.1 had the most unintuitive unmanageable console ever.

How can you go through 10 screens to configure a jdbc connection?

and Global Security... fine, but why not give Console security only instead of messing with app resources and having to use J2C when you just want to protect the console?

and btw, you ALWAYS want a clean redeploy, because you want to make sure the right code is running.

I hope IBM hired some user interface guys and fixed the console.

WAS 5.1 had the most unintuitive unmanageable console ever.

How can you go through 10 screens to configure a jdbc connection? and Global Security... fine, but why not give Console security only instead of messing with app resources and having to use J2C when you just want to protect the console?and btw, you ALWAYS want a clean redeploy, because you want to make sure the right code is running.

Off the top of my head, to configure a jdbc datasource it takes 7 pages for minimal configuration and 9+ to add other configuration parameters.

peter

Agreed that the gui is a bit cumbersome, but in a real production environment, you should not rely on the gui.

So what if it takes 7 screens to config a JDBC resource, i would advocate 2-3 lines of jython and automate (which you should/would have to do).

Also, hot deploy is overrated in prod enviroments, great for dev though

I agree that you can use scripting.
Some customers I know, though, won't allow it because they are more difficult to check and see what you want to configure in their systems.

And it does not solve the main problem of checking configuration, you still have to go through the screen to see what is the current situation.

My company migrated 30 web applications from Websphere 4.0 to Websphere 6.0.2.3

Websphere 6.0 has been a significant improvement over Websphere 4.0. Our web administrators are happy with the improved admin console and utilities.

The biggest disappointment has been IBM Rational Application Developer 6.0 The performance in this IDE is so bad that our team refuses to adopt it. We installed all of IBM's fixpacks but the performance was not any better. We've decided to stick with Websphere Studio 5.1.2 until IBM can improve the RAD 6 IDE.

J2SE 5 sounds great but what our team really needs is a productive development environment. I want IBM to focus their energy on improving the RAD6 product.

The biggest disappointment has been IBM Rational Application Developer 6.0 The performance in this IDE is so bad that our team refuses to adopt it. We installed all of IBM's fixpacks but the performance was not any better. We've decided to stick with Websphere Studio 5.1.2 until IBM can improve the RAD 6 IDE.J2SE 5 sounds great but what our team really needs is a productive development environment. I want IBM to focus their energy on improving the RAD6 product.

+1

RAD 6 runs ok for me and others. Having 2 gig. memory is ideal but nt essential.

RAD 7 is due out later this year.......

Would nt it be nice if TSS started a new item that asked why does RAD and / or WS suck so much ? Why are they both or individually such poor products. Buggy, slow, unstable and all sorts of other attributes that someone with Rational and their RUP methodologies and experts should surely have fixed by now.

IBM if your listening stop worrying about new standards and fix your old product line so its actually usable and not a complete waste of time and pain in the ####.

Wouldn't it be nice if TSS started a new item that asked why does RAD and / or WS suck so much ? Why are they both or individually such poor products.

So IBM consultants can come to your rescue, of course.

Its a wonder how or why IBM craft RAD to be so unusable. Do they have an internal team that takes a reasonable IDE and then tweeks its so its utter crap ?

Then again the only requirement to work for IBM is too have a suit, sit at your desk, produce lots of useless diagrams and maybe just maybe they might actually get something done. Then again consultants only exist to suck your money and keep their job by talking to dull managers who havent a clue.

:)

I thought that was what the fixpacks where for, it really scares you when you see 1GB fixpacks. Even more hilerious when you have to ask system admins to download it for you cause of company restrictions on bandwith and they give you the nickname 'gibabyte boy'......

My fav regression is the server console which worked flawlessly in 5.1.2, but has regressed to the most infuriating piece of s###. On rare occasions it works, but mostly it doesn't work for days on end, and then when it does suddenly want to work, it has to catchup with itself. Its amazing to watch 2-3 day old log entries appear in the console. This also is a trap for young players due to the confusion it creates, imagine trying to debug issues and seeing exceptions that are 3 days old....

Its amazing to watch 2-3 day old log entries appear in the console.

LOL oh yeah i almost forgot about the logging. I pulled my hair out wondering where the exceptions came from, after a while i saw that they were from the day before. I feel very sorry for the guys that will have to work with version 7, that will propbably be worse than 6 as 6 is worse than 5. What are those people doing at IBM? They dont give a **** about developers. Another story is that one of my colleguas was so courages to decompile code and fixed bugs in RAD. When he submitted the bug fix he was told that decompiling and fixing their filthy code was an illegal action and he was forbidden to do so. My only question is why they can get away with all this. You can fool some people sometimes but you cant fool all the people all the time. When you buy RAD you are fooled big time!

I thought that was what the fixpacks where for, it really scares you when you see 1GB fixpacks.

+1

Would nt it be nice if TSS started a new item that asked why does RAD and / or WS suck so much ? Why are they both or individually such poor products. Buggy, slow, unstable and all sorts of other attributes that someone with Rational and their RUP methodologies and experts should surely have fixed by now.

IBM if your listening stop worrying about new standards and fix your old product line so its actually usable and not a complete waste of time and pain in the ####.

+1

The biggest disappointment has been IBM Rational Application Developer 6.0 The performance in this IDE is so bad that our team refuses to adopt it. We installed all of IBM's fixpacks but the performance was not any better. We've decided to stick with Websphere Studio 5.1.2 until IBM can improve the RAD 6 IDE.J2SE 5 sounds great but what our team really needs is a productive development environment. I want IBM to focus their energy on improving the RAD6 product.

+1

+1000

RAD crap has made my working days nighmarish the last 2 years. Since 6 months i am on bare eclipse and i feel relieved. When I smell RAD i run away as hard as i can, it must be the most unproductive buggy environment ever made. I'd rather clean shit houses with my bare hands than to work with that filth.

My company migrated 30 web applications from Websphere 4.0 to Websphere 6.0.2.3Websphere 6.0 has been a significant improvement over Websphere 4.0. Our web administrators are happy with the improved admin console and utilities.The biggest disappointment has been IBM Rational Application Developer 6.0 The performance in this IDE is so bad that our team refuses to adopt it. We installed all of IBM's fixpacks but the performance was not any better. We've decided to stick with Websphere Studio 5.1.2 until IBM can improve the RAD 6 IDE.J2SE 5 sounds great but what our team really needs is a productive development environment. I want IBM to focus their energy on improving the RAD6 product.

ditto on the +1

I spent some time looking into this. It looks like in addition to a bunch of Rational's half-finished/half-tested code, the fact that the product is based on Eclipse 3.0 is a big part of the problem. Moving to Eclipse 3.1 or 3.2 will solve these problems, and Java 5 support comes along for the ride.

We've adopted the 6.0 release, with a set of guidelines on how to not poke the bear, so to speak. :-)

The biggest disappointment has been IBM Rational Application Developer 6.0 The performance in this IDE is so bad that our team refuses to adopt it. We installed all of IBM's fixpacks but the performance was not any better. We've decided to stick with Websphere Studio 5.1.2 until IBM can improve the RAD 6 IDE.J2SE 5 sounds great but what our team really needs is a productive development environment. I want IBM to focus their energy on improving the RAD6 product.

RAD 6 is great if you enjoy spending valuable developer resource hours staring at an hour glass cursor. I found RAD 6 the most productive when I disabled nearly all of its features than at that point I might as well just use Eclipse with a few plugins.

And, yes I have 2 gigs of memory.

I convinced the team to switch to Tomcat and only use WebShpere and RAD 6 when we have to, which is rare.

Some of use use Eclipse and some of us use IntelliJ, but almost none of use choose RAD 6, which was our standard dev env.

I've used a lot of app servers and java development environments... RAD 6 and WebSphere is one of the slowest I've used.

I hate waiting a few minutes to see the "Open for eBusiness message".

Speed! I want speed.

On another note, EJB3 is retarded and RoR is a joke (see blog entries for more details).

--Hadji Murad
King of all Java Media

Selected Readings:

Boycott EJB3: http://www.jroller.com/page/killjoy?entry=boycott_ejb3


RoR is a Square Turd: http://www.jroller.com/page/killjoy?entry=ruby_on_rails_is_a

I haven't found RAD too bad though I have 2GB RAM and a 2.6Ghz Intell to run it on. I have mine updated to 6.0.1.1 with a WASv6.0.2.5 server.
One of the unfortunate default WASv6 server settings is to have the performance monitoring infrastructure switched on. This caused a long delay (>1Min) during server startup where nothing was happening. Switching this off made my server startup in about 20-22 seconds with my application installed.
Look in the WAS console under PMI and see if turning it off improves your situation too.

Cheers

6.0.1.1 was focused on improving performance, according to IBM. That might explain your somewhat better experience.

' On another note, EJB3 is retarded and RoR is a joke (see blog entries for more details...'

    I have a question about the viability of Spring in Enterprise applications. From what I have been able to determine, the standard Spring interfaces are not clusterable. This would be a concern to me in building an application for failover purposes. I understand that BEA has extended the basic Spring interface so that they can be clustered, http://dev2dev.bea.com/lpt/a/448 , 'With the certification of Spring 1.2.5 on WebLogic Server 9.0, we have extended the JndiRmiProxyFactoryBean and associated service exporter—so that it supports POJO proxying with any J2EE RMI implementation, including RMI-IIOP and T3. Included with this support is a WebLogic RMI deployment descriptor that enables clustering on the proxy RMI interface, so POJO calls can be load-balanced across a WebLogic Server cluster.'
    If the Spring interfaces cannot be clustered and you are not using Weblogic, then I would think this would give significant weight to using EJB 3.0 over Spring. My question to someone implementing Spring is has this been an issue to you and if so, how have you addressed it?

Thanks,


Michael Butler

First of all, those changes mentioned in the BEA announcement have been made to Spring core. The standard Spring distribution contains all of this since 1.2.5.

Clustering is indeed not built into Spring's core programming model, and we believe it shouldn't in the first place. The application should concentrate on business logic, without much consideration of infrastructure concerns like clustering. However, that does not stop Spring for leveraging a clustered environment.

Clustering typically sits at the entry point level, not within the application. For example, many applications cluster at the HTTP level, which Spring supports through leveraging the standard Servlet API. That model scales very nicely and does not impose artificial burdens on the application's business services.

An alternative clustering scenario is a clustered cache. This works nicely with Spring as well, since clustered cache APIs usually are as simple as a Map interface and can easily be used with Spring's dependency injection facilities, that is, injected into every business service that needs access to the cache.

We use the notion of "remote exporters" to expose a Spring-managed service to remote clients. Among the supported protocols are HTTP-based ones like Hessian and HTTP invoker as well as any RMI-based protocol like JRMP, IIOP and T3. Spring provides a POJO service model for all of those, in a very consistent fashion.

All of the supported remoting protocols work nicely in a J2EE 1.3+ environment. On some servers, the application's deployment descriptor needs to explicitly activate RMI-based clustering (just like it needs to do for EJB-based clustering), which is what we ship out-of-the-box for WebLogic (as a convenience).

Alternatively, I would recommend to use HTTP-based clustering even within a server environment. Modern HTTP-based remoting protocols are efficient, extremely easy to set up, and able to transparently leverage clustering (plus encryption etc) at the HTTP level. This is a common scenario nowadays, in particular in the Spring world.

FWIW, you can of course use Spring in combination with EJB 3.0 Session Beans and/or MDBs as well. And of course, you can use the Java Persistence API (aka "EJB3 persistence") with Spring - as effectively yet another O/R mapping strategy within Spring's programming model. So this is by no means an "all or nothing" choice.

Juergen

Juergen,

   Thank you for your informative reply. HTTP clustering is a viable option for many applications but not all as you point out. I have several more questions for you along the same vein as the interfaces. I will follow up with you on the Spring site. The answers to those questions, like the interfaces, would be very helpful in evaluating Spring as a solution.

Thanks,


Michael Butler

The biggest disappointment has been IBM Rational Application Developer 6.0 The performance in this IDE is so bad that our team refuses to adopt it. We installed all of IBM's fixpacks but the performance was not any better. We've decided to stick with Websphere Studio 5.1.2 until IBM can improve the RAD 6 IDE.J2SE 5 sounds great but what our team really needs is a productive development environment. I want IBM to focus their energy on improving the RAD6 product.

I absolutely agree. RAD6 is a nightmare. It's slow, resource-cost, and unstable.
There is too much garbage inside. WSAD5.1.2 is much better and much faster.
It's bad to rename it to Rational. Everything coming from Rational means complex, resource-costing, slow, and garbage.

RAD6 is also buggy and resource intensive from my experience (with 2 GB), a real backward step from 5.1.2 which quite good. I used all versions since 4 and rad 6 is neatly as bad as that release (Maybe its something about the even numbered releases).

We had to migrate to RAD6 to use J2EE 1.4 features like messaging and JSF. We used it since 6.0 base release which could be best descriped as completely unusable.

The biggest loss has been the ability to cvs our server configs. Each developer has to spend hours setting up a server and we have to maintain a document for each project to document the settings. Previously a senior developer who understood the config would administer this and sync it with cvs and everyone could simply catchup.

Starting and stopping server has become interesting with the new profile feature, if you miss configure the launcher then you can easily create orphaned processes.

Add to that the debugger dies frequently and seems to stop on noexistant breakpoints, which drives you insane.

On the server side we discovered that the JSF implementation is bundled with the WAS libraries, making it next to impossible to use another implementation like myfaces, without jumping thru hoops to alter the classloading settings, and playing with this always brings its own swag of issues.

There are some good new features like the ability to use a decent messaging engine in WAS 6, but I cannot find many good feature in RAD 6..... imho

Steve

RAD6 is also buggy and resource intensive from my experience (with 2 GB), a real backward step from 5.1.2 which quite good.

+1

The biggest loss has been the ability to cvs our server configs.

Agreed! We are exporting the profile as a car file and sticking that in CVS. We then use cvsmail to send an e-mail to all the developers to pull in the change and import it. This is still a pain though as it is easy to forget to import the car file. Seems like something that the IDE could do behind the scenes.

The biggest loss has been the ability to cvs our server configs. Each developer has to spend hours setting up a server and we have to maintain a document for each project to document the settings. Previously a senior developer who understood the config would administer this and sync it with cvs and everyone could simply catchup.

That one is actually fixable, though why the new behaviour is the default is beyond me.

1. Open Window -> Preferences
2. Select the "Server" category
3. Tick the "Create server resources in workspace" box
4. Create a new server definition

Otherwise in 100% agreement about RAD. The generall slowness is overwhelming, but I also miss a bunch of features from Eclipse 3.2. Another pet peeve: its habit of changing stuff in .classpath, .j2ee (what's that for anyway?) and .websettings (and that?); they're always polluting the CVS view.

In all fairness... the deployment descriptor editors and things like that are very useful, oftentimes a lot nicer than editing the XML by hand, e.g. automatically gathering the security roles of all child applications into application.xml (most/many of the same editors are available in WSAD 5.x). The XML editor is a pig, but that's symptomatic for Eclipse in general, not specific to RAD. I tend to use Cooktop and/or UltraEdit there. My primary issue with RAD are the various validation builders that for no apparent reason take minutes to complete. Nothing quite like waiting five minutes for a "tacit migration builder" to complete in between a CVS Commit and a CVS Tag.

The biggest loss has been the ability to cvs our server configs. Each developer has to spend hours setting up a server and we have to maintain a document for each project to document the settings. Previously a senior developer who understood the config would administer this and sync it with cvs and everyone could simply catchup.

Our team shares your disappointment. In WSAD 5, we could store the server config in version control and all of our developers could checkout the latest server config. In RAD6, each developer has to manually config the server in their IDE. This is a waste of time for our development team. I hope IBM addresses this shortcoming in RAD 7.

I just found a document Rational Application Developer Performance Tips from IBM, and it introduces some tips to improve the performance of RAD. It may help you if you follow some tips from this document.

Been there...done that; it doesn't help that much. Besides I wouldnt' get your hopes up about RAD7. My portal group tells me that it will be licensed separately and no longer included with $250k+ licensing fee for Portal...what crap!

We were using 6.0.2 before and kind of gotten used to having a mother directory (WAS_HOME/lib) and having all the .jar files under it. We are a standard java environment trying to have our applications deploy on weblogic and websphere. And, so no tieups with any vendor specifc studios. Now having switched to 6.1, I see that I had to include ${was.home}\plugins\com.ibm.ws.runtime_6.1.0.jar in my development environment (IdeaJ) to get some of the classes to compile. (We have custom code using "com.ibm.websphere.ant.tasks.WsEjbDeploy" etc) And I had to include "${was.home}\plugins" in WAS_EXT_DIRS variable while running a stand alone java client (because it couldn't find WsnInitialContextFactory) and so on. I have been looking at the setupCommandLine.bat file, trying to figure out how all these jar files are getting included in the classpath mix from their new locations. When it was under WAS_HOME/lib and when WAS_HOME/lib got included in WAS_EXT_DIRS, it was easy to see. How does the server see these classes from their new locations?

I would like to read this page again.