Blogging Roller

Tonight at the TriJUG meeting, Andrew Oliver gave an interesting talk on what initially seemed to me to be a boring topic - reading and writing OLE2 Compound Document Format, Excel, and Word documents from Java. He was talking about the Poor Obfuscation Implementation (POI) project. The talk was good because Andrew is a good speaker and because he included a nice introduction to Cocoon in his presentation.

I know about Castor and Torque, but I need learn about Object Relational Bridge, OFBiz Entity Engine, Hibernate, JGrinder, JRelationalFramework, and Abra. Maybe I should just skip all that and go straight to Sun's JDO - it is going to rule the roost, right?

Mike is right, Doc Searls' Anarchy & Infrastructure presentation is excellent.

They are talking about this CRN article over at The Server Side: Former Employee: WebGain Out Of Business. If you hadn't nailed him to the perch he'd be pushing up the daisies! His metabolic processes are now history! He's off the twig! He's kicked the bucket, He's shuffled off his mortal coil, rung down the curtain and joined the bleeding choir invisibile! This is an ex-parrot!

I have to admit that when I heard about J#, I was massively skeptical. After spending an evening with it, I have to admit that it appeals pretty deeply to the subversive in me. The thought of running Ant, Xerces, or Tomcat under the CLR is pretty interesting. Don Box

How is Don going to run Ant, Xerces, and Tomcat under the CLR? All of those things require Java 2, but Microsoft's Java VM is still stuck on JDK 1.1.4 - isn't it? I think he has his beer goggles on. And another thing: did he "probe every corner and alcove" of Stephanie or Visual J#.Net? That article is really creeping me out!

As Jeff has correctly pointed out, XDoclet will start on a major refactoring to Velocity after the next release (in a month or so we hope). XDoclet 1.1.2 has 17000 downloads and many contributors. vDoclet is the third tool in its category (EJBGen, XDoclet and vDoclet). My question is: Do we need another one? Aslak Hellesoy from the Velocity mail list

Aslak is encouraging the VDoclet team (a team of 1?) to join forces with the XDoclet team. XDoclet is preparing to drop the weird XML templates and adopt Velocity as it's templating language. That is great news. Those XML templates are butt ugly.

I just uploaded a new new source release because the 0.9.4 release did not include build scripts! The build scripts do so much code generation that there is really no way to do a build without them and nobody complained (that is an interesting data point). Anyhow, after you down load roller-src-0.9.4.1.tgz and roller-tools-0.9.4.1.tgz, doing a build should be as easy as this (and ignore the XDoclet warnings):

   % tar xzvf roller-src-0.9.4.1.tgz
   % tar xzvf roller-tools-0.9.4.1.tgz
   % cd roller
   % build all

The machismo of this kind of U.S. business prose trickles through from the rhetoric of the sports field. The CEOs are the multimillion-dollar sports stars, the employees are encouraged to think of themselves as part of a winning team, and sales and marketing departments are heavy with the motivational chatter of sports. Playing the game, Red Herring

I hate that motivation chatter but the fact that it is based on sports cliches is what really makes it unbearable.

From Business 2.0. Steve Balmer's Dance Monkey-boy performance is #9

[Struts] developers have a differing opinion on whether the Action class is part of the controller or the model. The argument for it being part of the controller is that it isn’t part of the “real” business logic. If Struts were replaced with an alternative framework, chances are the Action class would be replaced with something else. Therefore, it really isn’t part of the model domain, but rather tightly coupled to the Struts controller. It doesn’t make sense to put business logic into the Action, since other types of clients can’t easily reuse it. From Chapter 3 in the Server Side's Jakarta Struts Book Review Project

In the alternative framework WebWork, the action classes could be considered to be part of the model because they include no presentation specific code (i.e. no Servlet API calls). Some would argue that this is better and allows re-use of actions, but I think it is just different. For Roller, the WebWork approach really does not buy me much - my 'business logic' is already re-usable and encapsulated in my model interfaces and classes.

Karl wants to try Roller. With his content management background, he might have some cool ideas for the Roller roadmap.

If your president did not know the meaning of the word hypothetical would you think that he is an idiot? I can read your mind. You are thinking that 1) I should not talk politics on this weblog and 2) I had better shut-up before I am declared an enemy combatant.

From an article on CNET. Apparently Oracle bought only TopLink and left the rest of the WebGain products to rot in the hot sun while the (Oracle, Sun, etc.) buzzards circle overhead. Also, Rogers Caldenhead pointed out this ominous part of the CNET article:

On Tuesday, the parking lot at WebGain's Santa Clara, Calif., headquarters was empty and the entrance to the building was locked.

Interesting article on Struts by by Harry Rusli and John Yu.

Mike has responded with a nice little write-up of WebWork. This is the point that I find most interesting:

Not tied to the web - WebWork (despite it's unfortunate name!) is not tied to the web at all. As I said, we use it as a general purpose command framework within our application. There is also a ClientAction which allows you to run any action over RMI. There is a project to tie SOAP into WebWork at a very low level.

I've been taking a little time to read about WebWork as an alternative to Struts. I've been googling around for information, and I have not come up with a whole lot. I found a post by the Struts master, some evangelism, and the WebWork docs on some dodgy server somewhere. I found the most informative summary of WebWork on the Barracuda site:

WebWork - WebWork is a web application framework for J2EE. It is based on a concept called "Pull HMVC" (Pull Hierarchical Model View Controller). The basic idea is to separate the site programmers' and site designers' tasks.

It has a small API and easy to use tag library in order to make it as easy to use as possible. It supports the notion of actions that are just JavaBeans. This means that they have setX() methods for incoming parameters, an execute() method to perform the action, and getX() methods to retrieve the result data. This ensures that all actions only contain the code that is required to perform the logic.

The tag library has also adopted this philosophy. They only provide tags to extract and iterate over the result data so that formatting the output as HTML, WAP, or what ever is as straight forward as possible.

If you've got pointers to WebWorks articles, postings, threads, etc. please send me a couple.

I returned to work today after two wonderful weeks of paternity leave. I still have one week of leave remaining, but I have a bug list a mile long, a deadline looming, and a product that is screaming for release after 18 months of development. I need to get back to work. Things might get a little slow around here for a couple of days.

If you've been following Blogging Roller for long, you know what a terrible time I had with CQHost, my previous Servlet ISP.  CQHost was cheap at $12.50 per month, but the technical support was terrible, the server manager software was broken, and the Servlet engine spent more time down than up.  After a week or so of constant server crashes and daily trouble-tickets, CQHost actually blamed my Roller software for the crashes and I knew it was time to leave.

I had been following a thread on Servet ISPs at the Java Lobby, so I already had a short-list of ISP alternatives.  Luckily, Kattare was on that list.  After reading the rave reviews of Kattare in the Servlets.COM ISP list, I signed up for Kattare's Level 3.1 plan for $29 per month.  The plan includes 100mb storage, 4 mailboxes, private Tomcat JVM, MySQL, FTP/SCP, Telnet/SSH, and lots more.

Kattare had me up and running in hours and they answered my every question with a thoughtful and informative reply.   Their systems are fast, well configured, and stable.  Roller has been running at Kattare for a week now and I have have not experienced one server crash, outage, or other inconvenience.   So, if you are looking for a place to host your Roller-based weblog (or any other webapp) look no farther than Kättare Internet Services.

No new features, only bug fixes:

• #576004: Blogger API posts do not flush cache
• #576719: Calendar in Weblog:Edit page not working
• #576902: Error in web.xml, rollerdb datasource was omited
• #576157: RSS feed items are not ordered properly
• #576731: Database jars should go in common/lib

Download it at the SourceForge download page.

Surfing around tonight I found that the #8 link on blogdex was a list of weblogs presented by the Guardian. After noticing with delight that Be Blogging was one of the half-dozen tech blogs on the list, I followed the link to Corante: The Bottom Line and found this little gem:

It is probably true that the corporate scandals would not have happened had the graphical user interface not been invented, limiting the use of computers to people who deal with command lines

I'm not sure I agree with that, but I like the argument and I'll be back to Corante.