Musings of a Trained Monkey: Tag java

Accessibility of Applets versus Flex RIAs...

Steve Longdo — Sat, 26 May 2007 18:58:00 +0000

I’ve been doing some research into providing support for section 508 accessibility compliance for a simple file browser available inside a web browser. There really is a dearth of information available on the topic from a software implementation stand point.

Java Applets seem to pay lip service to providing support for accessibility by the addition of tons of code labels that would need to be maintained with each release. Certain activities don’t seem to be supported at all for users with accessibility needs (opening tree nodes for example). The most recent documentation I could find on Applets and Accessibility on Sun’s website is called What’s new in Accessibility and is dated October 30, 2000 clearly Sun took their role as an industry leader in web technology very seriously…back in 2000. (As an aside is referring to a user of an applet as a Robot just a little offensive to anyone besides me?)

Adobe Flex, however, seems to view accessibility as a core issue in the design of their product. So much so that they provide a very simple xml attribute for MXML files (<accessibility>true</accessibility>) to turn on a host of accessibility features. This requires almost no extra effort on the part of a software developer. There are a small number of caveats to their support, but they have taken the steps to detail them and best practices for accessibility on their site including a functional understanding of how Flex works with JAWS and Braille screen technologies. Adobe even maintains a current website dedicated to the accessibility of their products.

Jetty and WebLogic JNDI sitting in a tree...

Steve Longdo — Thu, 17 May 2007 04:26:00 +0000

I am a huge fan of Jetty for Java web application development. Love how fast it loads, reloads, and generally kicks the crap out of Tomcat. The Jetty team seems to follow the JEE specifications that they do implement (not all of them) much more strictly than Tomcat. This makes it a very good testbed for developing applications that will be eventually hosted on WebLogic Server .

Until you need to do something like run Message Driven Beans with WLS special performance enhancing sauce. Jetty has it’s jetty-plus configuration which supports JNDI and some other JEE niceties. If you really want to use the full on BEA JMS/MDB combo though you have a problem.

Including the weblogic.jar in Jetty’s classpath is not the answer. This will cause all sorts of confusion about which implementations of java. * and javax. * packages to use. Within the WebLogic Server distribution’s server/lib directory there is a wlclient.jar. Again this won’t work right in your overall classpath, but it will work if you put the wlclient.jar inside of a webapp’s WEB-INF/lib directory. Your code will just need to make a small accommodation for connecting to WLS JNDI constructing the InitialContext with the WebLogic specific properties:


    Properties props = new Properties();
    props.put(Context.INITIAL_CONTEXT_FACTORY,"weblogic.jndi.WLInitialContextFactory");
    props.put(Context.PROVIDER_URL, "t3://localhost:7001");
    InitialContext ctx = null;
  try{
      InitialContext ctx = new InitialContext(props);
       (...Lookup JMS Queues, place Messages, etc...)
  } catch (Exception e) {
     (do something!)
  } finally {
       try{ ctx.close() } catch (Exception e) { (eek! just have to eat this one!)}
  }

This gives you the flexibility of using Jetty configured JNDI objects as well as leveraging WebLogic JNDI objects without having to eat huge WebLogic redeployment startup times for your web application.

Running BEA WebLogic Server on OS X...

Steve Longdo — Sun, 29 Apr 2007 19:31:00 +0000

People have trying installs of BEA WLS on OS X out for awhile. I remember installing WLS 9.2 back in 2005. I recently ran across a blog entry about installing WLS 10.0. I’m still using a PPC Mac with not nearly enough RAM, so I thought I would try to see how the 10.0 performance would be compared to my 9.2 experience.

Surprisingly 10.0 is much slower to start up. I appreciate that OS X and Apple’s JVM aren’t a supported platform, but still it took almost 7 minutes for the MedRec sample domain to start up! Under 9.2 it was only 2 minutes. Back when I installed 9.2 Apple’s JVM was 1.5 and I have 1.6 running now.

I wanted to do a fair comparison, but apparently installing 10.0 with Apple’s 1.6 JVM overwrites the PointBase configuration such that 9.2 won’t even start up the MedRec sample domain (incompatible serial uids for javax.xml.namespace.QName if anyone cares). Also Apple says that uninstalling their 1.6 beta is a bad idea?!? I’m sure I could probably clean it up and get it running, but that would exceed my ten minute or so attention span for working on Java stuff at home.

Code Generation STILL in Action

Steve Longdo — Sat, 17 Feb 2007 15:30:00 +0000

— The Past —

Jack Herrington wrote a book Code Generation in Action way back in 2003. It took advantage of the little programming language that could, Ruby. I remember having a discussion with a colleague back then about the value of generated code. We were of the opinion then that it would only apply to simple repetitive cases and wouldn’t make much of an impact.

If only I could go back in time and slap both myself and my colleague in the face! We missed out on being early adopters of time saving technology. In fact almost everyone using Java was already taking advantage of code generation whether they realized it or not , Eclipse would generate getters and setters and had its refactoring capabilities, Spring was beginning to handle externalizing dependencies, and the then fledgling Hibernate Framework was manipulating bytecode to enhance Java methods.

At the same time my colleague and I were hammering out code with the tried and true cut-n-rape technique that had been popular since before we were born.

— Modern Day —

I have the “opportunity” to work on porting some old JSPs which, in true 20^th century Model 1 style, link directly to each other and the database via scriptlets, to a more modern Java MVC framework. Rather than just wade in blindly, manually translating JSP scriptlet page to Java MVC framework code, I decided to make the project interesting and have been writing a Ruby JSP parser/code generator. It can identify scriptlet code and extract it for me. I have been enhancing the parser to handle generation of Spring MVC code and am working on having it generate Spring Web Flow flows from reading the form action attributes in the JSPs to model application flows.

Here is a 2006 appearance by Jack Herrington at Google on the subject of Code Generation that is worth a watch.

Spring now in 3D

Steve Longdo — Tue, 12 Dec 2006 01:50:00 +0000

I have resisted posting anything about the Spring Experience 2006 until now. Mostly because postings on my blog have been Ruby-centric or me-centric, but let’s break the embargo for a post or two.

The opening keynote by Rod Johnson was a loser. He started out by checking his bank balance and then doing a UK patent search to demonstrate that some websites in the world use Spring under the covers! Duh!

While Rod was dumbing it down for those in attendance, a table at the back of the room featuring some of the conference speakers as well as a baby, began to reminisce about other conferences they’ve gotten money for and such. They were loud enough it was hard to hear the keynote. This was especially rude because the speakers are only there to present because people like me (well my company) paid money for them to be there. Glad you guys enjoy being around each other, but take it out in the hall next time with the baby if you want to be loud.

Also I think some limited mockery of Domain Driven Design is called for. People need to stop making up ?DD acronyms. I’m all for having a software methodology and following its doctrines when they make sense, but you just know some idiot somewhere is going to post about how This JEE Web 2.0 website , is now brought to you in 3D!.

The Spring guys seem to have bitten into this D³ stuff hook, line, and sinker. Martin Fowler blessed the book with his own brand of stink so it must be good right? There was a dedicated D³ track available at the conference on the topic. I can’t figure out why there wasn’t enough Spring content to fill out that track as well. It was the Spring Experience after all!

Anyway there was some good content presented, Matt Raible did a great job writing it up for the Java faithful. I’ll leave my non-mocking commentary on the conference some other time.

JDK 6 - Dec 7th, 2006

Steve Longdo — Wed, 29 Nov 2006 04:24:00 +0000

Java SE 6 is to be released this December 7th, 2006 (a day that will live in infamy indeed!). Wow! How out of date are the Java applications you are working on? I feel almost as bad as I did when JDK 5 came out (December 2004!!!) and I wasn’t able to utilize it for ~~months~~ years. I fear the lag time will be much longer this time.

Java, Spring and Why I Haven't Posted Lately...

Steve Longdo — Fri, 06 Oct 2006 03:37:00 +0000

Things have been busy on the work side of my activities lately. Work is currently revolving around Java and using Spring MVC. Since I have primarily focused on Ruby and Rails information on my blog, I've been reluctant to blog about where my head is at in the Java space. If someone is interested, leave a comment and I'll see what I can come up with.

Which isn't to say I am not tracking recent developments in the Rails community, just that I haven't had time to analyze their impact and vomit forth a blog entry or two :-)

Okay, one little morsel of my experience with Spring relates to one of the core tenets of Rails: "Convention over configuration".
It finally found a way into a Java web development framework!

Java the Surrealist Way...

Steve Longdo — Sat, 22 Apr 2006 23:52:00 +0000

Like great art, this post by Darren Hobbs will long be remembered as the one that described Java in a way that most anyone can appreciate :-)

Also if you already have any programming experience with Java it will make milk shoot out of your nose, even if you aren't currently drinking milk!

XSD is dead from a POX upon its house...

Steve Longdo — Sat, 11 Mar 2006 10:44:00 +0000

If you’ve experienced the Enterprise SOA push from vendors at the large company you work for, there is some seriously funny reading to be had over at Patrick Logan’s website. I read them out of order having neglected Patrick’s feed for far too long.

Anyway start with Bad Idea, It keeps on raining… and finish off with This is Enterprise Stuff Look Out. Patrick makes a quote worthy quote in the last one:

Yes, why list actual reasons why XSD is good when you can just claim the top of the hill and dare people to knock you off. I think you might fall down on your own.

Having worked with SOAP/WSDL/UDDI/SOA J2EE shops and the “fun” that could be had by implementing different vendors tool frameworks because of the supposed possibility to interoperate. .NOT’s doesn’t play very nice with Java’s All of these radical redeisgn projects are overhauling either Mainframe systems that continue to shove data down the wire as they have for decades. By cobbling together a standards based “confederation” of servers running the latest OS’s with more memory and better IO times to large SANs, and yet even with it all set up, the Mainframe often remains the core piece because no one knew that it did “XYZ” and we didn’t buy the right software from a Vendor for this situation.

Manager’s begin to go bald as costs exceed even the most out there cost projections for the project. Vendor Consultants are hired to come in and clean up the implementation. The begin by dissecting it and do design documentation. This design/discovery/recovery period goes on for awhile… Until finally, the Bald Manager, with only his eyebrows intact, cancels the project!

Too bad no one at the company would adopt a new way of doing things or even admit that SOAP/WS**/SOA stuff is really just an attempt to ressurect CORBA. Don’t get me wrong the Vendor’s dolled CORBA up with some XML implants and tightened up some of the old source code wrinkles. Made quite the sales pitch for the new *Sexy cO*rbA* and managers all across the globe fell for it again.

Thanks for letting me drink from the well of truth you have over there Patrick I won’t be neglecting your feed any longer.

Tomcat 5.5/Java 5 and SSL...

Steve Longdo — Mon, 12 Dec 2005 20:30:00 +0000

I ran into a puzzling situation trying to setup Yale's CAS and the Spring Framework/Acegi contacts sample application on my workstation. I had CAS deployed on Tomcat 5.5 and the Acegi sample application deployed on BEA WebLogic 9.0, both running under Sun's JDK 1.5.0_06. I kept getting errors like:

Your CAS credentials were rejected.

Reason: HTTPS hostname wrong: <localhost> should be <127.0.0.1>

I couldn't figure this out because the server.xml for Tomcat included both the keystoreFile and truststoreFile parameters for the SSL connection. The JVM JRE cacerts file included self-signed certs for both the BEA server and the Tomcat server which should be picked up system wide whether or not I specifically include it in the server.xml file. I am not sure why but putting the following option on the Tomcat command line solved the problem:

-Djavax.net.ssl.trustStore="C:\Program Files\Java\jre1.5.0_06\lib\security\cacerts"

You can also see from this that it is a Windows workstation... I am documenting this for the benefit of other programmers and Google (which doesn't have much of anything about this particular problem).