typed by Hans

Barry Schwartz wrote the book The Paradox of Choice: Why More Is Less more and gave this TED talk on why too much choice can be a bad thing.

To summarize, too much choice is bad when:

• It paralyzes decision making because you fear making the wrong choice.
• You recalculate the benefit of your choice by subtracting the benefits of the choices you didn’t make (i.e., subtracting the opportunity cost).
• Choices leads to escalation of expectations.
• You blame yourself for not making the right choice (if you didn’t have to choose then you couldn’t be blamed for making the wrong choice).

A couple months ago I signed up to do the Hood To Coast relay run, which is next weekend.  I’m running with a really well organized and experienced team.  The team captains are taking care of all the organization, much all I have to do is show up and run so this is a fantastic opportunity!

However, since I signed up for the run I’ve had some regret because I’ve discovered I’m missing out on several other things (the company picnic, the first Chinese class of the new semester, an annual 100 mile cycling event, and watching the last couple of days of the Olympic games).  Why should having all these options available to reduce the benefit of my decision to do the HTC run?!  Except for missing the class, none of the other options would even remotely be valid reasons for not doing the run!

Thanks to Barry, I now understand why I feel a little bit of regret, even though I shouldn’t.  However, in his TED talk Barry didn’t say what I should do about it…  it’s not unlike Daniel Gilbert, who explains in his book - Stumbling on Happiness - why we are often unhappy about our choices, without really explain what to do about it!

This article describes the directory structure and command shell environment configuration I use on my development machine to help manage tools, source code and dependent libraries.

Directory Layout

On Windows I keep all my development tools and source code under c:\dev.  On Linux I use the same layout under $HOME/dev/

Inside the “dev” folder I have the following sub-directories:

…/dev

/tools - contains executable application, tools and utilities, like Eclipse, DbVisualizer, Tomcat, Maven, HSQLDB, etc.  I only place tools here that you can install manually simply by unzipping them (for example, Ecipse, Maven, Tomcat).  For the tools or applications don’t ship in a zip file but only come with installers (for example, Java, Oracle, TortoiseSVN) - I let these tools install to their default locations (for example, under “C:\Program Files” or  “/usr/local/”).

/lib - contains binary distributions of 3rd-party libraries and documentation, like GWT, JUnit, SpringFramework, Hibernate, etc.  These days this directory is actually pretty empty - since I now mostly rely on the Maven to download 3rd party libraries and store it in the local Maven repository.  However, it is still sometimes useful or necessary to have the documentation and examples handy.

/os - contains source code for 3rd-party Open Source libraries and tools.  I also use it as a sandbox for unpacking and executing 3rd party sample code.  The entries here are usually transient.

For example, on Windows I have:

    c:\dev\
            tools\
                    \apache-maven-2.0.9
                    \apache-tomcat-6.0.14
                    \eclipse-jee-ganymede-win32
                    …etc…
           lib\
                    \gwt-windows-1.4.62
                    \htmlunit-1.14
                    \spring-osgi-1.0.2
                    …etc…
           os\
                    \eclipse-equinox-SDK-3.3.2
                    \pax-samples
                    \servletbridge
                    …etc…

I also place my own source code projects directly under “dev”, for example:

    c:\dev\
            apps\
            spikes\
            voluble\
            …etc…

BTW, “spikes” is a nice “catch all” place for trying new things out before moving them into “real” projects. Even though “spikes” only contains throwaway or proof-of-concept code, I tend to commit the contents to a SVN repository. Some people use the term “sandbox” instead of “spikes”.

Shell Environment Configuration

To set up my environment variables I create a command file in the “dev” folder - usually called “setenv.cmd” on Winows and “setenv.sh” on Linux.  The file sets up the path, environment variables, and on Linux it also sets up aliases.

On Windows I find it far more convenient to manage the environment settings through a command file rather than going through the Control Panel (System > Advanced > User Profiles > Environment Variables) dialog.  It’s also very useful if to set your environment through batch files if you sometimes need to run a different versions of a tool (for example, if you use Java 6 for new projects but sometimes have to run Java 1.4 for legacy projects).

This is an example Windows setenv.cmd file:

SET JAVA_HOME=c:\Progra~1\Java\jdk1.5.0_14
SET ANT_HOME=C:\dev\tools\apache-ant-1.9.0
SET CATALINA_HOME=C:\dev\tools\apache-tomcat-6.0.14
SET MAVEN_OPTS=-Xmx512m
SET DERBY_HOME=C:\dev\tools\db-derby-10.3.2.1-bin
SET SPRING_OSGI_LIB=C:\dev\os\spring-osgi-1.0-rc2\lib

@echo off

SET PATH=%JAVA_HOME%\bin
SET PATH=%PATH%;C:\WINDOWS
SET PATH=%PATH%;C:\dev\tools\apache-maven-2.0.9\bin
SET PATH=%PATH%;C:\dev\tools\putty
SET PATH=%PATH%;C:\dev\tools\apache-ant-1.7.0\bin
SET PATH=%PATH%;C:\WINDOWS\system32
SET PATH=%PATH%;C:\dev\tools\bnd-0.0.238
SET PATH=%PATH%;C:\dev\tools\db-derby-10.3.2.1-bin\bin
SET PATH=%PATH%;C:\progra~1\Subversion\bin

echo on
echo PATH=%PATH%

Introduction

I have developed a small sample to demonstrate how to use Hibernate in OSGi with Spring Framework and Spring-DM.  The sample works in either Felix or Equinox. It builds with Maven 2 - the build process automatically downloads the Felix and Equinox jars so there is no need to install either framework separately.

Starting Hibernate in OSGi with the normal Hibernate SessionFactory would allow you to use Hibernate in OSGi - but that would not address the dynamic behavior that an OSGi system is capable of. This sample demonstrates how a Hibernate SessionFactory can be updated dynamically in an OSGi environment. You can dynamically add and remove entity classes to the Hibernate configuration by stopping and starting bundles that contribute clasess to the Hibernate configuration.

This implementation contains a small Swing UI which allows you to see which entity classes are present in the Hibernate configuration. The Swing UI also allows you to issue rudimentary SQL and HQL queries.

The solution is based on this OSGi and Hibernate blog entry by Peter Kriens. The design uses an entry in the manifest file to declaratively add classes to a Hibernate session using an extender model.

Get the code

To get the source code for the hibernate sample, use SVN:

svn checkout http://voluble.googlecode.com/svn/trunk/osgi-samples/hibernate/ osgi-hibernate-sample

Run the code

Run “mvn install” in the new osgi-hibernate-sample folder and see all projects build successfully.

The 2nd last project that builds is “integration-tests”, which uses Spring DM to run an integration test suite in an OSGi container. The integration tests are run against an in-process HSQLDB server.

Continue Reading »

This is a quick and practical task-based introduction to OSGi. We avoid any in depth discussion about the purpose or benefits of OSGi or even the tools referenced in this article. Many high level introductions on the technology are available; for examples, please see the webcasts at Eclipse. There’s also a good webcast on the OSGi efforts at Weblogic.

With OSGi it’s possible to run multiple versions of the same classes in the same container. We’ll start from scratch and build services that demonstrate this ability.

I used Apache Felix as an OSGi container. The other two well known open source OSGi container implementations are Equinox and Knopflerfish. Apache Felix actually seems to be the least popular of the three - I wanted to get some experience using the underdog!

Our immediate focus is to create and deploy applications in an OSGi server. We’ll do the following:

• First we’ll set up a simple OSGi deployment environment.
• We’ll create a simple OSGi bundle that will export classes and a service, making it available to other bundles. We’ll call this Bundle A, version 1. We’ll install and start the bundle in an OSGi container.
• To demonstrate the power of OSGi, we’ll create a second version of the same bundle and install it alongside the first version. We’ll call this Bundle A, version 2. This will demonstrate OSGi technology’s ability to seamlessly manage multiple versions of the same classes at the same time, even though the class and package names of both versions will be identical.
• We’ll create another set of bundles which will import and use the classes and service exported by the first set of bundles. We’ll call them Bundle B, versions 1 and 2. Bundle A, versions 1 and 2 will export services which will be imported by Bundle B versions 1 and 2 respectively. This will demonstrate OSGi technology’s ability to dynamically manage modularized applications.

We will use Spring Dynamic Modules for OSGi(tm) 1.0 (with dependencies), Apache Felix 1.0.3, Apache Felix Maven Bundle Plugin 1.2.0, Apache Maven 2.0.8 and Java 1.5.14. Please refer to the individual project websites for more information and background on these projects. We are avoiding discussion of these tools as we are only focusing on accomplishing the tasks outlined above.

I used Windows XP to create this tutorial. When writing the development and deployment steps it seems much easier and less error prone to reference full path names, rather than relative names. For example, I prefer referencing c:\dev\tools\felix-1.0.3, rather than “your Felix home directory”. You will notice I use the following directory structure for my development environment.

   c:\dev\
                  lib\   (Libraries)
                              \spring-framework-2.5.1
                              \spring-osgi-1.0
                              …etc…
                  tools\   (Executable tools)
                              \felix-1.0.3
                              \maven-2.0.7
                              …etc…

You can change the steps in the tutorial to match your environment, or copy the environment I lay out here. If you use a different layout, simply substitute path references to match your environment. For example, substitute “c:\dev\tools\felix-1.0.3” with whatever you chose for “your Felix home directory”.

To start out I’ll detail where to get the components to build the OSGi deployment environment. I will then use a command file, “setenv.cmd”, to configure my environment variables and system path when operating from a Windows command prompt. I’ll provide a sample “setenv.cmd” file.
Continue Reading »

Shujie & I returned home yesterday after an 11 day trip to the People’s Republic of China. I uploaded my 400+ photos and Shujie’s 115 photos.

Here are some touristy photos of Shujie and me in the Palace Museum / Forbidden City:

I learned so much about China, Shujie, her family, and also a little bit more about myself. I learned that China’s incredibly vast and rich 5000 year old history is not only found in great walls, gates, pagodas, tombs, literature or artwork, but also on the table at every meal!

It’s hard to summarize my impressions of the country or the people, but I can say that I felt completely safe the whole time, everyone seems friendly beyond measure, proud of their heritage, proud of the current pace of development, excited about the future, determined to succeed, well spoken, void of criticism, eager to make visitors feel welcome and unselfishly willing to help out whenever they can.

On the down side, traffic in the cities and towns is a mess - your life hangs in the balance every time you cross the street. Also, pollution is staggering, but as far as I can tell this will improve very rapidly over the next couple of years.

I’m finally going to China, for the first time, tomorrow. I have been dreaming about this for years!  I have seen several documentaries on life in China and have traveled a bit around the globe, so I think I have a pretty good idea what it will be like. However, I still have a feeling that this experience will still touch and change me in unexpected ways. The Twin Peaks quote comes to mind: “When you see me again, it won’t be me.”

My fiancée, Shujie, will be my tour guide. She grew up in China in a small town, Fengqiu, in the Henan province. We actually found the town on Google Earth! It’s ferry cool, and a bit freaky. Here’s the GPS coordinates of a circle in the middle of town, which Shujie says she walked past to school every day: 35.04009 N, 114.41635 E.

We’ll visit Beijing for a couple of days and then go to Fengqiu, where I will meet my future mother-in-law.  Shujie will have to interpret for us, as her mother doesn’t speak English and I speak neither Manderin nor their local dialect.  I’m hoping to take down and memorize some polite words though while we are there.

Today I went through the exercise of setting up a new blog. I prefer to install the blog software myself at a domain that I own and this begs the question: Which blog software should I use?

After a couple of web searches and skimming some results, I looked a little closer at b2evolution, Serendipity, Nucleus, TikiWiki (mostly the blogging module) and WordPress.

b2evolution looked promising, but there are little animated emoticons on the blog editor which I found rather distracting and silly. I couldn’t figure out how to hide them, and promptly gave up.

I’ve tried Serendipity before, but never tried to do any customization. The default theme was rather humdrum. Also, on their feature list they reference WordPress integration, which really made me think I should look more closely at WordPress instead.

I turned away from Nucleus because the screenshots link on their websites took me into the Gallery application. I like Gallery, but I really want a blog system that will make it easy for me to post photos inside the blogs and I reason that, if the Nucleus website doesn’t use the Nucleus software to display screenshots, then it’s probably a little too cumbersome to do.

I’ve played with TikiWiki in the past and it looks extremely powerful. However, after installation, it took me several minutes just to figure out how to enable the blog module, so the whole system seemed a little too much like Linux in the 90’s… it’s very powerful, you have full control, but you really need to put some effort into learning how to tame the beast.

So that left me with WordPress. It’s very popular and extremely well supported, but I used it about 2 years ago and remember that inserting images into blogs were really hard. So I played around for a while on the WordPress site and noticed an example of an image in the Wysiwyg editor. Very encouraging!

I’m very impressed with the wide range of plug-ins for WordPress and the plethora of themes! I settled on this very cool Paalam 1.1 theme by Sadish Bala (”made free by Web Hosting Bluebook“). Uploading my own banner image is dead simple, and WordPress even has a built in image editor to help crop the image. The default Wysiwyg editor I’m using right now (I believe it’s called TinyMCE) is also really cool. I’ve gotten so used to editing Google docs documents - editing WikiText is really awkward by comparison.

The blog research and implementation took me several hours… but I’m happy with the result. Now, lets see how easy it is to add an image to this blog…

Wow! That was fantastic! There’s a file upload form right below the editor. After uploading, a button lets you insert a link to the uploaded image into the document. It couldn’t be simpler (well, without drag & drop at least). I’d like to figure out how to make the text flow around the image (without having to edit the HTML and stilysheet), but this is a great start! BTW, it appears that switching between “Code” and “Visual” input is also really seamless in the Wysiwyg editor.