Software Architect

Guice Dependency Injection

I've been reading more and more about Google's Guice in different places.  I finally decided to visit the source of all these news reports and blogs posts to understand for myself what it was all about.  Guice is Dependency Injection (DI) through attribute tags in code.  DI essentially changes the way you think about the use of factories and object building.

I've embraced DI in the past with my use of Spring Framework.  The difference seems to be that Guice eliminates the need for mapping objects to one another to handle the injection.  In that respect, it reminds me of the Microsoft Composite Application Block (CAB) way of handling service dependencies.  Under CAB, simply register your service with a WorkItem and then attribute a property on the class being built and the framework will automatically inject that service.

Guice takes a slightly similar approach that uses the concept of Modules. According to the Guice User Guide: "With Guice, you implement modules. Guice passes a binder to your module, and your module uses the binder to map interfaces to implementations."  You then attribute your code to specify where the injections of the implementations should take place.

I think I need to read some more before offering any more thoughts, but I do like the direction Guice has taken.

OLEX Rating Open Source

I don't think that this is the first service I've come across like this, but OpenLogic Exchange (OLEX) offers what many organizations need in their decision to go to an open source product: Support, Update and Indemnification Services.

What I find most interesting and unique about their service, is that it falls in line with something that I've blogged about a few times now in reference to SourceKibitzer and Ohloh.  OLEX puts all the open source software in the library that they support through a 42-point certification process.  That certification process is described here.  They essentially rank the open source projects that they support.

The certification process is not public, but it calculates a score based on three areas OLEX believes are important: Adoption, Community and Licensing.  It's unclear how easily it is to affect this score and the disclaimer at the bottom of the page makes it clear that the rating is only a guide and "not a replacement for your own due diligence".  I assume that since it bases it's scores on the same data as everyone else, that it should be quite simple to get an odd ranking here and there.  Of course, it is my impression, as I browsed the projects in their library, that OLEX is focused on larger name and longer lived projects like MySQL, Apache and JBoss so the chance of an off rating would probably be relatively slim.

Either way, I'm glad to see more resources for IT folks to make more informed decisions on open source and hopefully the support offered through OLEX and organizations like it will allow open source to become more pervasive.   

New SQL Server JDBC Driver

I think that there are a lot of people who hope that this new Microsoft JDBC Driver turns out to be a big improvement on the previous release.  I can think of many times where I was forced to use jTDS to access SQL Server.  "Forced" might not even be the correct word, as the jTDS driver stacks up quite nicely to other options that are available.

I wonder when the guys at jTDS will re-do their driver benchmarks now that this new version is available.

Measuring People in Open Source

I've made several posts now discussing how to evaluate an open source project prior to working it into your architecture.  Free is great, but the zero on the price tag could add up to a far more expensive venture if you're left with a spotty product and no support. 

The posts that have discussed this issue are those that have focused on one of two web sites, Ohloh and SourceKibitzer, which both present analysis of open source projects.  Both of these sites offer insight into an evaluation variable I haven't mentioned directly before.  I think a good measure might be the people involved in the project that you are evaluating. 

Consider some of these questions: Are they recognized by their peers as experts? How many recognized experts are on the project?  What is their most recent contribution and how big was their overall contribution?  How long have they been on the project?  SourceKibitzer even goes so far as to offer a code complexity metric.

While you have access to all this data through these web sites, there is a word of caution to consider and I've mentioned it in previous posts.  Some of the project metrics can lie or tell only part of the truth, so in the end these evaluation points should only be used as part of a larger evaluation strategy prior to making a decision.

"Contributors Brand Yourself"

"Open source contributors, brand yourself."  That's the title of an article I just read which was actually an interview with the co-founders of SourceKibitzer, Anton Litvinenko and Mark Kofman.  SourceKibitzer is a site that gives recognition to developers who contribute to open source projects.  It is not much unlike, Ohloh, a site I discussed in a post some time ago, though Ohloh appears to be further down the development path in terms of functionality that they expose.  Ohloh also seems to be more focused on  being project-centric whereas SourceKibitzer is more contributor-centric.   

I did find it interesting that SourceKibitzer follows a User Programmed Service (U-PS) model.  To quote from the article a U-PS is "Open Source [software] adapted to the development of a software as a service (SaaS). The source code is open, infrastructure is built in order to support participation, but the software itself is deployed as a service."

This is the first time that I'd heard of this type of model.  I decided I'd find out more about it.  A quick google search later and I realized that wikipedia didn't even have an entry for it yet.  In fact, the only place that had any reference to it was SourceKibitzer.  So it appears that this model is more new than I had first expected.  It will be interesting to see where this project goes and how this model sustains itself.

New Look for Spring

I really like the new look for the Spring Framework .NET logo.  Definitely a good change for them.  If you're using .NET, and you're not familiar with Spring consider giving it a look.  I've met too many .NET developers who haven't and they're making a mistake by not knowing what they are passing up.

DataTable Expressions

Day-to-day programming almost always leads you to a problem that you're not quite sure how to solve.  Luckily the Internet is there to save the day and for every problem I've had there have to be at least several thousand developers who've had the same issue.  Of those several thousand developers, there has to be at least several dozen who have written a blog post, quick tutorial or some other web content that explains the issue and how they resolved it.

As a result, I never feel the need to publish many solutions as most of the time I'd just be rehashing content that's already out there.

Once again, that is the case now, but I was so happy to notice this solution that I thought I'd rehash a fairly simply concept: DataTable Expressions. 

I don't do much with DataSets.  I use them quite sparingly and usually only for quick programming tasks.  I had such a programming task the other day and needed to perform some simple, but dynamically generated calculations on the data within the DataSet.  Clearly there were a number of ways to accomplish this, but I came across one of the overloads on the Add method on the DataColumnCollection class, exposed through the Columns property on the DataTable that solved my problem quite simply.

This method took three parameters, the name of the column, the data type for the column and finally an expression.  The expression allows you to add an on-the-fly calculation field to the DataTable. 

So in a situation where you might have a two column DataTable one called Price and another called Quantity, you could add a third column to the table named Total and given it an expression of Price * Quantity.

My tests so far have shown it to perform reasonably well given the conditions I have it working under.

Is There Beauty In Code?

I realized just how big a nerd I was when I opened the first pages of my newest purchase from Amazon, Beautiful Code: Leading Programmers Explain How They Think.  As an architect might look at a building, see beauty and put it into words, the programmers in this book try to do the same.

I've only read through the first three chapters, the titles of which give you an idea as to just what kind of person might read this book:

• Chapter 1: A Regular Expression Matcher
• Chapter 2: Subversion's Delta Editor: Interface as Ontology
• Chapter 3: The Most Beautiful Code I Never Wrote

While I have only read though less than one-tenth of the text, I feel like there's value in what I'm reading.  Sure, understanding why an implementation of QuickSort is beautiful to someone isn't exactly useful to my everyday development efforts.  It doesn't answer a problem or even address the domain or programming abstraction level that I usually work at.   It does, however, help further open a dimension to the perspective in which I view the product of my workday.

The moment you can find beauty in your work whether you build furniture or build software, your job becomes immediately less of a "job".  I first saw beauty in software very early in my career.  It appeared in the form of a data model .  I could tell it was "beautiful" from across the room where the words were so small as to be completely illegible and the entities simply wired boxes with thin highways of lines connecting them together.  For all that abstraction from that sky-high view, I can still remember marvelling at it with its chief designer, as we threw one implementation scenario after another at it, looking for holes.

While I've spoken of beauty in code with some colleagues, this book represents one of the first times I can remember seeing something like this in print.  As I wrote earlier, I don't know how much practical use this book is, but it allows programmers an opportunity to see their work from a perspective we rarely take the time to see it from.

Attacking Java on The Server Side

Nothing quite starts a discussion on The Server Side like the guy who starts a post that plays to the fear/passion of every Java developer: "Java is Doomed".  I honestly tried not to read all the replies, but I did give it a quick scan as those kind of posts typically leads me to links to and information on new innovations in the Java world from folks who want to defy those who believe that Java is on its way out.  I'm sorry to say I had no such luck on this thread.

NDbUnit Got Me Thinking

I forgot to mention in my last post what sparked my thoughts on .NET ports of open source Java projects.  I was floating around the Internet and came across NDbUnit, which I immediately recognized as a port from the the JUnit extension DbUnit.  It's amazing how quickly someone can turn around these ports.  As fast as I can consider helping out an open source project with a port to .NET, I immediately find out that the work is done.