22nd July 2011
I've been writing Java in one sense or another for a few years now. I learnt stuff at university, then used it in a few jobs. I've written Beans and Applets, and various bits of stuff in between.
It's fairly safe to say, that I like Java.
One thing however, that's been a pretty consistent bugbear in all the time I've been writing Java, has been the classpath, and dependency resolution. Luckily, all that can now change.
I used to think that Ant was a pretty neat build tool. All the IDEs supported it, it kinda worked most of the time, but sometimes, building was a bit of a ballache - Some stuff had to be in your lib/ folder, sometimes in Ant's lib/ too.
Lately though, and this week in particular, I've been playing with Maven.
Maven is a pretty fucking cool build tool for Java applications. I suspect it probably works with other languages, but it's "designed" for Java.
I don't think I really have the expertise or knowledge to explain how Maven works, partly because I haven't studied the inner workings that deeply, but also, because it's far better explained here (http://maven.apache.org/what-is-maven.html).
Instead, I'm going to dive right in, and explain what I've been working on this week.
The company I work for currently, is making a pretty radical shift away from using PHP for everything. Instead, we've been investigating Java for creating a middleware layer that everything can talk to.
I'm pretty chuffed with this, but I do wish that it had come a lot earlier on. If it had, I might not have been so decisive to leave when offered a better job.
Basically, when we came up with this project, I insisted that we do it properly, for a change.
I suggested that a good workflow would be something like: Netbeans IDE -> Maven Project -> Git SCM -> Jenkins CI -> Maven Repository (We chose Artifactory, but I did test Sonatype Nexus too, but didn't like it).
This is a good pattern for the Joel Test's "Can you make a build in one step?"
I basically wanted to create a demo project that can be used as the basis for all future FR projects, I do the R&D to make the initial POM work, then everyone else can clone this, or inherit from it..
This decision was twofold, I also wanted to figure out JPA/Hibernate and have some clue how that works for reverse engineering the classes from an existing database, the answer to that is: Pretty well, actually. - But that's another story.
My IDE of choice is Netbeans. I've been using it since I was at university, except for a small android-related foray into Eclipse, and an experimental nosing around IntelliJ IDEA.
Stuff I did:
That's pretty much it. Here's the finished POM, with various bits of secret removed.
When I edit something in Netbeans, and commit a change, there's a post-commit hook (post-receive) that calls the Jenkins API, and builds the project. Jenkins then deploys the artifacts (a fat JAR and a POM) to the Artifactory.
Some of these may be partially falsified, but Tom doesn't know either.