Naming Conventions in Java, Maven and Gradle by@nedtwigg

Naming Conventions in Java, Maven and Gradle

Read on Terminal Reader
react to story with heart
react to story with light
react to story with boat
react to story with money
Naming Conventions in Java, Maven and Gradle give guidelines on how to name packages. Java Language Architect Brian Goetz says names matter, so names matter. JitPack turns any git commit into a maven artifact at a jitpack. Gradle gives guidelines that are very good - no unnecessary nesting, no unnecessary names. The naming convention is based on the reverse-domain pattern used for Java packages, for example
Ned Twigg HackerNoon profile picture

Ned Twigg

A diff a day keeps the bugs away.

twitter social icongithub social iconlinkedin social iconfacebook social icon
A central aspect of Java's philosophy is that names matter.
Brian Goetz, Java Language Architect and author of Java Concurrency in Practice


If you work for "" and you're working on a project called "foo" then the convention is for your root package to be

Being a forward-looking bunch, we tend to add lots of "grouping" packages. Our foo project is a utility library, and we might make other ones, so we put it into
, to make sure there is space for other utility projects. If we're really forward-looking, we'll take into account that foo is a utility library for manipulating text, so we better put it into
. On the other hand, YAGNI.

In the javascript world, there are no packages, you just get one name - 

 - and that's it. I'm glad that I do most of my work on the JVM rather than a runtime which was designed and built in 10 days, but whenever I start nesting my root package deeper than
, I try to remember that all those gobs of software being written in Node.js are getting by without any nesting at all, so maybe I can get by with 3 or 4 levels of nesting for my root package, rather than 5 or 6.

Maven groupId:artifactId

One nice thing about 

 files is that you don't have to pick names for them - they get their name automatically from a 1:1 mapping on the 
 file they came from. Unfortunately, maven asks you to pick two names, so it can't be that simple. Luckily, maven also provides not only one, but two slightly different conventions for how to pick these names!

Accordingly, if you ask StackOverflow, you'll see two popular answers. And when a major library like RxJava ships a new version, you'll need a 12-screen-long debate to figure out what the names ought to be.

Luckily, there is a mechanistic answer out there! JitPack turns any git commit into a maven artifact at
, which I've found to be huge improvement over 
 for integration testing. The naming convention that it uses is:

  • com.github.{user}:{repo}
    , and
  • com.github.{user}.{repo}:{subproject}
     for multi-module builds

If you use JitPack's custom domain integration, then you can replace

 with com.acme for a professional touch. Even if you don't use JitPack, using this convention will mean that you could, and it lays down a simple rule that works well enough for all these people.

Gradle plugin id

In Gradle-land, you can apply a plugin like this: 

plugins { id '' }

Gradle provides an excellent guideline:

As a convention, we recommend you use an ID based on the reverse-domain pattern used for Java packages, for example org.example.greeting.

The trouble is, a huge number of the plugins which have been published so far are named... gradle-plugin. It seems reasonable while you're writing the plugin, and you don't realize how silly it is until you use it from a distance:

plugins {
    id ''
    id ''
    id 'beetlejuice.beetlejuice.gradle'

What kind of car is that? That's the Chevy Corvette Car. What kind of phone is that? It's an iPhone XS Phone. Wouldn't this be better?

plugins {
    id ''
    id ''
    id 'beetlejuice'

It's interesting, because the guidelines that Gradle gives are very good - no excessive nesting, no unnecessary 

, just the bare minimum. And yet, a lot of the people who use it feel like they should add a 
 or two, just in case. Probably the only way to save us from ourselves would be for the gradle tooling to search for 
 in the id and throw a warning, to help us think about it a little before we publish.


Designing namespaces is a rare opportunity. Most of us never do it even once in our entire career. And since nobody has experience in it, we're still making lots of beginner mistakes, even in the quarter-century-old world of java. It's hard! Typosquatting, IDN homograph attacks, there are so many pitfalls. If I ever end up defining a namespace, I'm gonna try to remember these lessons:

  • namespaces are helpful for identifying the author/maintainer: 👍
  • namespaces are overkill for fine-grained categories: 👎
  • defining the name to be a tuple of two other names is 👎
  • if the names are all plugins to foobar, people are just gonna name them my-foobar-plugin, and it might be good to warn them ¯\_(ツ)_/¯

Previously published at

react to story with heart
react to story with light
react to story with boat
react to story with money
. . . comments & more!