Why is it so Cool to Develop with Java?

2018 01 01 head

Why is it so cool to develop software using Java, a modern IDE and agile tools?

Because the environment empowers you and incredibly speed-up your development outcomes. In seconds, you know if the next code increment adds value to your product.

Test driven development and static quality tools are integrated in your IDE. Each time you check some code into your Git repository, all configured checks shall be executed before allowing the commit.

Liberated your team works adding functions and no more spend unaccounted hours in setting up infrastructure.

This trend is propagating to other technology stacks such as Python or C#, but less for PHP or Javascript/Node.js worlds. I still experience higher maturity in the Java toolchain.

Below, our current set of techniques and tools to multiply our results and bring earlier products to your customers.

Java Quality

Solution Quality

The most efficient approaches to improving the intrinsic quality of your software solution are

Use Pair Programming

Write regularly together source code. Decide that all complex code parts shall always be developed through pair programming. If you feel adventurous, give a try to mob programming. Pair Programmers learn:

  • To perform a form of continuous review.

  • To provide a measurable increase of quality.

  • To evaluate more options than solo programmers.

  • To smooth the low points of your partner and level-up the high points.

  • To produce simpler solutions.

Use Refactoring

Before pushing your code to the Git repository, always take a look and refactor your code. You are a craftsman and shall be proud of your result,

Use Java 8 Lambdas and Streams

Actual Java code builds on lambda expressions and stream constructs. Your for loops are replaced by stream pipeline, your algorithms extensively use lambdas.

Use Java 9 Modules

It is finally possible to clearly define public interfaces of a whole software component. Configure the dependencies to other libraries and identify specific modules as privileged participants.

Do not write Java code as it was done last millennium. Use the actual constructs to write solid and resilient code.

Pairs complete a task in 60% of time taken by a solo programmer. So two people are almost but not quite twice as fast.

Programmers working in pairs produce higher-quality output. Teams with high scores in stability and throughput spend 44% more time on new work.

No better way to spread learning in a team!

Code Quality

The most efficient tools to detect glitches or legacy constructs in your code are

SonarLint

The plugin for your favorite IDE provides all the SonarCube checkers to apply on your source code. Purify it before pushing it into the continuous build pipeline. The plugin supports download of tailored rules from your company SonarCube instance.

SpotBugs

The successor of the wildly popular FindBugs tool can be integrated in your build pipeline to improve the odds of correctness.

PMD

Is another tool to automatically detect potential quality problems.

IntelliJ IDEA build-in Checks

You are lazy and do not want to use additional tools. Please use at least the build-in checkers of IntelliJ IDEA to improve your code. It is very unprofessional to have equality problems or boolean condition errors in modern code.

Test Quality

Test Driven Development - TDD

You are serious about development quality code. So you are working with the test driven approach TDD and need to write unit tests. Each time you write some code, you go through the loop "red - green - refactor - commit".

Junit 5

is the standard and best unit test framework for Java solutions. The newest version supports lambdas is a pleasure to use.

AssertJ

enables the writing of legible assert statements in your unit tests.

Mockito

Tests must be lightning fast, thousands of unit tests shall run in a few seconds. Remove all dependencies to external components from your source code, mock these elements with Mockito.

Hamcrest

Your test conditions are complex. You want to have a legible code, start using hamcrest to write clear assertions in your tests.

Please never access a database or a file system when running unit tests. If you low-level tests with files, use an in-memory file system mock. [1]

Each time an error is found in productive code, write first a test to reproduce it, then correct the error. This approach guarantees the same error will never again occur.

Acceptance Test Driven Development - ATDD

You are serious about delivering the functions your customer needs. Each story you implement has acceptance criteria. These acceptance criteria define your requirements. See specification by example. Your acceptance tests document and validate these requirements.

Cucumber

Write your acceptance criteria and document your requirements in one step.

Remember stories are not requirements, they are work items. Requirements are coded and documented in your acceptance tests.

Continuous Integration

Agile means reacting to all changes. Lean means build-in quality. Your continuous pipeline certifies after each change that your solution is correct and has the correct behavior.

Gradle

Is the new standard for building your artifacts.

Jenkins

Is the regular tool for a local build pipeline.

If you have a choice, use cloud continuous integration for team activities. Decide if you prefer the cloud approach – e.g., with TravisCI or Bitbucket Pipelines – or Jenkins within Docker for your local integration activities.

A smooth running CI pipeline is a boost for team morale and motivation. To know that your application compiles and runs through all your automated tests anytime is such a good feeling.

Continuous Delivery and Deployment

Agile means to react fast the changes. You need to deliver and deploy your new solution to your customers in minutes or hours.

Docker

Virtual machines were a powerful construct, but they are slow and require a lot of resources. Docker is the answer. Start a testing instance in 300 milliseconds.

Tomcat and TomEE

Embedded servers deployed in docker image in a matter of seconds. Actual versions of Tomcat or TomEE support it.

hsqldb

In memory database are lighting fast for integration and test staging areas. Later move to a file-based database for the production area. The database shall be a configuration of the delivery and not require a new compilation.

Google Cloud AppEngine

To deploy various integration environments

Design Quality

Important decisions and overall structure shall be documented for new team members. Architecture Decision Record: A lightweight approach to document architecture decisions using Markdown documents and traceability through the decision history. UML: Sometimes I would like to draw some UML diagrams to explain a design solution. I am currently using a textual tool to define such diagrams.

Development Quality

Analog Board

Nothing beats a big analog board in the team room as the ideal information radiator. If you want your collocated to work effectively and solve problems as a team, you shall provide an analog board.

Trello, Stride, BitBucket, Hipchat and others Atlassian tools for distributed teams

The first advice is to have a collocated team. The productivity and performance gains are unachievable in a distributed environment. We have splendid experience with Atlassian and Google App solutions to mitigate the drawback of distributed teams and establish a team work atmosphere. But a collocated team still rocks.

IntelliJ IDEA

This IDE is the choice for refactoring, powerful context suggestions, and efficient programming. If you use Eclipse or Netbeans, try a friendly refactoring competition with an IDEA convert.

Git

Stop using Subversion, CVS, TFS or other geriatric tools. Agile development, efficient integration in delivery pipelines, and extensive refactoring require Git. Migrate now to Git and leave dreaded message Tree conflict detected, please solve manually behind you.

These tools shall be used in the agile way. You do not have long-lived branches in your source code management tool. Either you have only short-lived local branches or better, you work against the trunk. Complex branching strategies are just a waste for a lean practitioner.

I decided to collect all these hints in the same long post. They belong together. Every day you design a solution, you code it, test it, improve it and guarantee its quality.

These activities are tangled together to producing a product you are proud of.

And yes, I know about GitHub, Gitlab, Slack, Ant, Maven, Jetty, Amazon Cloud, AssertJ, Subversion, Eclipse, Netbeans and other good tools. But I still prefer the ones above; I am convinced to be more productive with them.


1. After years of neglect, a Jimfs 1.2 was released in January 2021 and published on maven central.