Ensuring code quality when your software team is growing rapidly is a huge challenge. But even with a constant number of software developers, maintaining code quality can cause headaches.
Without tools and a consistent system, the whole project can accumulate a huge technical debt, causing more problems in the long-term than it solves in the short-term.
The best thing is that you don’t have to be a rocket scientist to avoid this (sure, it’s not a problem if you are one).
Some parts of this post might seem obvious, but the value lies in how the parts connect and build up a working code quality assurance system.
- What is code quality exactly?
- Why should you care about code quality?
- Build a code quality assurance system for your team
- How to measure tests
- User interface test
- Use continuous integration tools
- Post production code quality
Code quality is a group of different attributes and requirements, determined and prioritized by your business. Here are the main attributes that can be used to determine it:
- Clarity: Easy to read and oversee for anyone who isn’t the creator of the code. If it’s easy to understand, it’s much easier to maintain and extend the code. Not just computers, but also humans need to understand it.
- Maintainable: A high-quality code isn’t overcomplicated. Anyone working with the code has to understand the whole context of the code if they want to make any changes.
- Documented: The best thing is when the code is self-explaining, but it’s always recommended to add comments to the code to explain its role and functions. It makes it much easier for anyone who didn’t take part in writing the code to understand and maintain it.
- Refactored: Code formatting needs to be consistent and follow the language’s coding conventions. Some code refactoring tips here.
- Well-tested: The less bugs the code has the higher its quality. Thorough testing filters out critical bugs ensuring that the software works the way it’s intended.
- Extendible: The code you receive has to be extendible. It’s not really great when you have to throw it away after a few weeks.
- Efficiency: High-quality code doesn’t use unnecessary resources to perform a desired action.
A quality code does not necessarily meet all of the above-mentioned attributes, but the more it meets, the higher its quality. These requirements are more like a priority list that depends on the characteristics of your project.
Great authors write books with compelling stories that are easy to read and understand. From some aspects, the job of an author is similar to that of a developer. The main difference is that developers use different jargon.
As the author’s writing has to be easy to read and comprehensive, so should a software developer’s code.
I know it’s pretty hard to pay attention to code quality when you’re under pressure to meet your next deadline, but if you’re thinking long term, you definitely need to produce code that’s readable and maintainable. Here are three main reasons why code quality is important:
- Readability: Make the code more readable and easier to comprehend for everyone working on the project. It’s much harder to read and understand a bad quality code than to write it.
- Maintainability: It’s easier, safer and less time consuming to maintain and test quality code.
- Lower technical debt: Good quality code can speed up long-term software development since it can be reused and developers don’t have to spend that much time fixing old bugs and polishing code. It also makes it easier for new project members to join the project.
In this part, I will show you how we use version control, style guides and automated testing to make sure our code meets our predefined quality standards.
By following these methods, you can easily replicate our system and radically improve the code quality your team produces. You just need to go through the following steps:
- Setup version control
- Determine conventions
- Run functional quality tests
Version control tool is the foundation of our system.
The most popular tool for version control is Git. It also offers a branching style guide called GitFlow that enables seamless collaboration between team members and makes it easy to scale up the development team.
It provides an easy-to-track system that separates live product from the less stable developer branch with unpublished features.
When a developer from our team finishes a feature, he/she sends a pull request on GitHub. This describes the content and the details of the request.
This system makes sure that no unreviewed code will be merged with the master branch.
Here is how our process looks:
- One team member sends a pull request to the development branch.
- This will appear in a ready-to-review section waiting for a project member to review (peer review).
- A team member reviews the request, and if it meets the requirements, it will be merged to the development branch.
This a great system for controlling versions and making everyone’s work fully transparent. There are lots of GUI extensions for Git, such as GitKraken, which supports Gitflow. Here you can see how easily can enable it.
But how do you decide if the code is good enough?
In the next part, I show you tools to track code quality and metrics that can be used to measure code quality.
A style guide is a collection of best practices and conventions. Using a style guide ensures that every developer's code looks exactly the same, making the code easier to review and work with.
Fortunately, you don’t have to create your own style guide. There are many style guides available for free, focusing on different programming languages and scopes:
- Project: There could be varying conventions across different projects or products within a company. We don’t really recommend project-based style guides since it makes it much harder for people switching between projects.
Use linters to automatically test code style
Linter is part of the style guide. It’s a small piece of software that automatically checks if your code meets the predefined code convention rules. You don’t have to manually go through the code base to check style.
There are linters available for almost every programming language, just to mention a few:
- TypeScript TSlint
- Python pylint /flake8
- Sass/SCSS sass-lint
- Go golang lint
- Bash ShellCheck
Many code editors have support for configurable linting, such as VSCode. Here is a guide how to set your own linter up.
EditorConfig helps developers define and maintain consistent coding styles between different editors and IDEs. The EditorConfig project consists of a file format for defining coding styles and a collection of text editor plugins that enable editors to read the file format and adhere to defined styles.
While style guides with linters test how your code looks, functional quality tests show if your code actually works or not.
This simple pyramid shows how a test process should be structured and your efforts directed.
Generally, we can say that you have to run lots of unit tests, fewer integration tests and even fewer end-to-end tests.
With a unit test, you inspect one module of the software by mocking out dependencies. An integration test shows how different components work together while an end-to-end test checks the full client-server round.
Here are two funny visual examples why integration tests are necessary:
For running unit and integration tests, here are some tools you can use:
For end-to-end tests, we recommend using:
Additional reading: Mobile Labs Inc put together a cool checklist to use before deploying any applications.
The best way to measure test effectiveness is to track test coverage. It shows what portion (%) of the code is covered by the testing algorithm. To get a better understanding, it’s worth breaking down test coverage:
- Statement coverage (%): number of statements executed during a test divided by all statements
- Branch coverage (%): number of executed conditions divided by all conditions
- Function coverage (%): number of executed functions divided by all functions
- Lines coverage (%): number of lines ran during a test divided by all lines
UI test can be also automatized but it demands more resources, especially when the component is changing so the full test environment should be rewritten.
There are cool applications for automated user interface tests: Monkey test for Android UI stress tests, Saucelabs for cross-browser testing and Protractor for a more comprehensive end-to-end test (including user interface).
Our philosophy is to always have feedback on the condition of the software we’re building. This is where continuous integration comes in the picture. One of our favorite bloggers, Martin Fowler, nailed the definition of continuous integration.
“Continuous integration is a software development practice where members of a team integrate their work frequently, usually each person integrates at least daily - leading to multiple integrations per day. Each integration is verified by an automated build (including test) to detect integration errors as quickly as possible.” Martin Fowler
Here is our process:
- Continuous integration platform will run the linters on the code. If it fails, the process will stop here and the developer have to fix the style-related issues.
- It will run the functional test and move to the next step if the code runs according to plan.
- Then it starts calculating test coverage. If it doesn’t meet the predefined threshold, it will fail.
- If the request is merging with the master branch, it should be deployed as well.
Guide: Here are 10 things you should consider while choosing a CI platform.
Here is how to integrate Shippable with Github:
1. Go to Shippable.com
2. Log in with Github
3. Select the team you want to work with
4. Click enable project
5. Select your project from the list
You really shouldn’t let quality tracking go after the product goes live. Tools such as Sentry and Newrelic real-time monitor errors so you don’t have to ask users to report crashes and bugs since you will be notified automatically. All you need to do is add a small piece of code to your app.
That’s it. This is how you can create a workflow for your team that makes sure everyone’s code follows the same style guide and goes through predefined tests to check functional quality.
Apart from testing code quality before release, you also need to monitor what happens with your app when users start using it.
We know that writing bug-free code is impossible, but following the process mentioned above will definitely improve the quality of your team’s code.
This post was written by Akos Szokodi, Balint Csak and Tamas Torok