9 Best Practices for Effective Code Review
Code review, when done properly, allows developers to deliver optimal source code while learning more about effective code writing and debugging.
Join the DZone community and get the full member experience.Join For Free
The code writing process requires great mental effort and concentration, which makes it very difficult to achieve a flawless result. However, it’s important to remember there is no such thing as perfect code — programmers should strive to achieve "better" code that constantly can be improved. Reviewing the code author’s work can’t guarantee the prevention of all bugs, but it can protect users from experiencing it. That’s why integrating the code review into the development process helps to deliver high-quality applications that companies expect to get while outsourcing their software development.
Code review is a systematic examination of software source code used to detect mistakes, spot early bugs, track possible defects, and improve general code quality and consistency — this process helps code remain more maintainable. Conducted by fellow programmers and quality assurance specialists, code review aims to accelerate and streamline the software development process. The clean, well-designed, and easy-to-maintain codebase is the foundation for any high-quality, secure, and reliable application.
9 Best Practices for Effective Code Review
It’s important to make sure that team members are all clear about the rules and guidelines on how to conduct code reviews in the company. Peer code review is about combining forces for higher productivity, not competition.
#1. Know What to Look For
When it comes to the review process, the developers have to know exactly which things they should cover. These are design, functionality, style, logic, structure, consistency, test coverages, the complexity of the code, etc. Some of the aforementioned characteristics of code review can be examined automatically due to static code analysis (e.g. structure, logic), while others (e.g. design, functionality) require manual reviews.
To make the code review process more efficient, developers who check the code after their peers should focus on the following questions:
- Can it be clearly understood what the code does?
- Does the code function as it was expected?
- Does the code follow regulatory compliance?
If they can answer these questions with a confident "yes", the code can be regarded as a good one.
#2. Automate Before Review
It’s worth paying attention to continuous integration, which is the practice of running a series of automated tests while building and testing the code each time the change occurs. Continuous integration must be conducted before a peer review. When a series of automated tests pass, developers receive a code with fewer errors and the process of manual check becomes smoother and quicker, saving developers time.
#3. Limit Review Sessions
The effective review can only be conducted when the developers are 100 percent focused on the performing task. Most studies show that when people are engaged in any activity that requires high concentration, their effectiveness tends to drop off after 60 minutes. Code reviews don’t have to be rushed. It’s better to break this process into short sessions that will give developers an opportunity to reset, thus improving the quality of the codebase.
#4. Check Under 400 LOC per Hour
A single line of code (LOC) reviewed improperly can have bad consequences for the entire system. That’s why trying to check as many lines for one session as possible can lead to the failure of the whole development team. Code review cannot be done in a hurry — otherwise, it loses its point. Trying to focus on a maximum of 400 lines for each review session will result in a more rigorous and efficient review process.
#5. Give Constructive Feedback
As mentioned before, peer code review aims to enhance the team’s productivity, not to incite competition between code author and reviewer. The code has to be reviewed anyway. And if developers perceive it as a learning process, not a criticism of their work, it contributes to the overall project success. Receiving constructive feedback will motivate developers to learn from their mistakes and expand their capabilities. Also, reviewers are welcome to leave comments with the prefix “Nit:”. It means they don’t necessarily have to be fixed by code authors. Such comments have an educational purpose and help developers to continuously polish their skills.
#6. Set Up Goals and Capture Metrics
If the goals of the code review process are defined in advance, it’s much easier to measure code effectiveness and decide if the code review brings value and helps to achieve expected results. Setting up external and internal metrics using SMART criteria allows developers better access code quality. Examples of external metrics can be “reducing the percentage of defects reported by end-users in half”, or “decrease the defects found before the product launch by three times”.
Internal metrics include:
- Inspection rate: the speed of the code review process
- Defect rate: the average number of errors detected during one review session
- Defect density: the average number of errors detected per line of code
#7. Annotate the Code Before the Review
Code annotation means that code authors look back at what they have coded and leave explanatory comments for the reviewer. The annotation guide provides information about what each code line or section is trying to accomplish. It also can show the reviewer what changes were done, which code modification methods were used, and the reasons why doing so was beneficial. This practice offers a deeper understanding of the code by the reviewer and simplifies the overall code review process.
#8. Use Checklists
Let’s not forget about the human aspect of peer review, as some parts of the code might be overlooked by the developer. A checklist will make the review process more consistent, as it will be a constant reminder of what should be reviewed. It’s especially useful for the reviewer as it helps to move through the review process with specific criteria in mind. However, there is also such a concept as a personal checklist. If the code authors are aware of their common mistakes and flaws in the work, they can compile a checklist to continuously improve the quality of their code. Also, the code review checklists are very important in detecting omissions that are the hardest defects to review simply because they are not there.
Some checklist items include:
- Is the code easy to understand?
- Does code formatting meet the agreed project convention?
- Is the code structure modular enough?
- Does the chosen solution match the requirements?
- Did we check for all logic and bugs?
- Are there any security leaks?
- How is its performance? (Is there any moment obvious for optimization?)
- Is all code covered by tests?
- How effective are PR formatting, naming, and descriptions?
#9. Include Everyone
The more eyes are there to catch a bug — the better. People tend to perform better if they know their work will be reviewed. It doesn’t matter what the qualification level of the software engineer is — everyone should be included in the code review process. Junior developers can learn new techniques and alternative methods of doing something from their senior colleagues, while seniors can perfect their programming skills by writing a code understandable and readable for everyone.
A solid code review process allows companies to get their projects on time and bug-free. Code review is one of the most important parts of web and mobile software development services. It should be integrated into the development team’s workflow and how the team works on making code review and tooling more efficient.
Published at DZone with permission of Dariya Bogretsova. See the original article here.
Opinions expressed by DZone contributors are their own.