Peer code reviews as a process have increasingly been adopted by engineering teams around the world. And for good reason — code reviews have been proven to improve software quality and save developers' time in the long run. A lot has been written about how code reviews help engineering teams by leading software engineering practitioners.
My favorite is this quote by Karl Wiegers, author of the seminal paper on this topic, Humanizing Peer Reviews:
Peer review – an activity in which people other than the author of a software deliverable examine it for defects and improvement opportunities – is one of the most powerful software quality tools available. Peer review methods include inspections, walkthroughs, peer deskchecks, and other similar activities. After experiencing the benefits of peer reviews for nearly fifteen years, I would never work in a team that did not perform them.
It is worth the time and effort to put together a code review strategy and consistently follow it in the team. In essence, this has a two-pronged benefit: more pair of eyes looking at the code decreases the chances of bugs and bad design patterns entering your code base, and embracing the process fosters knowledge sharing and positive collaboration culture in the team.
Here are 6 tips to ensure effective peer reviews in your team.
Code reviews require developers to look at someone else’s code, most of which is completely new most of the times. Too many lines of code to review at once requires a huge amount of cognitive effort, and the quality of review diminishes as the size of changes increases. While there’s no golden number of LOCs, it is recommended to create small pull-requests which can be managed easily. If there are a lot of changes going in a release, it is better to chunk it down into a number of small pull-requests.
Code reviews are the most effective when the changes are focused and have logical coherence. When doing refactoring, refrain from making behavioral changes. Similarly, behavioral changes should not include refactoring and style violation fixes. Following this convention prevents unintended changes creeping in unnoticed in the code base.
Automated tests of your preferred flavor — units, integration tests, end-to-end tests, etc. help automatically ensure correctness. Consistently ensuring that changes proposed are covered by some kind of automated frees up time for more qualitative review; allowing for a more insightful and in-depth conversation on deeper issues.
A change can implement a new feature or fix an existing issue. It is recommended that the requester submits only those changes that are complete, and tested for correctness manually. Before creating the pull-request, a quick glance on what changes are being proposed helps ensure that no extraneous files are added in the changeset. This saves tons of time for the reviewers.
Human review time is expensive, and the best use of a developer’s time is reviewing qualitative aspects of code — logic, design patterns, software architecture, and so on. Linting tools can help automatically take care of style and formatting conventions. Continuous quality and static analysis tools can help catch potential bugs, anti-patterns and security issues which can be fixed by the developer before they make a change request. Most of these tools integrate well with code hosting platforms as well.
Finally, be cognizant of the fact that people on both sides of the review are but human. Offer positive feedback, and accept criticism humbly. Instead of beating oneself upon the literal meaning of words, it really pays off to look at reviews as people trying to achieve what’s best for the team, albeit in possibly different ways. Being cognizant of this aspect can save a lot of resentment and unmitigated negativity.