DEV Community

Cover image for How to Review Pull Requests 3x faster

Posted on • Updated on

How to Review Pull Requests 3x faster

How much time do you spend on average for reviewing a pull request - 3min, 10mins, or 30mins?

Even experienced programmers spend between 10 mins to 30 mins on average to review a pull request from their colleagues on some project.

Throughout various enterprise projects (Typescript, Python, Golang, Java), I reviewed more than 100 pull requests and learned to review a pull request 3x faster and more accurately, and then get even faster with automated tool's help.

No.1: Use a pull request template

Have you ever thought the following in your heart?

What changes are made and what are not?

Where are the relevant documentation for functional requirements, specifications, and designs?

The solution to the above challenges is easy, but many newbie projects miss it.

Use a pull request template!!

It ensures that every pull request has the necessary information in sorted order.

At least, it's good to include the following points.

1. Description
2. A list of Changes
3. Future todos
4. Screenshots or movies
5. Test Cases and results
6. How to review your code
Enter fullscreen mode Exit fullscreen mode

You can check my repository for a complete pull request template.

No.2: Don't review business logic but prepare a test for it

Have you seen the following comments?

I tried your new feature on my local PC, but it did not work.


I checked your code, and the logic in line XX should be the opposite of the specification.

That would be generous but also too much consuming to guarantee business logic by manually checking the coding counterpart.

For example, look at the following control flow.

if (condition1) {
  //  10 lines of code here
} else if (condition2) {
  //  20 lines of code here
} else if (condition3) {
  //  20 lines of code here
} else {
  //  20 lines of code here
Enter fullscreen mode Exit fullscreen mode

To review the above code, you need to go back and forth between the pull request and documentation 3 or 4 times.

And the real problem of this code is not about the logic but many if/else statements.

Anyway, it is more efficient to use (automated) testings to assure functional qualities of your code.

The procedure is so simple.

1. Write test cases to a test sheet
2. Review if test cases are sufficient
3. Write corresponding test code
4. Run the test program in every pull request
Enter fullscreen mode Exit fullscreen mode

There are many automated testing frameworks in every programming language and CI/CD tools in code hosting platforms.

No.3: Configure a linter and a formatter properly

Have you seen the review comments like the below?

This variable should be a camelCase.


This JSON object should be indented.

Any project with comments like the above fails to introduce a syntax checker and a code formatter properly.

For instance, ESLint and Prettier are the de fact standard for javascript/typescript development. So it is important to configure eslintrc and prettierrc files properly.

Furthermore, it is also crucial not to allow exceptions without rationals. Think carefully about why a violated lint rule improves/secures your code, and your code could still be an exception.

The above No.1, No.2, No.3 practices make pull request reviews 3x faster than not.

It should take between 3 mins to 10 mins for a meaningful review.

You should only focus on:

  1. Readability
  2. Performance
  3. Abstraction

A useful rule of thumb is to keep every file less than 200 lines in the repository.

But how can you spend even less time for pull request review?

There is an automated code quality review tool.

No.4: Use Code Quality Review tool

A good Code Quality Review tool makes pull request reviews another 3x faster, so it is 10x from the beginning.

I prefer to use Codebeat because it gives a sufficient static code analysis, rich CI/CD supports, a managed service. Moreover, it allows us to focus only on poor but easily fixable code on their Quick Wins panel.

Image description

On the other hand, SonarQube is more configurable in-depth and would be helpful in the advanced project requirement.

I recommend two articles for detailed tool comparisons.

Thank you for reading, and good luck with your review!🙌

Top comments (0)