Why Only Pull Requests Are A BAD IDEA

Continuous integration is a crucial aspect of software development that can have a major impact on the efficiency and effectiveness of a team. In this blog, we delve into the benefits of maintaining a clear view of how changes integrate with the rest of the codebase and explain why trunk-based development is the most effective approach.

The Shocking Truth About Pull Requests: Why They Could Be Holding Your Team Back

As a devops company, we often begin our discussions by acknowledging that software can be complicated. It's essential to proceed with caution, as we're almost always "running with scissors." At our company, we place a strong emphasis on continuous integration, which allows us to maintain a clear understanding of how our changes work alongside everyone else's as we make progress.

According to the "State of DevOps" report from the DORA group at Google, trunk-based development leads to higher throughput, stability, job satisfaction, and lower burnout. This approach also allows for fast and clear feedback, which is crucial for continuous delivery.

Code reviews and pull requests are commonly used in software development, but they can be a barrier to getting fast and clear feedback. Some people can't even imagine code review without pull requests, but it's important to understand that continuous integration values fast and clear feedback. The only way to get a definitive answer is to try out changes and see if they work with everyone else's changes.

Continuous delivery is even more demanding in terms of the need for fast and clear feedback. The data shows that working in many small steps and getting a clear answer after each step is the most effective way to develop software. It doesn't matter if you personally like this approach or not, the data speaks for itself. If we are being rational, then it means we need to optimize our processes to deliver this level of feedback on a consistent basis.

One of the biggest barriers that teams face when trying to achieve this level of feedback is injecting more people into the feedback cycle. Code reviewers, testers, and other team members can be valuable, but they can also be slow. Let's consider a classic pull request approach: I make a change on a feature branch and create a pull request asking you to take a look and approve my change. You're probably busy and need to finish whatever you're working on right now before you can look at my pull request. Asking how long this takes is obviously an impossible question, but let's be generous and say it takes 10 minutes for you to wrap up your current train of thought and start looking at my pull request. What do I do for that 10 minutes? Actually, it's going to be longer than 10 minutes because you might not be the only person reviewing my change.

There are many ways to organize code reviews, but the problem is that if you practice continuous integration, you want a definitive answer to the question "Does my change work?" multiple times a day. The only way to get that definitive answer is to try it out and see if it works with everyone else's changes. This can be difficult to achieve if you rely on code reviews or pull requests, as these processes can be slow and cumbersome.

Instead, we recommend working in small steps and getting a definitive answer after each step. This allows us to continually and consistently deliver high-quality software. It's important to be rational and approach software development problems in a way that optimizes for fast and clear feedback. By doing this, we can ensure that we are consistently delivering high-quality software and meeting the demands of continuous integration and continuous delivery.

But how do we actually achieve this level of fast and clear feedback? One way is through automation. Automating tasks such as testing and deployment can help us get feedback more quickly and efficiently. This doesn't mean that we should completely eliminate the need for human intervention, but rather that we should focus on automating the tasks that can be automated in order to free up time for the tasks that require human expertise and judgment.

Effective collaboration and communication are crucial for achieving fast and clear feedback in software development. When team members work closely together and are open to feedback and suggestions, they can identify and resolve problems much more quickly. This can be achieved through a variety of methods, including:

  • Pair programming: Pairing up with another team member to work on a task can provide an opportunity for real-time feedback and collaboration. Pair programming can also help improve code quality, as it allows for more thorough review and testing.
  • Regular team meetings: Regular meetings, whether in person or online, can be a valuable opportunity for team members to share progress and get feedback from one another. These meetings can be used to discuss challenges, brainstorm solutions, and identify any issues that need to be addressed.
  • Collaborative tools: Tools like Jira, Asana, and group chat apps like Slack can facilitate collaboration and communication among team members. These tools allow team members to share updates, assign tasks, and provide feedback in real-time.
  • Peer review: Encouraging team members to review each other's work and provide feedback can help identify and fix issues in a timely manner. This can be done through code review tools like Stash or through more informal methods like pair programming or regular team meetings.

Overall, fostering a culture of collaboration and communication within the team is key to achieving fast and clear feedback. By working closely together and being open to feedback, teams can identify and resolve problems more efficiently, resulting in higher-quality software.

Want to hear more about how we can help your business with a tailored DevOps approach?

Ready to Get Started?

Let's Change the future together.