Failures and downtime are part of our day-to-day life. I had a problem with one of the services that started crashing a few times a week. We noticed that it crashes because the memory usage reaches its limits no matter how high the limit is. Debugging memory leaks is hard and time-consuming. As a temporary fix[^Nothing is more permanent than a temporary solution] we decided to restart the application once a day.

Writing linters is simple. I was surprised how it’s easy to write a Go linter. Today, we’ll write a linter that will calculate the cyclomatic complexity of the Go code. What is cyclomatic complexity? Cyclomatic complexity is a software metric used to indicate the complexity of a program. ref The idea is simple - every time we find any control flow statements we increase the complexity by one. I know I oversimplified it a bit but I don’t want to overwhelm you with unnecessary details.

In the last article, we wrote a few tests for a project to make sure that our refactoring won’t break anything. To understand the project better, we will separate the part of the domain and add a test to it. This will make the test more authentic. There is a problem with the end-to-end (e2e) tests: database under the hood. This attitude is not carefree. Firstly, those tests are rather slow.

When we talk about software design we very often use very generic and abstract words. But, how about the practice? How does it look in a real-world project? Today, you and I will start refactoring a small to-do app for better testability and maintainability. In this article, we will make the application testable. We’ll write black-box tests that will prevent from some bugs and make future refactoring easier and safer.

In this article, I explain how you can detect if the interface you’re using is getting too big and requires splitting into smaller ones. Smaller interfaces help to improve the maintenance and readability of the code. What’s more, it helps with understanding the code. Interfaces in Go are different than those known in Java, c#, PHP etc. In those languages you define interfaces up-front. In other words, at the moment of creating a class you need to know how the class will be used.

The context package in Go is quite simple and well-known. On the other hand, there are some misunderstandings while using it. Today, I’ll try to explain all the most popular concerns and make more clear when and how use the Context. Let’s start with what the context is. Package context defines the Context type, which carries deadlines, cancellation signals, and other request-scoped values across API boundaries and between processes. ref: https://golang.

From time to time, I receive an email from a scammer that says he has X million dollars/euro for me. At the very beginning, I removed those emails but at some point, I decided to answer them. Here’s what I found. Every scammer starts very typically. There’s a very reach person who’s dying or very sick. They found my email on the Internet and learned that I’ll be the person who will spend the money wisely.

Go has plenty of different web frameworks. When you are faced with choosing a framework for the first time, it may turn out to be quite a challenge to choose the best one. This article is intended to help you choose the best one. It is full of personal judgments that you may disagree with. However, I believe you will find it most helpful. Martini The first framework is Martini. Honestly, it shouldn’t be here as it’s been under development since 2017.

Making changes to all HTTP requests can be handy. You may want to add an API key or some information about the sender like app version etc. No matter why you want to do that you have a few options to achieve the goal. The first approach is building a factory method that will add the required header. func newRequest(endpoint string) *http.Request { req, _ := http.NewRequest(http.MethodGet, fmt.Sprintf("https://%s", endpoint), nil) req.

When I started my career as a software developer and published the first production application what I did was staring at logs and look for some fatal errors. It was a monolith application. Every log saying that something’s wrong had to be fixed. ASAP. This approach worked for some time. However, when the scale increased, and I started building microservices, I couldn’t get rid of all of them. Network issues, database failures, and more - it happens all the time.