I decided to try Wayland due to some issues I suspect might be related to Xorg. So far, so good. However, it comes with its own problems. Spare layouts don't seem to be working as expected. It's a super useful feature for those who have more than two layouts because it makes cycling through them much easier. According to the documentation:

Spare layouts allow to toggle between small number of layouts easily while keeping more layouts handy close by. For example you might use 3 languages: English, Ukrainian and German but first two are used often and third one just occasionally. In this case you can configure first two as main layouts and German one as spare one - when you toggle with keyboard and left mouse button the switch will happen between main layouts only but you can always choose 3rd layout with context menu.

Here is my configuration:

I want alt+shift to only switch between English and Russian. Unfortunately, this setup doesn't work as intended: all layouts are in the rotation. Therefore, switching from Russian to English takes an extra step through German. Ideally, I want to use German only when I'm practicing it via the applet on the taskbar.

I found a workaround for this bug!

Playwright is a relatively new automation library for browser testing. Despite that, it has already gained some user base, with articles and tutorials regularly appearing on Software Testing Weekly. It comes with many features necessary for UI tests, but crucially, it can also work with API requests made by the web app under test. This allows us to either completely mock responses from a backend server, or alter them - depending on what we want to achieve. In this post, I will show how to do both things in Java.

Aviation professionals say that a fire on board is one of the most dangerous emergencies that can occur during a flight. While I don't have any expertise in this area, I concur that being burned alive is something I'd prefer not to experience. Recently, a passenger's bag with a power bank and e-cigarettes ignited during boarding, forcing an emergency evacuation. This made me wonder: what would happen if my phone or laptop caught fire while the aircraft was airborne?

Probably most of us have used software to store data in a spreadsheet and visualize this data. However, these tools can offer much more than that! Recently, I learned that LibreOffice can find out a mathematical function which can be used to predict future values. For instance, based on historical data, we can try to forecast the growth of the average salary over time. Naturally, the accuracy of this method is not guaranteed, but that is not the point of this tutorial.

I believe one of the best ways to learn a language is to use it every day. However, it can be hard to incorporate a new language into your life at the beginning. You simply don't know it well enough to chat online and read news. This leaves you with rather boring learning materials. I've tried to solve this problem with LLM chatbots, and so far, the results are rather encouraging!

Today, aviation accidents are quite rare, especially in small countries like Estonia with our relatively low volume of air traffic. Nevertheless, one event resurfaced in my memory a few days ago. What made it interesting for an average person like me was the fact that the involved aircraft landed on the then frozen Ülemiste järv, which isn't only a lake next to the Tallinn Airport but also the main source of water for the city.

A couple of months ago, Amazon raised prices for their Lightsail VPS service which I was using. I decided to move to another provider to save some money and get more performance. My Lightsail plan wasn't that powerful at all, and eventually I got x4 RAM with Hetzner for almost half the price!

In the post you'll find the table I used to compare various providers.

I was looking for a tutorial about photo editing in GIMP. I found a video that touches multiple tools and explains how to use them, so I thought it could be useful to jot down a checklist based on that info. Most of the mentioned tools are located in the colors menu. For details, please refer to the video.

Docker compose introduced a new feature in a recent update called the compose menu. It adds an extra line in the terminal while running docker compose up in the attached mode, i.e. without the -d flag.

Image source: https://github.com/docker/for-win/issues/14021#issuecomment-2161370349

The issue is that the menu breaks the active session. In my case, after stopping the containers, I could not see what I was typing in the terminal, as if I was entering a password: the text was not visible.

As this menu doesn't bring anything useful to the table, I figured I should be able to disable it. According to the documentation, there are two ways to turn it off.

DuckDuckGo, a privacy-friendly alternative to Google and other search providers, has recently launched a new product: AI Chat. It offers anonymous access to popular AI models, including GPT-3.5, Claude 3, and open-source Llama 3 and Mixtral. While I cannot make a reliable assessment of their claims, using this service allows you to work with these models without registration, which is a good starting point.

I decided to compare them to ChatGPT 4o. There are many ways to do this, but I didn't aim to make a professional and thorough comparison. As a user of these tools, I wanted to see how they could handle my daily requests. Since I am learning German, sometimes I need to clarify certain words, phrases, or how to apply different cases in various situations.

The prompt was inspired by my mistake on Duolingo. To put it simple, I thought that the German "in" was equivalent to the English "to". However, it turned out that "in" can change its meaning depending on the case.

Let's see how various LLMs explained the difference.

I decided that if this is my blog, then I should be able to post anything, including silly memes!

I want to recommend a couple of courses on Udemy that I took over the last couple of months. Both helped me gain confidence and knowledge in tools that I use every day: Git and Docker.

The state of racing simulators on Linux today is not as bad as I expected. Last time I tried them about ten years ago, it was still a plain WINE without DirectX to Vulkan translation layers, Valve hadn't released Proton yet, and other improvements by numerous contributors from the open-source community weren't available. Games like rFactor simply didn't work at all.

Recently, I decided to give it another shot. Even though Linux still might not be a viable option for all simracers, I found it doable!

Music streaming services provide us access to huge libraries, which is great! Algorithms study what we like and offer useful recommendations or let us discover something entirely new.

However, we can only access what has been officially released. Some "live in ${city.name}" DVD release that you enjoy might not be available on a streaming service. Occasionally, artists and labels end up in legal battles, and the content that was released in the '80s and '90s suddenly can't be distributed anymore. Over time, I started to notice that a playlist with my favorite music lost at least a third of the tracks I liked to listen to! That makes other sources of obtaining music relevant again.

YouTube is one of them. The quality is not fantastic due to irreversible compression, but there are so many live recordings, obscure releases, and covers that I find it worth it. As I store music on my home server, I needed a solution that wouldn't pollute it with hundreds and hundreds of packages. Yes, I'm looking at you, ffmpeg! Probably most tools that can download and process videos from YouTube depend on ffmpeg.

So if you too want to utilize yt-dlp in an isolated environment, I can offer dockerization!

I recently started playing racing simulators and joined a league that mimics the real Formula 1 championship. These cars are really challenging and unpredictable, especially with my experience. Another key aspect to consider is tire strategy. There are several compounds that offer different levels of traction, wear, and overall speed. A good strategy can help you gain positions, while a bad one can undermine solid driving.

I decided to create a tool that would help me to find out the most common strategies.

In March, I wanted to add code generation to a project I was working on. I had a publicly available OpenAPI contract of considerable size, and using it I needed to integrate with a third-party service. I decided to stick with the generator I was already familiar with. Unfortunately, it was incompatible with the specification at my hands!

A couple of years ago, I created a simple website designed to assist QA engineers in practicing some of the most commonly used test design techniques: Equivalence Partitioning and Boundary Value Analysis. The site provides a specification of a feature along with multiple implementations; one is correct, and the others are not. The objective was to show that the correct test input can effectively detect all errors. However, my initial intent was not clear to the end users because I did not provide correct answers or any explanations! Let me do that in this post.

To no one's surprise, posting regularly to a blog when you have a job and when you don't are two different things. Nevertheless, I'm trying to reincorporate writing into my daily routine. If you're reading this, then I have succeeded! In the previous entry, I briefly described that I settled on a Go backend and a React web app for this site, explaining how code generation based on the OpenAPI specification simplifies the synchronization of any changes between the two.

Once the code is written and tested, it's time to deploy it to my VPS. I used a couple of tools for that, with the actual commands stored in Makefiles. For the Go service, it looked like this:

1. Build a new Docker image.
2. Save it as a tar.gz archive.
3. Copy it to a remote server using scp.
4. Load this image to Docker.
5. Update the running container.

While I believe there's nothing wrong with this approach given the nature of the project, I wanted to switch to GitHub Actions and Packages. That would eradicate the need to define deployment properties on every machine where I code, let alone maintaining them up to date if. Moreover, even though the exact tools differ from company to company, this approach would be much closer to what a real business would do with their product. In other words, besides the personal interest, it is relevant to my job!

The transition was rather simple. I was already building Docker images, so all I had to do was to move this part to GitHub and push images to their registry. Here, I want to document what I did and think about what can be done next.

When I created this blog last summer, I had several goals in mind: to acquire new skills, learn about fresh technologies, and have fun in the process. Its simplicity allowed me to rewrite the project multiple times using different programming languages and libraries: Rust with Axios, Java with Next.js, Java with Thymeleaf, and finally, Golang with React. While the first two options reduce the amount of code by handling both the service logic and the HTML rendering, they also take on too much responsibility in this way. Not that this is wrong by definition, but I prefer to separate these concerns.

However, this approach increases the workload. I had to build models on both sides, consider not only the controllers but also the client implementation, where changes on either side can break the integration. In this post, I want to explore a way to eliminate this overhead and explain how to generate code for a Go server and React client based on an OpenAPI specification. The idea is not limited to this stack, it can be easily applied elsewhere with suitable tools.

The full demo project is available on GitHub.