Over the years, I've grown fond of the Nord theme. It all started with the Solarized theme years ago. Then, when I discovered Nord, I found it to be as pleasing in the eye, and yet more elegant. In any case, since I still prefer the GNOME Terminal as my terminal emulator, I had to see how to configure things there. The following steps worked both on Debian and Arch Linux:

# Install dependencies on Debian 
root# apt install dconf-cli uuid-runtime git

# Install dependencies on Arch Linux
root# pacman -S dconf util-linux git

# Clone the repo
root# git clone https://github.com/nordtheme/gnome-terminal.git

# Run the script to configure things
root# cd gnome-terminal/src
root# ./nord.sh

That's it! It worked fine on both distros. I must say though that Debian did show some dconf warnings. But it still worked, it seems. {link to this entry}

The last few weeks, I wasn't able to upgrade the Debian 13 (Trixie) kernel due to the fact that I was using the open Nvidia drivers packaged by the Debian team. The failed to build, and I had to temporarily put a lock on the kernel I was running at the time. It is a well known issue with the open drivers that, in theory at least, doesn't occur with the proprietary drivers. So today, finally, I found the time to migrate from the open drivers to the proprietary ones. However, after I managed to do that, runnihg apt update and then apt list --upgradable failed to show any kernels ready to install, even though I know they were available. It didn't take much research to find out what had happened. At one point, while I was trying to figure out a way to solve my problem with the open Nvidia drivers, the linux-image-amd64 somehow got uninstalled, and I had to install it.

# apt install linux-image-amd64

That took care of the problem. {link to this entry}

Honest, the more I use LLMs, the more I realized that, amazing as it is, its use is quite limited, at least for my use-case. I'm sorry, but I've been to too many sessions with customers where we are troubleshooting a problem, they consult AI trying to be helpful, and recommend command after command that are completely hallucinated. Perhaps it's because I work in a niche market, the HPC segment. But that is the case. But the anecdotal evidence never stops. Yesterday, I had a JSON file that I had to use to put together a set of commands to configure switch ports. Since the amount of ports was pretty high (somewhere in the hundreds) and the file I had was a pretty well structured JSON file, I figured that AI would be great at this. Well, wrong! It did correctly identify how many ports were assigned to a particular port policy, but then it listed the wrong ports. So, I tried with a second port policy, and ran into the same issue. At that point, I thought to myself I could just go back to the old combination of grep, awk and pipes. But I thought I might as well ask AI for help. While AI thought, I searched in Google, and came up with the answer faster than the LLM. Worse yet, the LLM just answers my question. But the search engine shows a few documents that tend to be tutorials, which means that they also provide additional information on the topic and allow me to expand my knowledge. In other words, once again, I'm not sure I'm better off using AI for this purpose. Yes, it does have its uses. But, in my experience, the amount of cases where I find it useful keeps decreasing. {link to this entry}

While watching videos about the concept of an immutable Linux distro, it occurred to me that this is precisely one of Linux's strengths when compared to the BSDs. Precisely because Linux is in a constant flux, it is more chaotic, there is no doubt about that, but it's also where a lot of the innovation happens. Not only that, but the very fact that, as BSDers like to point out, "Linux is only a kernel", makes it easier to build distros that take very different approaches to things. All this may still be possible in BSD-land, but it's not as easy. Yes, the BSDs may be more stable. But Linux is where the action is. Take it or leave it. {link to this entry}

It looks as if GNOME may have disabled the middle click paste functionality by default in their most recent builds. The functionality is still there on my laptop running Debian 13 (trixie), but it's not working on Arch Linux, which has a newer version of GNOME. Running the following command re-enables it:

$ gsettings set org.gnome.desktop.interface gtk-enable-primary-paste true

{link to this entry}

I read on the InfoWorld website that the MATLAB programming language is sinking in popularity:

In the Tiobe index for August 2026, published August 8, MATLAB is ranked 25th, having fallen out of the top 20 for the first time in more than 10 years. “Is this just a temporary dip, or is it the beginning of a more permanent decline? Personally, I believe MATLAB’s days are numbered,” said Paul Jansen, CEO at Tiobe. One important reason for the decline may be MATLAB’s licensing model. “In a world where most programming languages are freely available, having to pay for a MATLAB license is becoming an increasingly significant disadvantage,” Jansen said.

Like it or not, the open source model is not only here to stay, but it appears to win against anything else in the long term. In the end, it's cheaper and easier to maintain code that way. {link to this entry}

A few weeks ago I wrote about a couple of minor issues I had with RSS. Yet, I must say that, after trying other approaches to my daily news feed, I ended up returning to RSS. In particular, I think it's worth mentioning my experience with AI. First of all, I must say that I don't truly understand the view some people are defending according to which getting your news from AI is preferable because one is less exposed to the distortions and manipulations of media outlets. I seem to remember Derek Taylor, the guy behind the popular DistroTube channel, stating something like that in one of his videos. I'm sorry, but I just don't understand the logic here. So, in order to avoid the distortions and manipulations of well established news media outlets that risk losing the reputation they accrued over decades of hard work I'm supposed to trust an AI chatbot provided by a single Big Tech company that completely controls the algorithms behind it and can do as it pleases with the LLM, aside from the fact that it cannot be held accountable for anything because they will just shield themselves behind the pathetic excuse that the model just "hallucinated" something? Am I understanding right?

I must say. I recently played a bit with using different AI chatbots (the main ones: ChatGPT, Claude, Gemini, even Lumo) to provide me with a daily news brief. At the beginning I was quite excited. It did provide a relatively short overview that I could quickly go through in a few minutes, get an idea of what the day looks like, and then ask further questions about any of the news shared with me. But then a couple of issues quickly moved to the front. First of all, as in the case of any other task AI may be trusted with, it also hallucinates when it comes to summarizing the daily news. In my case, I was lucky enough that the news was so preposterous that I was skeptical. It didn't take long for the LLM to acknowledge that it had just hallucinated the whole thing. Needless to say, that led me to consider what other news that I had been fed by the LLM were also fake. I distinctly remembered some that were way out of date. But one wonders how many others were just invented. Second, there is also the issue that, no matter how "personal" these chatbots might feel, the summary I was offered every day didn't truly respond to my interests. Actually, in some areas, it fell very short. And then, finally, of course, there is the issue that using a AI to get the news also means I'm not exposed to other headlines that may attract my attention even though they are not strictly related to one of my areas of interests (sort of like the idea of a sudden encounter of a book we had never considered when we visit a physical bookstore, instead of trusting an algorithm). One way or another, the thing is that I decided to return to RSS as the best way to get my daily news feed. {link to this entry}