Let’s go for my web review for the week 2026-15.

France Launches Government Linux Desktop Plan as Windows Exit Begins

Tags: tech, foss, politics, desktop, france, europe

Well, what can I say? This is excellent news and I’m excited to see it happen. Let’s hope more governments do the same. It’ll take a while of course, so we’ll have to be patient.

https://linuxiac.com/france-launches-government-linux-desktop-plan-as-windows-exit-begins/

The Free Market Lie: Why Switzerland Has 25 Gbit Internet and America Doesn’t

Tags: tech, infrastructure, economics

A good explanation and illustration of how natural monopolies work. This is why you want to regulate infrastructure properly.

https://sschueller.github.io/posts/the-free-market-lie/

You can absolutely have an RSS dependent website in 2026

Tags: tech, blog, rss

The stats are clear there. Beside in term of experience, RSS feeds are so superior to newsletters… I wish more bloggers would give up on the newsletter focus. There’s also a good point in this post: as soon as you have a newsletter you will sit on a database of email addresses, it’s definitely a liability.

https://matduggan.com/you-can-absolutely-have-an-rss-dependent-website-in-2026/

The Downfall and Enshittification of Microsoft in 2026

Tags: tech, microsoft, github, apple, linux, business, product-management

Indeed, the giant managed to make itself weak. This means opportunities for other ecosystems to grow faster than before.

https://caio.ca/blog/the-downfall-and-enshittification-of-microsoft.html

Let’s talk about LLMs

Tags: tech, ai, machine-learning, copilot, productivity, craftsmanship

Long but very precise piece about why you can likely ignore LLM for development purpose. Starting from older Fred Brooks work is spot on. Indeed whatever will remain of LLM based tools in the years to come, it’s much smarter to focus on fundamental skills than chase the new tools. At least, I’m trying to do my share in getting myself and others better at the craft.

https://www.b-list.org/weblog/2026/apr/09/llms/

Almost Half of US Data Centers That Were Supposed to Open This Year Slated to Be Canceled or Delayed

Tags: tech, ai, machine-learning, gpt, energy, economics, infrastructure

It’s getting clearer that the industrial LLM complex will have a hard time meeting its targets.

https://futurism.com/science-energy/data-centers-construction-supply

“Cognitive surrender” leads AI users to abandon logical thinking, research finds

Tags: tech, ai, machine-learning, gpt, cognition, bias

It feels like it’s supercharging an old bias… We tend to confuse confidence for competence.

https://arstechnica.com/ai/2026/04/research-finds-ai-users-scarily-willing-to-surrender-their-cognition-to-llms/

The machines are fine. I’m worried about us.

Tags: tech, ai, machine-learning, gpt, copilot, learning, science, research

Excellent piece, it show quite well the problem of skipping the “grunt work”. Without it you can’t really learn your trade (be it astrophysics or anything else). It also shows how the incentives on scientific careers are wrong. It’s not new, but when LLM agents become available, things are definitely changing for the worst.

https://ergosphere.blog/posts/the-machines-are-fine/

Giving LLMs a Formal Reasoning Engine for Code Analysis

Tags: tech, ai, machine-learning, copilot, prolog, logic

Definitely interesting approach. I think neurosymbolic approaches are what we ultimately need so I’m probably biased. At least it means using LLMs for what they’re good at (language skills) and only that. Then rely on proper code symbolic models which do the reasoning heavy lifting. I’d expect it can give nice output with smaller models.

https://yogthos.net/posts/2026-04-08-neurosymbolic-mcp.html

Open source security at Astral

Tags: tech, security, ci, supply-chain

Lots of interesting measures to reduce the risk of supply chain issues. Definitely to be considered on your projects.

https://astral.sh/blog/open-source-security-at-astral

another memory corruption case

Tags: tech, hardware, memory, failure

Failing DRAM chips are real. Here is the case of debugging a single bit flip.

https://trofi.github.io/posts/347-another-memory-corruption-case.html

The Git Commands I Run Before Reading Any Code

Tags: tech, git, version-control, team, audit

Nice little commands to use to discover quickly the state of a code base… Or rather of its team.

https://piechowski.io/post/git-commands-before-reading-code/

Zsh: select generated files with (om[1]) glob qualifiers

Tags: tech, zsh, shell

Oh this is super neat and convenient! I didn’t know about those glob patterns modifiers in zsh.

https://adamj.eu/tech/2026/01/27/zsh-om1-glob-qualifiers/

Two little scripts: addup and sumup

Tags: tech, unix, shell, scripting

A friendly reminder that one can go far mainly with awk.

https://utcc.utoronto.ca/~cks/space/blog/sysadmin/LittleScriptsIX

All of the String types

Tags: tech, memory, unicode, encodings

So many string types! They all have a purpose of course. It’s a good reminder that something mundane like a string type is not that simple.

https://lambdalemon.gay/posts/string-types

Stamp It! All Programs Must Report Their Version

Tags: tech, version-control, debugging

Examples of how i3 and go stamp versions. This is indeed good habits to ease dealing with errors in production.

https://michael.stapelberg.ch/posts/2026-04-05-stamp-it-all-programs-must-report-their-version/

The MVC Mistake

Tags: tech, architecture, complexity

Shows the problem with layer cakes in applications or how you might want to go toward onion architectures.

https://entropicthoughts.com/mvc-mistake

The Mouse That Roared

Tags: tech, leadership, tests, tdd, agile, organisation

Cryptic title to be honest. But this is a good explanation of why any “agile transformation” better start close to the code and in particular with automated tests. If you can crack that nut (and it takes time), the rest will follow naturally.

https://codemanship.wordpress.com/2026/03/30/the-mouse-that-roared/

If you thought the speed of writing code was your problem - you have bigger problems

Tags: tech, productivity, organisation, leadership, ai, machine-learning, copilot

So much this… There are so many organisational problems that churning code faster is likely not what you need. When did we start to obsess with the number of lines of code?

https://andrewmurphy.io/blog/if-you-thought-the-speed-of-writing-code-was-your-problem-you-have-bigger-problems

Are We Idiocracy Yet?

Tags: satire, culture

Getting there, one day at a time.

https://idiocracy.wtf/

Bye for now!

I’ve been on the Akademy organizing team and contributing in various cat-herding capacities since 2023, but this is the first time I’ve joined other contributors for a Sprint.

My mission this week has been to scout locations and activities for the Akademy conference later this year. One of the members of our local organizing team let me (temporarily) adopt their stuffed Konqi, so I have been wandering around Graz and the state of Styria with a stuffed dragon taking a bunch of pictures, drinking Aperol Spritz, eating chocolate, and petting animals to make sure that all the places we visit in September will be fun and accessible for everyone who joins.

This year KDE turns 30, so we are planning a big celebration for Akademy. I have been thrilled to discover that Graz is very accessible. The town tourism website has a guide for navigating with a wheelchair or other mobility devices; many restaurants have mocktails or homemade juice/tea options for non-alcoholic drinks; the city is full of plazas you can sit and sip a coffee in for hours when you need a break from walking, and there is an abundance of parks and fountains that children can expel their energy playing in.

I can’t wait to introduce the KDE community to Graz this September!

Akademy 2026: Call for Participation

Akademy 2026 will be a hybrid event held simultaneously in Graz, Austria, and online. The Call for Participation is open! Send us your talk ideas and abstracts.

Why talk at #Akademy2026

Akademy attracts artists, designers, developers, translators, users, writers, companies, public institutions and many other KDE friends and contributors. We celebrate the achievements and help determine the direction for the next year. We all meet together to discuss and plan the future of the Community and the technology we build. You will meet people who are receptive to your ideas and can help you with their skills and experience. You will get an opportunity to present your application, share ideas and best practices, or gain new contributors. These sessions offer the opportunity to gain support and make your plans for your project become a reality.

How to get started

Do not worry about details or slides right now. Just think of an idea and submit some basic details about your talk. You can edit your abstract after the initial submission. All topics relevant to the KDE Community are welcome. Here are a few ideas to get you started on your proposal:

• How KDE can empower building robust communities in changing political climates
• Work towards KDE's goals: Streamlined Application Development Experience, We care about your Input, and KDE Needs You
• Giving people more digital freedom, sovereignty, and autonomy with KDE software
• Advice on how to participate for new users, intermediates and experts
• New developments/plans for KDE Frameworks, Plasma, Applications and other projects
• 30 years of KDE: achievements, highlights, and what’s next
• Anything else that might interest an audience of long-time, new, and potential KDE contributors!

To get an idea of talks that were accepted, check out the program from previous years: 2025 2024, 2023, 2022, 2021, 2020, 2019, 2018, and 2017.

For more details and information, visit Call for Participation.

In my opinion, KDE Connect is one of the finest pieces of software KDE has ever produced. It lets you easily pair your devices and makes them work together. Copy some text on your computer, paste it on your phone. A call comes in and the video you’re watching in your browser pauses. How cool is that? For the past couple of weeks I’ve been working on even tighter integration between KDE Connect and the Plasma desktop.

By adding a corresponding back end to Solid, KDE’s hardware abstraction framework, your phone’s battery status will show up in the “Power & Battery” popup just like a wireless mouse would. The key advantage of reporting a KDE Connect device as storage media is that it can show up in various places just like a USB drive would, including the Places panel in Dolphin and “Disk & Devices” in Plasma’s system tray. Right now, the Places panel entry is actually manually added by kdeconnectd creating a bookmark. This also means that any modifications done to the entry, such as hiding it, will be lost once the device disappears. If instead it is a proper Solid device with a unique identifier, the Places panel will remember that the device was hidden next time it is discovered. Additionally, PowerDevil, Plasma’s power management daemon, already automatically issues a notification when a connected external device runs low on battery.

Now that the fact that you can browse your phone’s storage remotely via KDE Connect is more obvious, I spent some time improving the user experience when doing so. Since KDE Connect is cross-desktop and cross-platform it uses sshfs to provide access to the phone. It doesn’t implement a proper KIO worker like we normally would since that really only works in KDE applications. This unfortunately comes with a couple of downsides: for example, the device is mounted into /run/user/<uid>/<deviceid> which is gibberish that will be shown to the user in the address bar. More importantly, though, it makes the file manager think it’s a local path (which is normally fast to access) and could lead to UI freezes when the connection is slow or unstable.

To improve this, I made use of KIO’s ForwardingWorkerBase. This is a tiny KIO worker that just rewrites a URL and forwards it to a different location. This way, the application sees a “remote” kdeconnect://device/path URL and keeps everything nice and asynchronous but under the hood still uses the sshfs infrastructure we already have. It’s what the magic desktop:/ URL uses that parses the name of .desktop files (so your get nice application names) but other than that merely reads from /home/<user>/Desktop. This also makes the entry in the Places panel match up with the URL being browsed and ultimately makes the device’s name appear on the address bar instead of an ugly UUID.

While at it, I also added a “KDE Connect Devices” link to the “Network” folder, like we already have for Bluetooth, MTP, and Apple devices. Finally, when there is only a single storage location on the device, such as “Internal Shared Storage”, it redirects into it automatically. This saves one click when opening the device and puts you directly where your stuff is. The back end has just been merged and will be released next month as part of KDE Frameworks 6.26. However, since there’s still a bunch of infrastructure work needed around it, the back end is disabled by default. We will likely need to have at least KDE Gear 26.08 and Plasma 6.7 released that will include some necessary changes before we can flip the switch.

If you run latest KDE git builds, please give it a try and let me know what you think! You need to set the SOLID_ENABLE_KDECONNECT=1 environment variable to use it. To enable all debug output to aid debugging, set QT_LOGGING_RULES=kf.solid.backends.kdeconnect*=true or use KDebugSettings.

Introduction

Another year, another successful Season Of KDE for 20 contributors! This article has been co-written with the input from all contributors.

Lokalize

During Season of KDE 2026, Tanish Kumar worked on cleaning up the UI in Lokalize, KDE’s translation tool. The main task was fixing an annoying bug where the menu bar kept jumping around whenever you switched tabs, which was solved by giving the menus a stable “skeleton” in the KXMLGUI .rc files. Along the way, a bunch of “ghost actions” were discovered - menu entries that existed only in XML - and actually implemented the missing ones like Cut, Copy, Paste, Alternate Translation, Save All, Revert All, and Close All. He also built a Bookmark Manager dialog so translators can see all their bookmarks in one place.

Navya Sai Sadu and Kumud Sagar fixed navigation inconsistencies in the Editor Tab where shortcuts like “Approve and Go Next” failed to respect active filters and custom ordering in the Translation Units view. They ensured consistent behavior across all related shortcuts, including Next Ready, Next Non-Empty, and Next Bookmark. They unified the traversal logic so navigation always follows the filtered and ordered entries and users of Lokalize can now navigate through Translation Units using the keyboard shortcuts or the options in Go menu properly..

Additionally, Kumud identified Lokalize’s on-disk file tracking mechanism and began improving its handling of external file updates.

Jaimukund Bhan fixed foundational issues with the Glossary file, ensuring it could be properly loaded, saved, and autosaved, and cleaned up the codebase by removing an obsolete Restore function that no longer made sense once autosaving was in place. Several UI bugs were resolved, including the editor failing to clear when all terms were deleted and incorrect pre-selection behavior when the app started on the Glossary tab. In the second half, Jaimukund improved the manual term addition workflow by replacing silent blank entry creation with a proper dialog prompt, preventing accidental accumulation of empty entries. He also fixed a broken keyboard shortcut for switching to the previously active tab by replacing an index-based tracking variable with a widget pointer, which remains valid even as tab positions change.

Aditya Sarna made a full Figma redesign for the Glossary tab, which was referenced in several places to implement comprehensive UI and UX improvements. The work involved close collaboration with translators to understand user pain points and iterating on designs based on feedback from the design group. It included replacing button text with icons and adding tooltips to the Add and Delete buttons to clarify their purpose. This was followed by designing a new delete functionality, which introduced a cross button for each entry to make deletion more intuitive. Several additional UI changes were implemented, including shifting elements and improving the overall structure and layout of the Glossary tab. Furthermore, the workflow for adding new terms was refined by moving note text into the dialogue box and enhancing its layout, resulting in a cleaner, more intuitive and user-friendly interface.

Varun Sajith Dass worked on improving the proofing capabilities of Lokalize and implemented a reactive character consistency check that alerts translators in real-time when special characters are mismatched between the source and target strings. This involved debugging Qt UI signals, resolving macOS build issues with KIO workers, and creating a persistent status bar warning system to enhance the overall translation workflow.

Vishesh Srivastava worked on adding Appium-based UI testing to KDE’s Lokalize. Vishesh started with a small bug fix and unit test to familiarize with the codebase and then built a complete Appium test from scratch, including basic tests and a full end-to-end workflow. Another thing done was adding accessibility ids in the UI so Appium could interact with the editor. The tests were integrated into the CMake system, ensured they ran independently of the user, and can run with kde-builder tests with a flag. By the end, Vishesh had developed a functional UI testing framework for Lokalize, along with documentation to help future contributors.

KDE mentorship website

Advaith Sathish Kumar project was transforming mentorship.kde.org into a proper onboarding system for new KDE contributors. On the homepage, the hero section was redesigned, placeholder routing with experience based navigation was replaced, social media links were added, and the news cards to include author, date, and tags were also redesigned. For the /mentees page, I added past mentee details, implemented pagination, and added client-side filtering by year, program, and technology.

Aryan's project was to make mentorship.kde.org better so that new contributors who want to work with the KDE ecosystem have a better onboarding experience. As part of this effort, a new "/programs" page was added. It gives a structured overview of the main KDE mentorship programs and links to help newcomers find their way around the opportunities more easily. He also changed the card template to better organize the repository, making it more structured, easier to maintain. Aryan also filled the /resources page with more useful resources for new contributors.

Documentation website

Mohit Mishra worked on decoupling the bundled dblatex fork from the docs-kde-org repository and fixing PDF generation for Chinese, Japanese and Korean (CJK) languages for KDE documentation. This involved switching the TeX engine to XeTeX from pdfTeX and then re-integrating the KDE styles. The outputs are now nearly identical, with CJK languages rendering correctly as well. There is still work in process to ensure there are no regressions and we can officially switch to this generation.

Scripty

Aviral Singh and Keshav Nanda worked on fixing KDE's translation tools so translators can easily locate where text belongs in the code. Keshav corrected the underlying logic to make sure these paths are always accurate, and Aviral built an automated testing system to validate the fix across KDE projects.

Marknote

Siddharth Chopra worked on Marknote to add source mode for notes. The Source Mode essentially allows users to bypass the rich-text WYSIWYG interface and directly edit the raw markdown. While working on this feature, Siddharth also made a major refactoring of the codebase (on both the QML and C++ sides). Spell checking using Sonnet was also added, among other small fixes and improvements.

KDE Eco

Hrishikesh Gohain worked on setting up KEcoLab's measurement environment with Wayland on Fedora 43 KDE Plasma Desktop. He ported the Okular measurement scripts from the X11-based xdotool to ydotool and kdotool, which work on both X11 and Wayland. His contributions can be found in this merge request. The original project plan had included measuring the KDE Plasma desktop environment itself. However, porting the Okular scripts to Wayland in the new lab setup turned out to be more complex than expected. Moreover, measuring a desktop environment may need some infrastructure changes to the KEcoLab setup. Hrishikesh is currently working towards it and will continue after SoK26 is over.

Automate Promo Data Collection

Chuyen Nguyen wrote automation scripts and created environments for them to perform some of the KDE promotional team's insight data collection tasks. The first script collects KDE's X, Bluesky, Mastodon, and Threads accounts' follower and post counts using a mix of API requests and web scraping methods alongside a local Nitter instance. The second script scrapes the KDE subreddit's Reddit Insights page for weekly metrics on page visits, unique visitors and its total member count and includes a Docker image that allows for headless execution. The final script collects articles related to KDE using Google Alerts emails and performs sentiment analysis on them using a locally run large-language model. The figure below shows output from the follower and post count scraper, Reddit Insights page scraper, and Google Alerts evaluator respectively.

Plasma Setup

Onat Ribar worked on bringing Plasma Setup, KDE's first-run setup wizard, to Plasma Mobile. Plasma Setup was built with desktop screens in mind, and running it on a phone showed overall accessibility issues including but not limited to overlapping components, content clipping on short window sizes, session buttons that remained tappable beneath wizard pages, and a timezone selector built around an interactive map that was nearly unusable on a small touchscreen. Onat worked through these systematically across the wizard's QML UI, resulting in an adaptive experience across screen sizes and input methods without affecting the desktop UX. Two MRs have been reviewed, merged, and are now part of the upstream codebase in repositories plasma-setup and plasma-workspace.

Falkon

Sairam developed a XMPP extension that adds a chat client directly into the KDE Falkon browser's sidebar. Written in Python using Slixmpp, the project lets users message each other and run interactive WebXDC apps right inside their chat window. It supports modern XMPP features like Message Carbons (XEP-0280) for device syncing, Message Corrections (XEP-0308) for edits, Emoji Reactions (XEP-0444), and HTTP File Uploads (XEP-0363). This setup turns Falkon into a communication tool without even leaving the browser.

J Shiva Shankar added XMPP bookmark syncing to the Falkon browser. The basic setup is working, so whenever you add or update a bookmark, it successfully syncs across your devices. Deleting bookmarks still has a few bugs right now, but they have been documented. He plans to keep contributing after SoK to squash these bugs and get the feature completely polished.

Mankala Engine

Sayandeep Dutta helped in redesigning the MankalaNextGen GUI with Kirigami and added designs fixing the Main UI and the Game with game boards and shells. He also added music with Qt to Mankala and made translations in Tamil and Hindi. Made assets for the game variants using Krita in their traditional designs and motifs. Started with the review process of MankalaNextGen with the CI build.

Pavan Kumar enhanced MankalaEngine by adding an opening book and investigated performance of multithreaded alpha-beta search using OpenMP, Pthreads and Taskflow. In addition, he enhanced MankalaNextGen by creating visual assets for game boards and seeds and also created logos for MankalaEngine and MankalaNextGen.

Very technical post, feel free to ignore if you're not working on KDE stuff.

Since I'm on KDE Linux, I use flatpak apps for pretty much everything, such as Kontact suite. I wanted to test out how my Breeze changes would look like in KMail and other apps, but I didn't want to build the whole PIM stack.

Here's how I tested it by building the flatpak-kde-runtime in CI, downloading the built files and using it as a repository for testing any changes to org.kde.Platform locally.

1. First make changes to the flatpak-kde-runtime, such as this here: https://invent.kde.org/packaging/flatpak-kde-runtime/-/commits/work%2Fakselmo%2Ftest-breeze
2. Run the CI
3. Check the pipeline ID for build-runtime-amd64
   • Click the green checkmark, its something like #1210828
4. Download the artifact from here: https://storage.kde.org/ci-artifacts/packaging/flatpak-kde-runtime/p/PIPELINE_ID_HERE/repository-x86_64.tar.gz
5. Extract the tar into some folder after downloading
6. Add it as a repository: flatpak remote-add --no-gpg-verify kde-platform-testing ./repo-x86_64
7. Uninstall the current platform: flatpak uninstall org.kde.Platform --force-remove
   • The force-remove makes sure we do not install any apps, just the platform
8. Run flatpak install org.kde.Platform
9. Select the kde-platform-testing as the repo to install from
10. Done!

Now any app that uses the version of platform (6.10 in my example) will utilize your custom built one.

Akademy 2026: Registration

Akademy 2026 will be a hybrid event held simultaneously in Graz, Austria, and Online.

Hundreds of participants from the global KDE community, the wider free and open source software community, local organisations and software companies will gather at this year's Akademy 2026 conference. The event will take place in Graz and Online from Saturday, 19 September to Thursday, 24 September.

KDE developers, artists, designers, translators, users, writers, sponsors and supporters worldwide will meet face-to-face to discuss key technology issues, explore new ideas and strengthen KDE's innovative and dynamic culture.

Register now and join us for engaging talks, workshops, BoFs and coding sessions. Collaborate with your fellow KDE contributors to fix bugs, pioneer new features and immerse yourself in the world of open source.

For more information about the conference, visit the Akademy 2026 website.

Changelog

• Correction bug 515099: Missing secondary currency value in Accounts totals
• Correction bug 514649: Fix initial layout when splash screen is not displayed
• Correction bug 514649: Double click issue on table when there is no selection
• Correction bug 438967: Labels in Skrooge Report axis and Legend tend to illegible
• Correction bug 517475: Alarms triggering even when conditions not met
• Correction bug 515186: Import should remember the directory used last when restarted
• Correction bug 430245: summaries display truncated after searching in Operations grouped by Category
• Correction bug 518461: charts missing in responsive monthly report
• Correction: Better filtering in categories page
• Correction: Fix various error in flatpak mode (import, download sources, ...)

Krita 5.3.1.1 is an Android-only fix for 5.3.1. It is exactly the same as Krita 5.3.1, but with two fixes:

• fixed a crash on Android 12 and older when trying to access any text controls
• fix stylus workarounds for Xiaomi and OnePlus to work properly

Check out the release notes for a full overview of all the new features in Krita 5.3 and 6.0.

Note: this release is only relevant for Android, so only Android APK's and source archives are available. You only need the source archive when building for Android yourself.

[!WARNING] One again, we consider Krita 5.3.1 suitable for productive work; 6.0.1 is, because of the many changes from Qt5 to Qt6 more experimental.

5.3.1.1 Download

Android

Krita on Android is still beta; and is meant to run on chromebooks and tablets only.

• 64 bits Intel CPU APK

• 64 bits Arm CPU APK

• 32 bits Arm CPU APK

Source code

[!NOTE] Note for distributions: if you package both PyQt5 and PyQt6, you will want to patch our source code with this patch: Disallow importing conflicting version of PyQt

• krita-6.0.1.1.tar.gz

• krita-6.0.1.1.tar.xz

md5sum

For all downloads, visit https://download.kde.org/stable/krita/5.3.1/ and click on "Details" to get the hashes.

Key

The the source tarballs are signed. You can retrieve the public key here. The signatures are here (filenames ending in .sig).