GitHub (and by extension its owner, Microsoft) is calling for a new EU tech fund to ensure critical open source software can be maintained.

Its proposal, detailed in a report, stresses that our the economy relies on open source software (OSS) for digital infrastructure, but that maintenance ‘continues to be underfunded, especially when compared to physical infrastructure like roads or bridges’.

GitHub’s research says 96% of all codebases contain open source code, and OSS contributes an estimated $8.8 trillion to the global economy.

Yet, most open source projects get little-to-no funding. A third of projects, many linchpin dependencies in billion-dollar operations, are maintained by just a single person. Others get by on drive-by donations and volunteer contributors – goodwill for the greater good.

Given how widely many OSS components are, and how vulnerable they are to developer burnout, abandonment, or socially-engineered hijacking, the costs in not ensuring open source is in a healthy states are, in a sense, greater.

Infrastructure underpinning day-to-day life, powering the modern economy, balances on precarious foundations. Sustainable open source maintenance will ensure ongoing improvements and all-important security diligence for everything resting on top.

A Sovereign Tech Fund for Europe

The solution, per Github’s report, is to scale up Germany’s Sovereign Tech Fund (STF) to a pan-European level.

Regular readers will be familiar with the STF since it has, in recent years, provided money to fund important initiatives within the GNOME project, ranging from accessibility work to new privacy and wellbeing features on the desktop.

This proposed EU-STF would give cash to widely-used, critical components to help sustain and support open source maintenance.

GitHub’s report recommends a starting budget of €350 million carved out of the bloc’s 2028-2035 budget, and using the money to action 5 key areas:

1. Identifying the EU’s most critical open source dependencies
2. Investments in maintenance
3. Investments in security
4. Investments in improvement
5. Strengthening the open source ecosystem

The funding would require the creation of a new, centralised EU institution (a ‘moonshot model’), or a group of EU member states willing to provide initial funding and then seek further resources from the EU budget (a ‘pragmatic model’).

As GitHub itself puts it: “The flip side of everybody benefiting from this open digital infrastructure is that too few feel responsible for paying the tab.”

The proposed fund would be directed at the ecosystem’s most critical, underfunded dependencies

Cold hard cash is the answer, and GitHub thinks the European Union should step in to spend public money to fill in the OSS funding gap.

Others in the Linux community already take a more direct approach.

Canonical is donating £120,000 in 2025 to the smaller open source projects it relies on to maintain the infrastructure, tooling and triage mechanisms in Ubuntu and adjacent projects. The sums are small but symbolic: accepting responsibility to fund the dependencies you use.

The proposed EU-STF fund would not be a free-for-all (you couldn’t pitch a desktop widget and expect a wedge of polymer in return). Rather, it’d be strategically directed at the ecosystem’s most critical, underfunded dependencies for maintenance, security and improvement.

Github suggests the EU-STF start with a €350 million pool, but adds that this amount “…would not be enough to meet the open source maintenance need”, merely “form the basis for leveraging industry and national government co-financing that would make a lasting impact.”

While €350 million is a huge portion of EU’s budget it is veritable chump-change if viewed against the profits that big tech companies—remember those stats from earlier—generate each year.

Funding is important, but so is who funds it

I don’t disagree with the idea of funding OSS with public money since it open source is a public good. And the idea that a co-ordinated, centralised and (ideally) transparent funding process is needed is, to my mind, the logical approach – no self-beneficial backhanders.

But the cynic inside me is giving this proposal some side-eye.

If open source gets funding from the EU, it must happen lock-step with the EU pushing for adoption of open source in the public sector

Microsoft, like all of the big tech giants, relies heavily on OSS to keep its own profits ever-inflating. The idea that national governments—i.e., taxpayers—must stump up “or else” is a little (pardon the pun) rich given the rich benefit from them doing so.

The Scrooge McDuck money pits were dug out with the labour of hundreds of thousands of open source developers, and each coin clattering down to fill it up part-earned in the same way.

Open source maintainers deserve monetary support. Absolutely. It is vital that valuable talent and critical software is rewarded. I worry the creation of an EU fund at the behest of big tech might be a way for big tech to offload its own responsibilities to fund OSS.

So if open source is to get sustainable funding from the EU, it should happen in lock-step with the EU pushing the use of open-source in the public sector. This fund shouldn’t act as a sleight-of-hand subsidy for big tech to coast on.

What’s interesting from its blog post is that GitHub (and Microsoft) is not paying lip service to this idea. It’s pushing for it. It plans to meet with EU legislators to advocate for such a fund, and warn of the risks of inaction in not creating one soon.

Something has to be done, and an EU Sovereign Tech Fund is one way to address it — but those who milk the community cow for profit ought to pay most for its feed.

You're reading GitHub Wants the EU to Fund Open Source, But Who Should Really Pay?, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

A big update to OBS Studio Snap package is available to test, with Canonical rebasing the software on top of Core24 and switching the source code it is built from.

The bump to the Snap core is needed as the latest OBS Studio 31.1 release drops support for Ubuntu 22.04 LTS, as Canonical software engineer Vanda Hendrychová notes:

“Until now the snap has been running on core22; however, official builds of OBS Studio version 31.0 and subsequent releases are only available for Ubuntu 24.04 and later.We have upgraded the base snap to core24 in order to ensure close alignment with […] upstream.”

Additional, the OBS Studio Snap package is now built from the source code used by the official OBS Project PPA, and not built from code provided by the unofficial OBS Studio Portable project (which is not actively maintained).

This shift will make it easier to maintain the Snap, but it’s not without a caveat: OBS Studio Portable includes “popular third-party plugins out-of-the-box”. The updated version, being built from vanilla source code, does not.

Users can continue to install plugins in the Snap manually, but Hendrychová notes that “OBS Studio plugins are shared libraries, they must be built with the same library versions as those present in the snap”.

To workaround this, she suggests using an Ubuntu 24.04 LXD container for building custom plugins.

Better Intel GPU Support

Intel GPU acceleration is improved in the new build – and AI plugins are planned!

Intel GPU hardware acceleration is improved in this new version due to the inclusion of packages from the Intel Graphics PPA, which is also maintained by Canonical.

Additionally, AI plugins “optimized for Intel hardware” are reportedly going added in a later release. I expect these to be Intel OpenVINO plugins that Audacity offers.

Anyone with hardware able to make use of them should keep an eye out, as they includes features like noise suppression, audio ‘super resolution’, and music generation.

How to test the improvements

Community users are being asked to test the rebased, retooled OBS Studio Snap by installing it from the candidate channel. Those with a stable version installed can switch by running snap refresh obs-studio --channel latest/candidate.

Should testing go fine, the update will roll out to all user on the stable channel in the coming weeks.

The OBS Studio Snap package is, like many apps on the Snap Store, not maintained by its upstream developers, but Ubuntu community members who work under the umbrella of ‘Snapcrafters‘. The OBS Studio offer an official DEB and Flatpak on Flathub.

If you plan on testing this update, and you need or want to report issues you find, you should do so by opening an issue on the Snapcrafters GitHub page for the OBS Studio Snap package.

You're reading OBS Studio Snap App Gets a Major Upgrade – AI Plugins Inbound, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Last year I reviewed the CrowView Note, a portable monitor with battery, keyboard and trackpad that makes it devilishly easy to use a Raspberry Pi as a laptop — albeit a laptop with a single-board computer sticking out the side!

As much as I love (and continue to use) the CrowView Note—the HDMI input lets you use anything with it, even a phone—it is far from elegant.

Argon Forty, makers of the most well-engineered and popular Raspberry Pi cases (the Argon ONE range), clearly took note and thought: “hold my fruit punch”.

Argon ONE Up Runs on a CM5

The Argon ONE Up is a is a laptop computer powered by a Raspberry Pi Compute Module 5. Yup, the CM5, not the full-size Pi SBC.

You remove a panel on the bottom of the laptop, slot in your CM5 (and an M.2 SSD or expansion board, like the AI+ HAT), and bam: a Raspberry Pi-powered laptop. No SBC dangling out the side, and no super-thick chassis to house it internally.

Though this is not modular in the way the Framework line of laptops are (you can replace everything in those), the Compute Module can be swapped out (I’m not sure if alternative boards using the same connector exist, but if so: more choice) so there’s potential.

Here’s a video of the Argon ONE Up in production because, why not:

The Argon ONE UP is going to launch via a Kickstarter campaign sometime soon. No price or battery life details are known, but the aluminium alloy chassis houses:

• 14-inch IPS LCD display (1920×1200)
• 1080p webcam
• Backlit keyboard with multitouch trackpad
• Built-in stereo speakers
• microSD card reader
• M.2 2280 slot (PCIe 2.0 – can run AI HAT)

Ports include:

• 2× USB 3.1 Gen1 (Type-A)
• 2× USB Type-C (PD + Data + OTG)
• HDMI 2.0 (4K@60 Hz output)
• 3.5mm audio jack

There’s also an additional add-on able to connect to the USB Type-C ports to provide an external 40-pin GPIO header, with power button — these folks know their market.

So when can you buy one buy and how much will it cost?

As yet unknown.

Argon’s been teasing an upcoming Kickstarter campaign for a few weeks—it unveiled the ONE Up at a conference earlier this year—and a preview page for the campaign is live¹ (but you can’t back it). Keep an eye there if you’re interested in getting one.

Argon Raspberry Pi products have a good rep in the community, and I’ve no doubt many of you reading will already own some – so shout out your kit in the comments!

1. Fun drinking game: neck fruit punch for every use of the word ‘level’ in the campaign blurb. ︎

You're reading Argon ONE Up Laptop Runs on a Raspberry Pi CM5, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

KDE Plasma is getting in on the rounded bottom window corners action in its next major release.

“Breeze-decorated windows now have their bottom corners rounded by KWin automatically! This feature is on by default, but can be turned off if you preferred the older style,” notes KDE’s Nate Graham in a roundup of recent changes.

Graham also who commented on the related merge request to note though the change is relatively small it “will make a lot of folks very happy.”

Here’s a screenshot of Dolphin, the KDE file manager, sporting those subtly smoother corners in current development builds of KDE Plasma 6.5:

The change is subtle.

KDE plans to ship this visual tweak in KDE Plasma 6.5 which is not out until later this year, so don’t go play hunt-the-rounded-bottom-corner in existing or older versions of the DE just yet.

It’s also not going to apply to all apps, only Breeze-decorated windows. Breeze is KDE’s default window decoration theme and is what provides the title bar, borders and window controls shown around around applications.

On Ubuntu, native GTK4/libadwaita apps already have rounded bottom corners. Though older GTK3 apps use sharp corners on the bottom of app frames, there are tools and workarounds to force them to use rounded bottom corners – the same as on KDE.

Third-party desktop customisation tools like KDE Rounded Corners are great at forcing apps to smooth out their jagged corners but they are not as efficient as a native, built-in approach.

Adding consistent corner radii to KWin (KDE’s window manager handles how windows appear and behave) will ensure content shown inside of a window isn’t clipped by radii; that system resources and video memory get used efficiently; outlines are drawn correctly, and so on.

Some popular apps use non-Qt toolkits or custom theme adaptations and may vary in whether they use these even when running on KDE Plasma 6.5. Firefox tested rounded bottom corners on Linux a few years ago, though the effort stalled and the browser recently binned its GTK and CSS theme tweaks.

Those with a preference for pointy window corners should not panic. Rounded corners will be enabled by default in Plasma 6.5 but KDE devs plan to provide a preference to allow users to pick their preferred approach.

If curvaceous bottoms are your thing, you’ll be able to opt for a sharper look.

You're reading KDE Plasma Adds Rounded Bottom Window Corners to Apps, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Mozilla Firefox 141 serves as the latest monthly update to the famed open source browser, and adds several new features – including AI-assisted tab grouping.

This release also delivers a reduction in memory usage for Linux users, adds a handy unit converter to the address bar, and expands Picture-in-Picture support to more online streaming services.

Last month’s Firefox 140 release was a big one as it’s served as both a regular monthly release and the latest cumulative ESR snapshot for home and enterprise users who favour stability over new feature churn.

Among its features were a tab context menu entry to force unload a site from memory, easier adding of custom search engines (from any search field), listing local AI models the browser has downloaded (for built-in features) on the about:addons.

The new Firefox 141 release offers a similarly focused set of changes.

Firefox 141: What’s New?

AI-Enhanced Tab Groups

Tab grouping was added in Firefox 137 and has, in subsequent releases, improved the feature. If you want to group tabs together based on theme, task or topic, you can. Then, you can collapse the group to free up space in the tab bar.

But what if you like the idea of sorting your tabs into groups, but not the cognitive burden and manual hassle involved in doing it?

Enter Firefox 141’s AI-assisted tab groups (for some, this feature was present in Firefox 140).

This feature is able to “automatically collect your tabs into groups”, generating a name for the tab group based on inferred content, and suggest other open tabs (it thinks) match the same theme/topic.

Mozilla says automatic tab groups will “help you to stay focused, save time, and reduce clutter”. As the AI-side of the equation happens on your device no details on which tabs you’re viewing are sent to the cloud for processing.

It also says the feature may not work reliably. You may spend more time manually adding tabs to/from groups, regardless.

Unit conversion in Address Bar

Something nifty: you can use the address bar in Firefox as a unit converter (I’d wager most of us already use search engines or the GNOME Shell overview to do this).

Firefox’s unit converter supports converting units of length, temperature, mass, force, and angular measurement, and timezones. Clicking on a result in the dropdown copies it to your clipboard.

Vertical Tab Tool Resizing

Building on the vertical tab improvements in Firefox 140, this update adds a resize grip to the bottom of the sidebar. You can drag the divider to adjust how many tool icons (like AI chatbot, AI summary, history, etc) are visible, giving more vertical space to your tabs

Drag the dividing line down to hide tools in an overflow menu, or drag it up to see more of those icons on show.

Lower Memory Usage on Linux

According to the official release notes, Firefox 141 “uses less memory”, and a “forced restart” is no longer required on Linux builds after an update has been applied by a package manager.

How is it using less memory than before, and why is a forced restart no longer required? Mozilla doesn’t provide any details, so I’m not 100 percent sure. However, a sift through Bugzilla suggests it’s might be due to the addition of a “fork server“.

This new fork server address the situation when the package manager updates Firefox even though it is running, and browser trues to spawn new processes from the updated build while the old version is in memory. This is why a restart was usually required.

The fork server keeps a “pre-initialized” process ready to go (a bit like a template), which new content processes can be forked from using copy-on-write memory sharing¹.

Instead of each new tab or process starting from scratch (and use more memory as a result), they share memory pages with the template process, only copying data when they need to change it. This means each process uses less memory.

More Streaming Services in PIP

Firefox 141 adds support for Peacock, SkyShowtime, Showmax, and Now TV to Picture-In-Picture (PiP) mode, along with optional formatting to captions displayed within the PiP window. If you subscribe to those services, this will be a welcome change.

Other Changes

Elsewhere, Firefox 141 enables the WebGPU API by default — but only on Windows. The Mozilla Graphics Team plan to ship WebGPU on Mac and Linux “in the coming months”, and then on Android.

WebGPU is available in Firefox Nightly builds on all platforms (apart from Android) already, so if you’re looking to test it, you don’t need to wait for it to hit stable.

Web developers may be interested to know Firefox 141 will clear the back-forward-cache when receiving the Clear-Site-Data: "cache" response header.

Besides that, some other changes:

• Pinned shortcuts on New Tab page only show pin icon on hover
• Text Fragment creation now uses a less memory-and CPU intensive algorithm
• Wayland proxy timeout increased (benefits graphics performance)
• Xdg-activation wayland protocol supported (ahead of focus changes)
• Address autofill enabled for users in Brazil, Spain and Japan
• Firefox now uses system font icons for caption buttons (Windows 11)

Plus, security fixes.

In all, this is another welcome batch of changes which both ardent fans of the free and open source web browser and those who use it begrudgingly will benefit from. Let me know what you think of this month’s feature drop by leaving a comment!

Download Firefox 141

Ubuntu user? You will be upgraded to Firefox 141 automatically, in the background, without you needing to do anything.

Linux Mint user? You can update to Firefox 141 via the Mint Update tool from July 22, as Firefox continues to be provided as a .deb package that is updated from an APT repo.

If you use Ubuntu but you don’t have Firefox installed, and want it, you have ample choices: the official Snap or Flatpak build; the Mozilla APT repo to install the Firefox DEB; and the option to download a distro-agnostic Linux binary from the Mozilla website directly.

1. I think. ︎

You're reading Firefox 141 Adds AI Tab Grouping, Reduces Memory Use on Linux, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Gradia’s latest update introduces a new option to generate eye-catching screenshots of source code snippets.

The feature can turn a custom snippet of source code into a colourful graphic for presenting in documentation, tutorials or sharing on social media.

Code Snippets can be displayed with a window frame or without; with line numbers or without; and support a variety of popular programming language syntax. A choice of colour scheme, Adwaita, Solarized, and Oblivion included, is available.

Gradia was created to make it easier for Linux app developers to make eye-catching graphics of their software for store listings, though it has expanded its feature set in recent releases to become a solid all-round image annotator and basic image editor.

Adding the ability to generate styled screenshots of source code feels like a nod back to its developer-focused roots, and the results sure do look good!

The code snippet feature isn’t the only change Gradia 1.7 brings. Other improvements include:

• Text and Stamp tools now support optional outlines
• Images can be rotated in 90-degree increments
• Colour pickers are no longer modal (more accurate colour picking)

There’s also a cropping tool for trimming an image prior to export; default export filenames are more descriptive (and include creation time); and the app no longer asks for confirmation on close if the open image has already been saved.

As a reminder, Gradia is not an app for taking screenshots, it is an app for making screenshots look nicer. It can hook into the system screenshot tool to take new screen-grabs in app, and you can configure it to auto-open when you take a screenshot – but its job is prettifying.

Install Gradia on Ubuntu

If you have installed Gradia from Flathub, pop open a Terminal and run flatpak update to update to the latest version and try the new features first hand.

If you want to install Gradia on Ubuntu as a Snap, you can as Gradia is now available on the Snap Store! Seek it out in App Center, or run sudo snap install gradia to snag it.

You're reading Screenshot Tool Gradia Adds Code Snippet Generator, Snap Install, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Keeping your Ubuntu install up-to-date isn’t difficult, even from the command line – but if you install packages from a slew of sources you often end up typing out a shopping list of commands to stay on top of things.

Which is where the Topgrade utility comes in. It lets you run a single command to update software from multiple sources, in one fell swoop.

Topgrade is not new, having debuted around 7 years ago (the original effort was archived in 2022 but a fork continues developed). I hadn’t heard of it; I saw a reader mention it in the comments of a recent article—thanks btw!—and felt it deserving of a spotlight.

What does Topgrade do?

A command-line tool written in Rust (because of course), Topgrade detects the package managers and updatable sources you use on your system, and then runs the correct commands to update the each time you run topgrade.

It manages APT sources, Snaps, and Flatpaks as standard, but developers will appreciate it can go beyond those common sources to also refresh and update a lot more besides, including:

• Cargo, Pip(x), NPM, Nix and Gem packages
• Vim, Neovim, and Emacs plugins
• JetBrains & VSCode/ium extensions
• Git repos (if configured)
• Homebrew/Linuxbrew
• GNOME Extensions & Cinnamon Spices
• ClamAV database
• wsl, winget & Chocolatey (Windows)
• Firmware (via fwupdmgr in view-only mode)

Plus a whole lot more.

You install it, open a terminal, type topgrade, and it does its thing. A one-stop-shop – but you customise and fine-tune its behaviour (I’ll get to that in a second).

Install Topgrade on Ubuntu

You can install Topgrade on Ubuntu in two main ways (not counting building from source): Cargo or a DEB package which is available for Ubuntu (and related Linux distributions, such as Linux Mint) on the project’s GitHub Releases page.

Using the Topgrade DEB saves any undue hassle. All necessary dependencies get fetched and installed for you, and the tool will integrate cleanly with your system’s PATH, and make it easy to uninstall at a later date (sudo apt remove topgrade).

Download the appropriate package from Github, then proceed install by double-clicking it and using App Center, or run this command (in the same directory as the DEB):

sudo apt install ./topgrade*.deb

Once done, you can open a Terminal and run topgrade.

However, the first time you use it run topgrade -n instead:

This performs a ‘dry run’, listing what actions/sources it would update from but without, y’know, actually updating anything you may not want.

There are other commands too. For a comprehensive overview of everything Topgrade can do, run topgrade --help.

To fully automate all upgrades (i.e., automatically agree/say yes to any package changes each source proposes), run topgrade -y and enter your password.

If you only wish update a specific source(s) or run them in a specific order, for example snap and Flatpak, use topgrade --only snap flatpak . To skip a specific source(s) but update all the rest you’d run topgrade --disable snap – this would be one-time only.

Fine-Tuning Your Upgrades

If you want to control which software sources and formats Topgrade manages every time you run the single command, without passing flags, edit the ~/.config/topgrade.toml text file. Here you can also add/control how source(s) like git repos are handled.

To exclude a source, for example exclude Flatpak, you’d add disable = ["flatpak"], and then save the file – always remember to save.

More customisation options are available, and the config file lists most of them.

You can add your own custom shell commands to run before or after the main process itself. For power users this open up a lot of possibilities, like letting you update a remote system(s) over SSH whenever you run it the command locally.

Topgrade is useful, but not essential

If you’re comfortable with GUI utilities like Software Updater and App Center, Topgrade is not something you will have need for. It also doesn’t do anything that can’t be done in other ways (some users like to create their own custom multi-line update scripts).

But Topgrade is worth checking out if you use a diverse range of software sources and want to abstract away the hassle involved in keeping all the things up-to-date. The low barrier and automatic detection make this a fuss-free, time-saving tool.

You're reading This Tool Upgrades Everything on Ubuntu with One Command, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Virtualisation choices on Linux are, as I’m sure you’re know, varied – even more so since VMware made its Workstation Pro software entirely free to download and use on Windows and Linux, even for commercial purposes, no license key needed.

This week, VMware Workstation Pro on Windows and Linux, and its macOS counterpart VMware Fusion, received an update with critical security fixes and a remedy to an issue affecting the (useful) Snapshots feature.

VMware Workstation Pro 17.6.4 patches four critical security vulnerabilities (CVE-2025-41236, CVE-2025-41237, CVE-2025-41238, and CVE-2025-41239) and include a fix a fifth flaw filed under ‘moderate’ severity.

The critical issues all require(d) a “malicious actor with local administrative privileges on a virtual machine”. If you’re reading this and panicking your whole system may be at risk because you keep a sly Windows VM around, relax: it’s unlikely you were.

But given that VMware products are heavily used across enterprise, businesses and cloud, any and all security vulnerabilities found and disclosed do require prompt resolution, which they’ve duly received.

This update also solves a snafu with the built-in snapshots feature in earlier builds, which was throwing a “vmx [msg.log.error.unrecoverable] VMware Workstation unrecoverable error: (vmx) vmx Exception 0xc0000005 (access violation) has occurred” error.

That issue? Now fixed. The cause? Convoluted, per VMware’s own description:

“During a VM power-off operation, if Ask me is selected in VM > Settings > Options > Snapshots, a specific pointer isn’t verified prior to back-end asynchronous function calls. As a result the pointer causes access violation, becomes invalid during the VM power-off procedure.“

The VMware Workstation Pro 17.6.4 release notes continue to mention ‘known issues’, including the ongoing broken network connections when installing Windows 11 in a virtual machine (switching the network adapter from NAT to Bridged is the workaround).

Multi-monitor features are not working correctly in certain situations, and hardware acceleration on Linux hosts with Intel Meteor Lake GPUs doesn’t work (adding mks.vk.gpuHeapSizeMB = "0" config value to the VM¹ or global config file² fixes it).

Download VMWare Workstation Pro Update

VMware Workstation Pro is powerful software that performs well on Linux systems – once it is up and running. The ‘bundle’ installer Broadcom provides often ‘fails’, reporting errors building core modules against new Linux kernels, requiring a workaround.

To get this update, login to the relevant page on the Broadcom portal to download an updated installer. Windows and macOS users may be able to update in app, although a while back that functionality was broken – so YMMV, as they say!

1. For a virtual machine I believe this involves finding the relevant .vmx file for the Virtual Machine and opening it in a text editor, adding that line, and saving it.  ︎
2. The location of which… Might be in /etc — you’ll need to be root to edit anything there. ︎

You're reading 4 Critical Security Flaws Fixed in VMware Workstation Pro, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Ubuntu 25.10 for the Raspberry Pi will ship with a minimal set of pre-installed software, a move that bring the Pi images in line with regular Intel/AMD desktop builds.

The change reduces the amount of disk space an Ubuntu install on Raspberry Pi uses by around 800MB. Users will (of course) be able to install the ‘missing’ apps from the repos or Snap Store, or not – choice is the key here.

Canonical’s resident Pi engineer Dave Jones, fresh off implementing a new A/B booting system in the Ubuntu Raspberry Pi builds, tested the software sweep and noted that the fresh install footprint shrunk from around 8.7 GB to 7.9 GB.

Why the Change is Needed on Pi

Unlike a traditional desktop installation, Ubuntu for Raspberry Pi uses a pre-installed image (IMG) that gets flashed directly to an SD card or other storage.

The regular desktop ISO is a live environment containing an installer that builds Ubuntu on a hard drive, while a preinstalled image copies an already installed system – a bit like an OEM install.

On first a setup, a wizard gets users to configure their hardware, set a location and create a user account.

Since the Flutter-based installer isn’t involved, there is no choice between having a ‘minimal’ or ‘expanded’ software set – hence this change to what’s baked into the Pi image.

On a PC, the ISO contains the full suite of “expanded” applications already. Choosing a minimal install there does not download less, but uninstalls those apps during the installation process.

Since the Pi images are ‘preinstalled’, the expanded software will not be included in the image – at least, from what I can gather; the changes haven’t hit daily builds yet.

What’s Being Removed?

The following apps will no longer be pre-installed on Ubuntu for Raspberry Pi from 25.10 onwards. All of these are available to install from the archives; they’ve not been expunged from existence:

Practical Decision

Most people who run Ubuntu desktop on a Raspberry Pi are not running it as their main, productivity work hub, but a platform for development, tinkering or other specific use-cases. Not having LibreOffice available immediately won’t crater their productivity.

The change also mirrors the approach taken with the official Raspberry Pi OS, which has long provided separate images: a slim install with only essential utilities and a ‘full-fat’ one with recommended software included, ready to use out-of-the-box.

In his reasoning for Ubuntu’s need to make a switch, Dave Jones says it: “will alleviate our image building issues, but it also gives users more choice in what they want in their installation.”

He adds that under the old system, uninstalling a “seeded” (i.e., preinstalled) application sometimes had rogue consequences, like uninstalling Ubuntu’s package of tweaks for Pi hardware – tweaks most users will want.

Shipping a ‘low fat’ Ubuntu Pi image caters to the way people actually use these devices, giving them a cleaner starter base to build on.

You're reading Ubuntu 25.10 Shrinks its Install Footprint on Raspberry Pi, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

If you need to compress a video to a specific file size on Ubuntu, which tool do you reach for — HandBrake or FFmpeg?

Both are solid, capable choices, but they are not the only Ubuntu video tools available.

Constrict, a new Python-based GTK4/libadwaita video compressor for Linux, recently hit Flathub promising to end the “manual trial-and-error of re-encoding at various bitrates” through a simple FFmpeg GUI that compresses videos to exact file sizes.

To casual users, HandBrake’s sprawling feature set can feel overwhelming, and harnessing FFmpeg’s power requires mastery of arcane commands¹ requiring precise input. It’s no surprise that free online video converters like CloudConvert are popular.

Constrict is an alternative to those tools, able to automatically calculate the ABR (average bitrate) resolution, framerate, and audio quality of a video to meet the target file size.

Rather than guess quality settings and hope the end file is small enough for what you need (to meet an upload limit on social media or email), you can specify exactly how large you want the final file to be.

The app says it will try to “retain as much audiovisual quality as possible for the file size given. However, extremely steep reductions in file size can cause significant loss of quality in the output file, and sometimes compression may not be possible at all.”

Features include:

• Bulk compress multiple videos and save to a specific folder
• Frame limits to opt for a more clarity or smoother footage
• Tolerance settings to provide some leeway in output files
• Choice of H.264, HEVC, AV1, and VP9 encoding codecs
• ‘Extra Quality’ toggle at expense of compression time

The app compresses and encodes files locally, on your computer, with no cloud connections or AI gubbins. This will mean compression is as fast as your hardware — on my Ubuntu Chuwi laptop, that is not very fast at all…

Constrict is not a replacement for FFpmeg or HandBrake, and it won’t be much use to anyone looking to compressing length video files down to a smaller size whilst retaining the best quality. This answers a different need, one where compromise is expected.

You can install the app from Flathub using the link below,, or grab the source code from GitHub.

Get Constrict on Flathub

1. Or rather, mastery of a search engine ;) ︎

You're reading New App to Compress Video to Target File Sizes on Ubuntu, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

A bug fix update that tackles a number of glaring issues and compatibility hiccups in the ONLYOFFICE 9.0 release has been made available to download.

ONLYOFFICE 9.0, released in June, brought a revamped user-interface, new AI features, the ability to view VSTX diagrams, and improved support throughout the suite when used in right-to-left (RTL) languages.

Alas, that update also brought a few gremlins with it too.

Which is why ONLYOFFICE 9.0.3 is out, shipping fixes for all sorts of issues – an annoying freeze on first launch if template preview generation was taking place; a non-functional Open document shortcut on the Home page; and an inability to paste code in Macros windows.

Other fix-ups across all editors include:

• Cursor and text selection is now limited by the size of the object/shape
• Icon scaling improvements (when scaling is set to 200%)
• Revamped autoshape Merge/Flip icons in the Modern theme
• Fixed the position of the editor version line in the About window
• Assorted conversion crash fixes

Dialling down into editor-specific fixes, Document Editor resolves errors using combine and compare text operations in the Document Editor if files contained images, and remedies quirks caused by editing footers in certain DOCX files.

The Spreadsheet Editor is, once again, able to save custom functions to a file, the app will no longer freeze if you go to export a spreadsheet to PDF files, and trying to reference a cell with a formula no longer crashes the app.

Presentation Editor can handle adding animations from the Emphasis Effects section to images without issues, and won’t crash when deleting slides sequentially from the preview panel. Copying all content on a slide in PPTX files works again too.

The Forms component won’t bug out when adding TextArt settings to text in the Signature field (though, morally, it should – who’d that?!); PDF Editor copes with cut and paste operations on the thumbnail panel; and Diagrams fixes a slide display issue.

More details can be found on the ONLYOFFICE website, with the links to download this update for Linux, Windows or macOS. If you installed ONLYOFFICE 9.0.x from the Snap Store or Flathub, you’ll receive this update from there.

You're reading ONLYOFFICE Desktop Editors Get a Bug Fix Update, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Ahead of the RISC-V Summit China 2025, Canonical and ESWIN Computing have announced the launch of a new RISC-V single-board computer (SBC) with full support for Ubuntu 24.04 LTS.

The EBC77 Series SBC is, like many RISC-V boards of this type, primarily intended for use in development, embedded systems and specific-task scenarios. It’s not going to work well as a general-purpose desktop PC (though it can with limitations) for consumers.

Indeed, RISC-V devices in general can’t compete with ARM-based SBCs like the Raspberry Pi on price or performance. Arguably, they don’t need to yet – it’s a fledgling tech and people who buy RISC-V buy it because they want to test or develop for RISC-V.

From every acorn, and all that.

Cheap RISC-V SBCs do exist and are able to cater to the curious, but it is powerful RISC-V hardware that will drive the tech forward in adoption, use-cases and abilities — and this new board trends in that direction.

ESWIN EIC77 Specifications

The ESWIN EBC77 specs include a quad-core EIC7700X 64-bit RISC-V SoC running at up to 1.8GHz, delivering, per its spec sheet, ARM Cortex-A75 class performance – which for RISC-V, is decent, putting it somewhere near a Raspberry Pi 4.

The EBC77 also boasts an NPU of up to 19.95 TOPS; includes up to 32GB LPDDR5 memory; and there is onboard multimedia decoding/encoding of HEVC (H.265) & AVC (H.264) up to 8K. The Imagination GPU supports OpenGL-ES 3.2, EGL 1.4, OpenCL 1.2/2.1 EP2 and Vulkan 1.2.

While only 8MB SPI flash storage is included, the microSD card slot provides quick storage expansion, as does the 4-lane PCIe Gen3 FPC connector for anyone looking to add an SSD.

Ports include micro HDMI out, Gigabit Ethernet, 4x USB 3.0/2.0 type-A ports, a 40-pin GPIO header, and Wi-Fi 5 and Bluetooth 5.0 are built-in for connectivity needs.

A fan connector is present.

The RVA23 Problem

The 4x SiFive Performance P550 RISC-V cores on board implement the RV64GC ISA and conform to the RVA20 profile. They do not support the newer RVA23 profile.

Related Story

Why Ubuntu 25.10 Won’t Run on RISC-V Devices

Why is that a problem? Because it means this brand-new SBC will not be able to run (or upgrade to) Ubuntu 25.10 or beyond.

Canonical has (proactively) chosen to only support RISC-V devices with the RVA23 profile (mandates Vector 1.0 and Hypervisor extensions) in new distro releases.

It’s doing so because the performance and feature set of RVA23-capable devices (though none are yet on sale) will surpass the fastest, priciest RVA20 boards on the market right now. Raising the profile requirements will raise the bar on what users can expect.

It might sound like a glaring negative, but the ESWIN EBC77 has full support for Ubuntu 24.04 LTS, which is a long-term support release receiving updates until 2029.

Availability and Pricing

https://www.youtube.com/watch?v=GaDt4PtczYEVideo can’t be loaded because JavaScript is disabled: Introduction to ESWIN's Single Borad Computer EBC77 serial SBC (https://www.youtube.com/watch?v=GaDt4PtczYE)

Jonathan Mok, Silicon Alliances Ecosystem Development Manager at Canonical, says the launch of the device “represents a significant advancement for the open-source community, empowering everyone from seasoned developers to newcomers to innovate on the RISC-V architecture.”

Price is from $149 on Amazon US, though other stockists of ESWIN devices are likely to have it too.

The EBC77 Series SBC is available to look at in person at the ESWIN Computing and Canonical booths during the RISC-V Summit China.

Canonical (via CNX Software)

You're reading New RISC-V SBC Supports Ubuntu 24.04, Won’t Run Future Releases, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Name: Ubuntu Dock radii.

Age: Brand new(ish).

Appearance: Ever-so-slightly rounder.

What’s this about? When Ubuntu Dock is in dock mode (not full height/width), corner radii are out of whack with the corner radius used elsewhere. Yaru icon padding (and hover effect) is also inconsistent. A community designer noticed the issues, submitted a fix, and that fix ships in Ubuntu 25.10. *mic drop, exits stage, world left stunned*.

That’s… it? That’s it — drop the mock shock, though: you are reading OMG! Ubuntu, a site founded on covering minute changes in desktop Linux docks!

How rounder are corners now? Well, the CSS fix bumps the dock’s border radii up from * 1.5 to * 3, so they’re now 2x as round as they were before. If that in any way makes this change sound more consequential…

It doesn’t. Fair enough. The original issue is “technically unnoticeable” by most since the Ubuntu Dock is in panel mode out-of-the-box. Those who do uses floating mode may find this change similarly imperceptible, but adds subconscious “something feels right” vibes. That is good, right?

Hmm… Okay, let me put it this way: there’s nothing wrong with people eating obsessive attention to detail to Linux UIs. The eagle-eyed user who spotted this inconsistency has form: he noticed Ubuntu’s icons used optically incorrect proportions. Then as now, he didn’t just point the issue out, he got stuck in and fixed them. Team work is dream work!

Any actual upside to this? You can finally use Ubuntu on your laptop in hipster coffee shops without fear of getting disapproving glares from Figma types, hunched over their MacBooks, who can sense your dock radii is off (assuming that actually ever happened, which it doesn’t).

Any likely downsides? The 3 people who noticed the discrepancy, and didn’t mind it, will have to adapt. There’s also a chance the YouTube drama industrial complex will machine-gun “Ubuntu Did WHAT?!” content, replete with shocked thumbnail faces, off the back of this change. Stay vigilant.

Focusing on pixel-perfection is a slippery slope. Next, you’ll rehash this laboured, chat-style article format to talk about how the radius of notifications is going to be consistent with the calendar dropdown. After that… Who knows. Where will this madness end?! Relax, that issue is still not fixed, and there are probably plenty of other tiny visual quirks no-one has yet noticed and won’t be fixed.

Something to look forward to in Ubuntu 26.04, then! Such snark.

Do say: “The visual consistency Ubuntu users unconsciously craved.”

Don’t say: “deSigNeRs sHouLd sPeNd mOrE TiMe fiXinG {$engineering_problem}.”

You're reading Ubuntu 25.10 Fixes Something You Didn’t Know Needed Fixing, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Kiro, a new AI-powered Integrated Development Environment (IDE) built on the open-source foundation of VS Code by a team from Amazon, is now available in preview for Windows, macOS and Linux.

Similar to tools like Warp, OpenAI’s Codex, and ‘vibe coding’ darling Cursor, Kiro integrates AI agents into the development workflow as an AI pair programmer not only able to turn ideas into production-ready code but handle mundane and routine tasks too.

Rather than immediately generate code from a raw prompt, Kiro will turn the request into a “spec” in which requirements, design, and tasks are laid out before code is written, a process Kiro’s developers describe as going from “vibe coding to viable code”.

Once the “spec”, a living and updatable document, is good to go, Kiro breaks the feature request down into a sequence of tasks and sub-tasks. Again, developers get to audit, refine and monitor this before and during action.

Kiro also uses “hooks”. These are event-driven automations which trigger on specific actions, like saving a file or committing code to a repo. Hooks can be used to generate documentation, update APIs, run security checks, check for errors, and so on.

AI coding and agentic workflows are becoming ever-more common. In any skills market, keeping pace with trends is a pragmatic necessity: employers are hiring fewer people due to expectations AI will make a smaller group of developers “more productive” (even if stats don’t back that up).

“Through Kiro, we reinvented how developers work with AI agents. We pioneered spec-driven development, where Kiro turns your prompt into structured requirements, design, and tasks that are then implemented by agents,” the team, working within Amazon Web Services (AWS), says.

Beyond specs and hooks, Kiro supports the Model Context Protocol for specialised tools, steering rules for AI behaviour, and (naturally) includes an embedded chatbot mode for questions (albeit with file and document context).

As Kiro is built on Code OSS, Microsoft’s vanilla codebase on which it builds VS Code, it supports VS Code settings, Open VSX plugins and other kinds of AI features.

Pricing, Platforms, and Availability

Kiro is free to use with “generous limits” during its preview period, so as to enable users to try things out properly before hitting bumpers. Once out of preview, Kiro will offer an encumbered Free tier, in hopes users shell out for paid Pro or Pro+ plans.

You can download Kiro for Ubuntu as a 64-bit DEB package or a standalone binary that will run on most Linux distributions without effort. Packages for Windows and macOS are also available.

Kiro can’t be used offline, and requires a sign-in with one of four sign-in providers (e.g., Google, GitHub,) to be used. A step-by-step tutorial is included to help introduce the workflow and key features.

Currently, Kiro only supports English prompt/specs/hooks but more language support is planned — and on programming languages, it supports most used by developers including Python, Java, JavaScript, TypeScript, C#, Go, Rust, PHP, Ruby, Kotlin, and more.

You're reading Kiro, Amazon-Backed Agentic IDE, Enters Public Preview, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Blender 4.5 is out, serving as the latest Long Term Support (LTS) release of the much-revered cross-platform 3D modelling and animation suite.

While a lot of hype has been generated over features planned for Blender 5.0, the the Blender 4.5 change-log is by no means meek, offering a mix of performance, usability and tooling tweaks.

“With 2 years of updates, full Vulkan support, and quality-of-life improvements, Blender 4.5 LTS is every Blenderhead’s best friend,” the launch page states.

The promise of full Vulkan support ‘on par with OpenGL’ is the big draw in this release, gaining support for more features on Windows/Linux like OpenXR, Subdivision, and USD/Hydra. Limitations remain, and it is not enabled by default – flicking a switch in Preferences required.

If trying the Vulkan backend in Blender 4.5 on Linux, those with AMD GPUs are advised to use Mesa 25.3 (or newer)—FYI: Ubuntu 24.04 is getting Mesa 25.0.7 as a HWE update—and Linux users with NVIDIA GPUs need driver version 550 or higher.

Other changes and additions in Blender 4.5 LTS:

• New Geometry Nodes focused on data access and string handling
• New Grease Pencil nodes have been added
• Denoise node now supports GPU devices
• Procedural texture nodes now available in the Compositor
• “Boundary Strokes” renamed “Fill Guides”
• Option to only show onion skinning for the active object in Viewport
• SVG exporter can now export animations
• Multi-threaded shader compilation enabled by default
• Liquid simulation performance improved by 1.25x – 1.5x
• Playhead element snapping available in all timeline editors
• Video Sequencer supports HDR
• Faster texture loading in EEVEE
• New shadow terminator bias 
• Support for writing ProRes codec videos
• User interface buffs, including large cursors for HiDPI on Linux

As for all recent releases, anyone looking to learn more can watch a video recap from Blender which highlights the key changes. Detailed release notes are available for those who prefer to read.

https://www.youtube.com/watch?v=wPhA0imjvVsVideo can’t be loaded because JavaScript is disabled: What's New in Blender 4.5 LTS! Official Overview (https://www.youtube.com/watch?v=wPhA0imjvVs)

For macOS users, Blender 4.5 is the last release to support Intel-based Macs. Blender 5.0 will be the first version only officially available for macOS on Apple Silicon (unofficial builds could appear as Blender is is open source software).

Installing Blender 4.5 on Ubuntu

Blender is free, open-source software available for Windows, macOS and Linux. 

While an older version of Blender is available to install on Ubuntu straight from the repos using sudo apt install blender, to benefit from the changes mentioned in this post users can install Blender from the Snap Store, or download a binary from the official Blender website.

For Linux, Blender’s system requirements ask for a distro equipped with glibc 2.28 or newer (Ubuntu has this), a quad-core CPU with SSE4.2 support, 8GB RAM, and a dedicated GPU with at least 2GB VRAM and OpenGL 4.3 support — Vulkan requirements listed further up this page.

You're reading Blender 4.5 LTS Released with ‘Full Vulkan Support’, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Most fresh graduates could be forgiven for taking time out to rest, relax and defrag their post-study mind, but one comp-sci student decided to devote their downtime by reviving KDE’s TV-focused front-end — and the results look great!

Plasma Bigscreen is a shell designed for use on televisions and other large displays, using a remote for navigation rather than mouse or keyboard. It’s been around for a while (I wrote about it in 2022) but if I’m frank—I’m not, I’m Joey—I’d forgotten it existed.

KDE developer Espi, a regular contributor to Plasma Mobile, had not.

But he found the project in a state of limbo. Though ported to Plasma 6, Bigscreen had seen scant development in the years since and had been dropped from the main Plasma release cycle and from the package repos of most Linux distributions.

In a detailed blog post Espi says that upon seeing this dearth in development effort he “sensed an opportunity” to do something about it, and decided to “take a swing at improving the project” over the course of a week.

And its new programming makes for some must-see TV.

Plasma Bigscreen Sees a Big Refresh

Espi’s big Bigscreen revamp brings the home screen UI much closer to the original Breeze Ocean mockups, with a cleaner, flatter design, a dynamically expanding clock widget, and a KRunner-based search tools to make finding and launching apps faster.

The Settings interface was rejigged, now sporting a sidebar layout, using vertical scrolling stacks for toggles and content, and making use of a custom-made component library (which other contributors and developers can use) for building TV-focused UIs.

Beyond what’s visible on screen are extensive changes under the hood, ranging from QML library updates and module fixes to greater use of tools and other Plasma components.

All of it is real, working code.

Espi tested the work on his television by compiling Bigscreen on postmarketOS and running it off of a Raspberry Pi 5.

Xbox controller support worked in lieu of a compatible TV remote, as did a number of apps and games, including the stock Bigscreen web browser Aura as well as apps from Flathub, like YouTube client VacuumTube, Kodi, and even SuperTuxKart.

Remaining challenges

Given that Espi only spent a week working on Bigscreen, the progress achieved is commendable and a solid reminder in the power of passionate volunteers in pushing open source projects forward.

But there’s more to do, like solving the lack of a virtual keyboard with arrow navigation (one is planned as a separate Plasma project) and improving the handling and configuring of modern TV remotes.

For now, Espi has returned to working on Plasma Mobile, but says he will continue to review merge requests to Bigscreen and help new contributors.

The project’s future will hinge on others getting involved, and would benefit from having a clear direction set out to focus development effort. It also needs to reconnect to the Plasma release cycle if it’s to reach more eyeballs and gain feedback.

Anyone interested in sampling the current state of Bigscreen will have to compile it from source using the source code on the KDE Invent or fetch it through KDE nightly repos – at least until other packages are made available.

You're reading Plasma Bigscreen, KDE’s TV Interface, is Back on Air, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Consider this post apropos of nothing given that nothing has been formally announced, but it seems that long-rumoured merger of Google’s two operating systems is indeed happening.

Sameer Samat, who works at Google as president of the Android Ecosystem division that oversees Android, casually mentioned the merger in an interview with TechRadar, saying:

“…we’re going to be combining ChromeOS and Android into a single platform, and I am very interested in how people are using their laptops these days and what they’re getting done.”

Combining operating systems would mean the end of one, since they are fundamentally different. It’d clearly be ChromeOS; Android is the brand with marketshare and mindshare, and does most of what ChromeOS can (Android 16 added GPU-accelerated Linux VM and a terminal).

Rumours about Google rebasing ChromeOS on Android have been kicking around for years, stretching as far back as 2014 when Google created ARC as a way to run Android apps via Chrome as ‘extensions’ (something you didn’t need ChromeOS for).

So why is a merger only being confirmed now, over a decade later?

Samat namecheck Apple in his chat, the blurring of iPadOS and macOS, and the productivity synergy of the Apple ecosystem (by virtue of running on All The Things – things Android could very easily run on too).

Anti-trust actions on the horizon may also hasten a merger.

If Google is forced to sell off its Google Chrome browser—and what happens to the Chromium codebase is unknown since Google is the biggest financial and technical contributor, idk—it would complicate Google’s ongoing involvement and support commitments to ChromeOS.

Users can already install Android apps on Chrome OS, but they run in a tightly-integrated subsystem and container (ARCVM). Users can also choose to enable Linux Developer mode to access a Debian-based container to install Linux apps on Chromebooks.

Why am I mentioning this news on a blog dedicated to a specific Linux distribution? Should I not blow the dust off mothballed sister-site¹ OMG! Chrome and write about it there instead?

Ubuntu is not an island, and splashes made within the wider Linux ecosystem do ripple out. Plus, the news may still be of interest more generally. Google is a key contributor to Linux kernel development, and each new kernel release adds or improves support for Chromebook hardware.

Phasing out Chrome OS and Chromebooks in favour of a desktop-friendly Android OS with better desktop-style multitasking could lead to a glut of newer, cooler Linux-friendly hardware on the market (whether said hardware can be made to run regular Linux, unknown).

So Chrome OS is a Linux-based desktop operating system. It runs a (heavily-patched) Linux kernel, but it uses a read-only, signed system image and atomic updates from a single, Google-maintained source. There is no end-user package manager, heavily restricted shell access, and a custom desktop.

It is not what I class a Linux distribution, but other opinions are available² ;)

Arguably, if this merger happens—no timelines, dates, other details have been revealed—it is overdue.

Chrome OS is directionless and without purpose. What began as a low-cost, thin-client and web-first OS has snowballed into a hodge-podge of containers and conflicting user experiences in an effort to appeal to everyone.

Chromebooks run web apps, Android apps and Linux apps, but none best-in-class.

Most Android apps are designed for phones; Linux apps behave, run and look better on a proper Linux desktop; the simplicity and opportunity of a web-focused OS is diluted by tacked-on subsystems running non-web apps.

In a sense, there is no point to Chrome OS: Android does everything Chrome OS, but better; Chrome OS does some of what Android does, but worse.

Being subsumed into a unified, more coherent Linux-based OS (even if it is more Android than Chrome OS) is clearly the smarter move — should this melding of machine systems happen, of course. Google does have a habit of having—or buying—a good idea only to fudge it.

1. For those who don’t know, I launched a spin-off of OMG! Ubuntu called OMG! Chrome that ran between 2011 and 2016 (sporadic posts thereafter as my enthusiasm for Chromebooks waned). ︎
2. As is Chrome OS Flex, which anyone can download to install Chrome OS on modern-ish Intel and AMD-based PCs and laptops. It lacks some of the more appealing features found in the mainline version preinstalled on Chromebooks, like Android containers. ︎

You're reading Chrome OS is (Apparently) Merging with Android, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Terminal emulator apps: ten-a-penny on Linux, right? All major Linux distributions (Ubuntu included) preinstall one, and most command line aficionados are loyal to their preferred client.

Because yes, every terminal emulator runs commands and displays output. They do the same thing in at first blush, which makes switching or swapping feel less urgent. Yet implementation choices, be it language and toolkit or convenient integrations, suit some more than others.

I often cover new or updated terminal apps, like Pytxis (the new terminal app in Ubuntu 25.10), speed demon Ghostty, and AI-infused agentic terminal-cum-IDE Warp.

Rio is yet-another terminal emulator that Linux (as well as Windows and macOS) users make kick the tyres on. Although not new—first release was in 2023—Rio was updated recently, and an even bigger update on the way.

Since I plan to cover that release when it lands, I ought to spotlight the who, what and why first.

What Makes Rio Different?

Rio is described as a Rust-built, hardware-accelerated terminal emulator powered by WebGPU (itself, notable) with a focus on being “fast and efficient“. On Linux, it works with both Wayland and X11, and offers native ARM64 builds for Pi users.

It’s the use of WebGPU that makes Rio a little different. WebGPU is a new web standard and JavaScript API that lets web developers build apps which can use the underlying system GPU for computational tasks, as well as fancy graphics.

Of course, Linux isn’t short on hardware accelerated terminal apps these days either.

Rio carries a crop of features more advanced users will appreciate such as font ligatures (where characters like -> are combined into a single glyph →), terminal splits (multiple panes in one window), and iTerm2 and Kitty image protocol support.

Rio also reuses code from the legendary Alacritty for its ANSI parser, events, and grid system, plus borrows a few features, like a toggleable Vi mode.

Much like the carnival, the Rio terminal brighter look will be pleased to hear the terminal emulator supports themes and transparency (with or without background blur. On Ubuntu you need to install the Blur My Shell extension to see blur in action).

It also has a couple of novel (or gimmicky, depending on your viewpoint) features like being able to apply RetroArch shaders, if you fancy CRT-style aesthetics, and it handles the Sixel protocol for in-terminal bitmap images.

Rio Relies on Keyboard Shortcuts

Rio is a cross-platform app so there is a hit to cohesiveness (a USP of the Ghostty terminal is that it uses native tool, despite also being cross-platform) versus native terminal tools.

Designed to be used with a keyboard, you also won’t find (m)any right-click menus in Rio and to customise the app (e.g., theme, behaviour, keybindings, default directory, etc) you need to edit the configuration file in a text editor as there’s no in-app preferences dialog.

Thus, using Rio requires learning a (fair) few keyboard shortcuts to get the most from it. The full list of key bindings on the Rioterm website is worth checking out, but here are few handy ones to swot up on off-the-bat (for Ubuntu):

• Shift and Ctrl and R – creates a right split
• Shift and Ctrl and D – creates a bottom split
• Ctrl and D – closes the focused split
• Shift and Ctrl and T – opens a new tab
• Shift and Ctrl and ] or [ – switches left/right tab
• Ctrl and - or + keys – change font size

Is it possible to resize splits in Rio? Per the list of keyboard shortcuts, no — for now, anyway!

How to Install Rio on Ubuntu

The purpose of this app spotlight is not to convince you to switch wholesale from what already works for you, but make you aware of choices – if not for you directly, so you can recommend and share the knowledge on with others.

While raw CLI utility can be had from a TTY, extra bells and whistles are appreciated by many. And for those whose command-line heavy workloads may benefit from the WebGPU speed, Rio is worth checking out.

Rio is open source software available for Windows, macOS, and Linux. Its cross-platform nature should make it appealing to developers who routinely move between different operating systems.

As Rio is on Flathub, it’s easy to install on a wide range of Linux distributions. Ubuntu users may prefer to download a DEB installer from the Rio GitHub. The DEB will not add an APT repository, so you will not receive automatic updates.

Have you tried Rio? Share your experience with it — or any of its rivals — down in the comments!

You're reading Rio is a Fast, WebGPU-Powered Terminal for Ubuntu, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Calibre, the dependable cross-platform e-book viewer, manager, converter and more continues to publish new update regularly, the latest arriving this week.

Anyone managing a well-thumbed libraries of e-books, comics, PDFs and other supported files will appreciate hearing that the Calibre 8.6 release improves the performance of restoring the database by “an order of magnitude”, per Calibre’s development team.

It does this by using SQLITE savepoints when restoring individual books, netting big gains. One contributor said it “brings the time to restore my 5.5k book database from over 5 hours to only 5 minutes” – which really is ‘an order of magnitude’ faster!

Coupled with the 30% faster opening of large EPUB files in the e-book viewer (a change added in the Calibre 8.3 release a couple of months back), it’s great to see Calibre’s contributors making such sizeable performance gains.

Other changes in Calibre 8.6.0 include:

• Content server user preference adds checkbox for password changing
• Tweaks gains option to show sort value for series in Tag browser
• Default output format for Kindle is now AZW3 (previously MOBI)
• Manage authors/items pages add ‘Search “not in”‘ & ‘Filter “not in'”
• Misc news source improvements

Plus, the usual bug fix footnotes are included, several focused on resolving regressions earlier builds introduced like broken fading of background images in the ebook viewer, and quirks in viewing PDF files in Calibre on Windows.

The tag browser also once again enables searching for books by the first letter of a series, and ensures that Next/Previous buttons in the metadata editor paginate to what the next/previous item in the order when the dialog was opened, not the order after a change gets made.

Installing Calibre 8.6 on Ubuntu

Calibre is free, open source software available for Windows, macOS, and Linux. Official installers for Windows and macOS can download installers for the latest release from the project website, while Linux users can get the new release in a few ones.

To install Calibre on Ubuntu using the official Linux binary package, run the following command (taken from the Calibre website) to download the Calibre installer script and move the binary it fetches to the relevant system location to easy launching.

sudo -v && wget -nv -O- https://download.calibre-ebook.com/linux-installer.sh | sudo sh /dev/stdin

On Ubuntu, it may be required to install the libxcb-package, depending on which packages have been added/removed since you installed Ubuntu.

Alternatively, go to the Calibre Github releases page and download the latest Linux release from there under the ‘assets’ heading. Extract the TXZ, enter the folder, and double-click to run. This won’t create an app launcher or shortcut, however.

For less fuss, a Calibre Flathub listing is available, albeit unverified and a little tardy in updating to the latest binary release (at the time I write this it’s still on v8.5.0).

You're reading Calibre 8.6.0 Delivers a Dramatic Database Speed Boost, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

You may have noticed that Ubuntu (rather, GNOME Shell) does not provide many ways to customise notifications out of the box. If you want to change where notifications appear, or how long they stay on screen, you seemingly can’t.

Well, actually you can — via GNOME Shell extensions¹.

Now, the way notifications look and behave in Ubuntu works well enough for most. Notifications appear at the top of the screen, slap-bang in the middle, right under the Date menu in the top panel — a menu that is also a notification hub where un-actioned notifications live.

But, if you want more control over the look and position of desktop notifications, like move notifications to the right of the screen, or a way to filter notifications before they even appear, there’s a brand-new GNOME Shell extension that can do it.

Meet the

GNOME Shell Notification Configurator

Notification Configurator is a relatively new GNOME Shell extension, developed by Artem Prokop. As it’s designed solely for customising notifications, and nothing else, it’s pretty comprehensive — with a few caveats.

While there have been various extensions to manage notifications in GNOME Shell over the years—some of those still work with the latest GNOME releases, but many don’t—having a new option that is focused on the particular task, is nonetheless welcome.

It supports things I’ve not seen in similar extension, too.

Artem’s Notification Configurator extensions supports GNOME 46, 47 and 48 (ergo Ubuntu 24.04 LTS to 25.04) and allows you to do the following:

• Set notification position – top fill, left, centre, right (it can’t move to bottom)
• Set notification rate limiting to not see alerts from the same app within a set period
• Block or hide notifications from apps, or alerts containing specific words
• Customise notification colour including only for individual apps
• Enable or disable notifications in fullscreen mode

It also has a ‘test notification’ feature, which makes it easy to preview/check changes you make, as you make then. If making use of the text-based notification filtering or custom themes, the test toasts are especially useful.

More features may be added in future updates, but there are underlying limitations to what this (or any) extension can do to the native notification system.

Try it Out

If you fancy installing this extension, it’s available on the GNOME Extensions website.

It is, as I’m sure you’re bored of be saying, easier to install (and browse, search, configure, remove) GNOME Shell extensions on Ubuntu if you use the desktop Extensions Manager app, which available in the Ubuntu repos.

Get Notifications Configurator on GNOME Extensions

1. GNOME cops flacks for “not offering many settings” but that’s not quite fair: it does, most are simply hidden. GNOME curates an experience, and every toggle and switch it puts in the GUI would mean another potential combination it has to test and support when, arguably, it doesn’t want to since it has an opinionated vision on how things should work. Permutations scale! Thus, when we choose to install extensions or apps or otherwise change a hidden setting ourself we do so knowing we take responsibility if something breaks. ︎

You're reading Customise and Move Notifications on GNOME with this Extension, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Existing users of Ubuntu 24.04 LTS will receive an updated hardware enablement (HWE) stack this month, ahead of the release of Ubuntu 24.04.3 LTS next month.

Canonical backports the newer Linux kernel and graphics drivers from its latest interim release to its currently long-term support (LTS) version periodically. This update for Ubuntu 24.04 LTS uses components from Ubuntu 25.04, released back in April.

This means the Ubuntu 24.04.3 HWE is composed of the gaming-friendly Linux 6.14 kernel from Ubuntu 25.04, plus the Mesa 25.0.x series graphics drivers, a major uplift over the 24.2.x series users received in the 24.04.2 HWE update.

Related Post

What’s New in Linux kernel 6.14?

The graphics stack includes other components tied to GPU support, including new DirectX headers, spirv (headers and tools, glslang, wayland-protocols (newer build to satisfy Mesa, not a plucky backport).

The spirv-tools update carries a bug-fix required by the Intel shader compiler, while spirv-headers and glslang are included as build dependencies.

For those with older AMD, Intel and NVIDIA GPUs, there’s an updated mesa-amber package, which continues to provide support for legacy graphics cards. While those drivers don’t see many changes, they often include the odd fix – better than nothing.

Of note, Ubuntu’s developer say: “libglapi library got merged in mesa-libgallium but it can’t be directly used by external dependencies, so instead we’ll have mesa-amber build libglapi-amber which provides libglapi-mesa”.

What’s New in Mesa 25.0?

The Ubuntu 24.04.3 HWE ships with Mesa 25.0.7, which is the latest bug fix update in the 25.0.x series and a notable uplift over the 24.2.8 release that noble currently provides.

Given the leap from 24.2 – 25.0, there are a myriad of improvements provided for Ubuntu gamers and users running the distro on newer hardware, including:

• Vulkan 1.4 API
• Initial support for RDNA4 GPUs
• AV1 video decoding in the ANV Vulkan driver
• Improved NVK driver
• Issues with VA-API, H.264 video decoding resolved

Plus, Mesa 25.0.7 rolls together a raft of remedies for glitches, quirks and performance issues across a wide range of Linux games, including Steel Rats, Cyberpunk 2077, Octopath Traveller II, Fenyx Rising, Tales of Arise, Elden Ring and Hogwarts Legacy.

Coupled with the gaming performance improvements provided by Linux 6.14, which can boost frame rates by 50-150% in some Windows games running on Linux, the 24.04.3 HWE combo looks set to deliver a significant uplift for LTS gamers.

When will users get this HWE?

The Ubuntu 24.04.3 release date is set for August 7, 2025.

That is the date when a freshly-spun ISO with Linux kernel 6.14 and Mesa 25.0 preinstalled, alongside hundreds of bug fixes and software updates released since the last ISO was generated, is made available to download.

Existing Ubuntu 24.04 LTS users won’t have to wait until then.

The Mesa updates and the Linux 6.14 kernel backport are currently in the noble-proposed repository for testing. They will be released to all users once this process is complete as a software update, likely sometime in the next few weeks.

You're reading Ubuntu 24.04 Users Get Major Kernel, GPU Driver Update, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Linux users can now install the official Mozilla VPN client from Flathub, making access to its paid-for, privacy-minded service more readily accessible to Linux users.

The Mozilla VPN Flatpak is current in the process of verification (reminder: Linux Mint does hides unverified Flatpak apps in its Software Manager by default, so if you’re reading from there, keep that in mind) but is an official upload, maintained by Mozilla directly.

The Mozilla VPN client is open source software and although it is already available to install on Ubuntu-based distributions as a DEB package from the Mozilla APT repo, packages for other Linux distributions were not provided, with such users advised to ‘compile it from source’.

That makes the arrival of the Mozilla VPN client on Flathub, the most popular desktop¹ Linux app store, all the more notable — the official Mozilla VPN extension for Firefox is Windows only, too.

Using an official desktop client is not strictly necessary to use most VPNs on Linux, but having one is certainly convenient (hence the positive reaction to the recent NordVPN Linux GUI addition).

Is Mozilla VPN worth using? I haven’t used it, and it is but one of many VPNs available to Linux users and beyond. I won’t be acting as a marketing arm of Mozilla to upsell you on it being the best VPN for Linux, but its features are as follows:

• Connect up to 5 devices
• More than 500 servers in 30+ countries
• Fast network speeds even while gaming
• No logging, tracking or sharing of network data
• No bandwidth restrictions or throttling
• Device protection, multi-hop routing and more

Visit the Mozilla website to sign up for or otherwise learn more about Mozilla VPN. There’s only one “plan” available, costing from $9.99 a month or, for those willing to pay upfront for a year’s access, from $4.99 a month (offer price; will go up).

If you’re an existing subscriber and want to install the Flatpak build, you can install it through your preferred GUI or via the CLI (assuming you’ve setup it up):

flatpak install flathub org.mozilla.vpn

Once installed, launch the app, proceed to login, and configure things as you need.

Get Mozilla VPN on Flathub

1. In terms of breadth of software and use across different desktop Linux distributions. Arguably, the Snap Store has to most widely used on account of 1) Ubuntu’s install base, 2) preinstalling snaps (which update automatically), and 3) Ubuntu’s IoT footprint that relies on snaps heavily. ︎

You're reading Mozilla VPN Linux App is Now Available on Flathub, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Fans of tiling window managers which aren’t Sway, Hyprland or i3 may be keen to kick the proverbial tyres on the latest version of miracle-wm, which is now available.

Miracle-WM 0.6 is a bumper update to the Mir-based Wayland compositor developed by Canonical engineer Matthew Kosarek, who aims to create “a flashy, cozy tiling window manager” in the vein of Hyprland but with less up-front complexity.

The “big” ticket additions in this update is the new libmiracle-wm-config.so shared library, which offers a C interface that folks can use to write their own GUI configuration tools for miracle-wm (as Kosarek has done himself using Flutter).

Miracle-wm 0.6 also adds support for the wlr output management protocol, enabling users to use apps like wdisplays to change output settings at run time.

Other new features in miracle-wm 0.6 include:

• Improved display configuration
• Window border radius support
• Window events are now supported
• Added miracle-wm-xdg-autostart.target
• Better rendering of borders
• Smoother animations
• New ipc commands:
  • fullscreen toggle
  • floating toggle 
  • rename workspace...

Plus, lots of bug fixes, tweaks and other code buffs. The Miracle-WM GitHub release notes are available for those interested in the nitty-gritty.

Many of the key changes in miracle-wm 0.6 can be seen in this release demo video:

https://youtu.be/l5ljNtbOaf0Video can’t be loaded because JavaScript is disabled: miracle-wm v0.6.0 release demo! (https://youtu.be/l5ljNtbOaf0)

Install Miracle-WM on Ubuntu

Want to try Miracle-WM on Ubuntu? It’s easy enough.

You can install Miracle-WM on Ubuntu using the official Miracle-WM snap, which does need to be installed with the --classic flag to run properly, like so:

sudo snap install miracle-wm --classic

Alternatively, the official Miracle-WM PPA provides pre-packaged builds for Ubuntu 24.04 LTS (only), however the PPA is slow to be updated.

Once installed, select the ‘miracle-wm’ session using the menu on the GDM login screen, and log in as normal — don’t expect to see much when you do!

Miracle-WM is a window manager/compositor and not a full-blown desktop environment. You’ll need to craft a configuration file using a text editor, and define the tools and apps you want to use. You can log out of miracle-wm using super + shift + e.

The introduction docs on the project wiki give pointers on how to get going, plus suggested tools and utilities, like launchers and panels to use (waybar, swaybg and wofi are commonly used).

It’s also possible to run Miracle-WM as a window (on Wayland) inside of another desktop environment, like GNOME Shell. This is useful when editing or tweaking config files since you can refine your setup without the hassle of needing to log out (or reboot if you get it wrong).

If all of that sound like hassle, that’s fair. Part of the appeal in using a tiling window manager, giving you fine-grained control and the means to customise and curate a desktop experience and style that suits your needs.

You're reading Miracle-WM 0.6 Release Brings More Refinements, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.

Ubuntu will use an A/B boot approach on Raspberry Pi devices to improve reliability and boot failures, albeit with a minor drawback for users.

You're reading Ubuntu is Changing the Way it Boots on Raspberry Pi, a blog post from OMG! Ubuntu. Do not reproduce elsewhere without permission.