Wake up and smell the future: Canonical's bold vision for Ubuntu, Flatpak friction, and the Core Desktop revolution. The Linux world is on the brink of a seismic shift, and Canonical is at the forefront, but not everyone is convinced.
During the recent Ubuntu Summit, The Register's FOSS desk sat down with Jon Seager, Canonical's Vice President of Engineering, to discuss the future of Ubuntu. This is a detailed exploration of our conversation, expanded for clarity and depth.
Here’s the bombshell: Seager believes that in the medium to long term—think five to ten years—the default Ubuntu Desktop will be a Core Desktop. But don’t hold your breath for 26.04 or even 28.04; this transformation is a marathon, not a sprint. When it happens, visiting ubuntu.com/download and clicking 'Ubuntu Desktop' will deliver a Core image, with an option to download a 'classic' Ubuntu version akin to what we use today. But here’s where it gets controversial: the path to this future is fraught with challenges, and not everyone agrees on the direction Canonical is taking.
Seager, a self-proclaimed 'immutable Linux bore,' recently penned a blog titled The Immutable Linux Paradox. He’s spent considerable time with NixOS and Universal Blue, and his insight is stark: there’s no such thing as a truly immutable general-purpose operating system—hence the paradox. Yet, Nix and Snap share surprising similarities in how packages are built, stored, and executed. For instance, Snapcraft uses patchelf, originally a Nix tool, to achieve some of its functionality. And this is the part most people miss: while there’s momentum behind bootc and OStree, Canonical isn’t following that path. Seager acknowledges its merits but asserts it’s not the right fit for Ubuntu’s vision.
Progress on Core Desktop has been slow, Seager admits. When initially promised, he wasn’t at the helm of Ubuntu’s engineering. Upon closer inspection, he and his team realized significant architectural work was needed to ensure the final product would be something they could proudly release. The result? A conceptual proof-of-concept showcasing a Core GNOME machine and brief engagement with the KDE community. But here’s the kicker: swapping out components like gdm for kdm or GNOME for KDE proved more challenging than anticipated, prompting a rearchitecting effort that’s still not as far along as Seager would like.
At the heart of this transformation is snapd, which remains under active development. A critical feature in the works is permissions prompting, slated for general availability in 26.04. This mechanism, already seen in experimental mode in 25.10, is key to unlocking Snaps on the desktop. It addresses a major pain point: apps crashing or failing silently when attempting actions outside their sandbox. This is where the Flatpak vs. Snap debate heats up: while both offer confinement, they do it differently, and Snap’s approach has often been criticized for its strictness. Permissions prompting introduces a familiar iOS/Android-style dialog, allowing users to grant or deny access to resources like cameras or microphones on a case-by-case basis. This is essential for Core Desktop, where all Snaps will be confined by default.
Another piece of the puzzle is TPM Full Disk Encryption (FDE), which has taken longer to implement in Ubuntu than in some other distros. Canonical is building a hybrid approach that works seamlessly on both Core Desktop and classic Ubuntu, ensuring a unified experience across platforms.
Seager has a post-26.04 plan to assemble a 'tiger team' focused on Core Desktop. His goal isn’t just to ship it quickly but to build a desktop that will last the next 20 years. But here’s the provocative question: can Canonical balance the Linux community’s love for choice and hackability with the need for a polished, 'just works' experience?
One risky move on the horizon is shipping Pipewire as a Snap in 26.04. This could backfire, as some users vehemently oppose Snap, going so far as to remove it entirely. If they do, they’ll lose sound functionality. Why Pipewire? It simplifies backporting support for new hardware like cameras and speakers, and its integration with Snaps offers automatic systemd unit management—a step toward Core Desktop readiness.
Now, let’s talk controversy: the perception that Snap is 'closed' and Flatpak is 'open' persists, but Seager counters that the tooling for building and distributing Snaps is entirely open. He acknowledges Flatpak’s importance, noting that recent issues with Flatpak installations in 25.10 weren’t intentional but a byproduct of shipping more AppArmor profiles. He’s since directed his team to implement tests ensuring Flatpak compatibility moving forward.
The debate over software stores is equally heated. Seager argues that multiple stores aren’t always the answer, citing Fedora’s struggles. Each additional store grants root access, raising security concerns. He draws parallels to the Apple App Store and Google Play Store, emphasizing the value of a vetted, centralized repository. Here’s the counterpoint: while some criticize Apple’s notarization and fees, the concept of a trusted distributor isn’t inherently bad. Canonical’s approach allows PPAs and manual Snap installations for those who prefer flexibility.
Seager challenges the Linux community’s 'unquenchable quest for choice' as a barrier to polish. He asks: How do we balance extensibility with delivering a product that 95% of users can install and use without friction? His personal experience using Snaps for apps like Firefox, VS Code, and Discord has been largely issue-free, though he acknowledges edge cases like native messaging between OnePassword and Firefox.
Canonical’s long-term support commitment is a game-changer. Imagine telling an ISV like Proton that their software will work seamlessly across Ubuntu releases for 12 years, with Canonical handling complexities like windowing, volume, and hardware access. But here’s the bold claim: Canonical is the only vendor capable of delivering an entire datacenter stack, from Kubernetes to bare metal, using exclusively their own tools. It’s a vertical integration dream, but the devil is in the details—each component is complex, and making them work together is no small feat.
Finally, Seager addresses the backlash over Rust-based coreutils and sudo-rs. He’s unfazed by online criticism, drawing parallels to the early days of GNU coreutils. His message? Innovation isn’t reckless; it’s necessary. The question for you: Is Canonical’s vision the future of Linux, or a step too far? Let’s debate in the comments.**
