LINUX Unplugged

692: Ubuntu’s Core Reset

1 hr 23 minVideo

Canonical’s Jon Seager joins us to unpack Ubuntu’s bold desktop reset, and what could be the biggest reimagining of Ubuntu in its history. Sponsored By: Managed Nebula: Meet Managed Nebula from Defined Networking. A decentralized VPN built on the open-source Nebula platform that we love. Incogni: They can't harm you if they can't find you! Take back control of your personal data today. Go to incogni.com/unplugged and use code UNPLUGGED for 60% off an annual plan. It's completely risk-free with a 30-day money-back guarantee. Namecheap: Use code UNPLUGGED for an extra 15% off VPS hosting. Hostinger: Snag Hostinger services at their lowest prices before the sale ends! Create your website today for just $1.99. Go to Hostinger.com/UNPLUGGED and use code UNPLUGGED for an additional 10% off Black Friday and Cyber Monday pricing. Support LINUX Unplugged Links: Share your Holiday Homelab - One hardworking box or a rack full of experiments? We want to see it. Tell us what you built, show it off, and share what makes it yours. Web Boost - Send us a boost via sats or USD Jupiter Uplink 💥 Gets Sats Quick and Easy with Strike 📻 LINUX Unplugged on Fountain.FM ConnecTen Internet: promo Jupiter35 MP3: Clanker Therapy 3: What Have You Been Building? YouTube: Clanker Therapy 3: What Have You Been Building? Bootleg: Clanker Therapy 3: What Have You Been Building? | Members Stream Fixes available in -proposed pocket for CVE-2026-98432 | Ubuntu CVE-2026-98432 | Ubuntu Canonical releases Ubuntu 26.10 Stonking Stingray Ubuntu 26.10 release notes Accelerating Ubuntu Core Desktop Canonical Reviving Their Immutable Ubuntu Core Desktop enioka group Accelerating Ubuntu Core Desktop Engineering Ubuntu For The Next 20 Years GamingOnLinux: Core Desktop restarted, gaming plans It's FOSS: Core Desktop + firmware HWE proposal Why Canonical Is Reviving Ubuntu Core Desktop (explainer) Releases - canonical/ubuntu-core-desktop - GitHub - Ubuntu Core Desktop 26 DANGEROUS developer preview. canonical/workshop - GitHub - Ubuntu-native development environments: secure, fast, composable, and agent-ready. The Immutable Linux Paradox - Jon Seager Ubuntu Core - Canonical's immutable, snap-based Ubuntu for desktops, servers, and edge devices. Snap confinement - How strict confinement and interfaces isolate applications. LXD - Canonical's system container manager, used for isolated development environments. uutils coreutils - Rust reimplementation of the GNU coreutils shipped in 26.10. bootc - Container-image-based OS updates, as used by Fedora's image-based desktops. ntpd-rs - Full-featured Rust implementation of NTP, including NTS support. imagecraft - Tool for customizing and building bootable Ubuntu images. Introducing Myna: Speech to Text for Ubuntu Desktop - Desktop Myna speech-to-text is now available as a technology preview Myna - GitHub - Myna is a lightweight speech-to-text application for Ubuntu Desktop. Myna - Snapcraft - Install the Myna snap. Torsher - One Docker container with one database and one interface, made for the TV first. koya-bar - JavaScript UI on Koya's own GPU renderer, without a browser underneath. koya-os - Live ISO if you fancy trying it from USB rather than changing your current setup. Pick: List All Tabs (Updated) - A Firefox extension that lists every open tab across all windows, groups duplicates and busy sites (respecting Multi-Account Containers), and lets you jump to or close tabs individually or by site. zachleat/list-all-tabs: Firefox extension to list tabs, find duplicates, and close them. Pick: Tangram - Tangram is a new kind of browser. It is designed to organize and run your Web applications. Each tab is persistent and independent. You can set multiple tabs with different accounts for the same application. Tangram on Flathub Pick: Pinchflat-NGX - An enhanced, self-hosted YouTube media manager built on Pinchflat with Material design UI, PostgreSQL support, OIDC SSO, and disk staging.

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show.

0:11ChrisHello, friends, and welcome back to your weekly Linux talk show. My name is Chris.

0:16WesMy name is Wes.

0:16BrentAnd my name is Brent.

0:18ChrisWell, hello, gentlemen. Coming up on the show today, we're chatting with John Seeger from Canonical about all of the big changes coming to Ubuntu. And then we're going to round out the show with some great picks and boosts and a little look at something special and a lot more. So before we get to that, we've got to do the right thing and say hello to our crew that's sticking with us through a long Sunday. Time-appropriate greetings. Virtual lug. Hello, mobile rope. Hello.

0:40BrentHello. That's better.

0:41ChrisThere we go. Yeah, yeah.

0:43WesBeen with us all day and we appreciate it.

0:45ChrisAlso, you know who's been with us for a while? Our friends over at Defined Networking. Go to defined.net slash unplugged and check out Manage Nebula. When it comes to your network, when it comes to something as important as your mesh network, I want you, I implore you to own that stack. That's why I love Nebula. It's based on the open source decentralized VPN that they created many years ago to solve some very, very big problems. And it doesn't depend on any third party staying online or special dashboard, nothing like that. Because you can host your own dashboards. You can host your own,

1:17Chrismake dashboards for days if you want. You can make your own lighthouses. You can do the whole thing. And you don't have to worry about any third party. But I understand that can be, that can be a bit much. So I want you to check out Manage Nebula because Manage Nebula, you can start it with 100 hosts, absolutely free, no credit card required. And they now also will provide a lighthouse to really make things easy. It's great. So check it out at defined.net slash unplugged. You can come and go from the hosted lighthouse. You're going to really be impressed with the flexibility, the simplicity, the speed, and the security.

1:47ChrisDefined.net slash unplugged. Thank you to Defined for sponsoring the Unplugged program. Now, you know that it's reaching the end of the year when we make this announcement. Ladies and gentlemen, if you can believe it, the great holiday home lab is coming back, for a second year. Ladies and gentlemen, we want your home labs. We want them. We want your home labs.

2:20BrentMostly just your RAM, really.

2:22ChrisSend us your RAM. Just send RAM. We're going to have the criteria for you. We just want you to be ready for this. We're going to set up the submission page at linuxunplugged.com. Was it holiday? I think it was what it was. It's as we record, not live, but it should be live, hopefully, by the time you hear this, or there will be a placeholder there. And we'll have all the criterion for you, and we're going to have a few more details. We're changing things up just a little bit. And it will be our last episode of the year will be the great holiday home lab special.

2:54ChrisIt's back. It's going to be fantastic. You're going to love it. It was a lot of fun last year, and we love senior setups. And we encourage messy setups. We also encourage pristine setups. We love all of the setups. And it turns out last year's episode was one of our most popular episodes of the year. So, you know, we got to do it again.

3:26ChrisSo get ready get your home lab ready tidied if you need to or make a bigger mess and send it in we'll have more details and again when it's ready it'll be at linuxunplugged.com.

3:35WesSlash holiday yeah i think right that sends to the old google form right now so it's not Shh.

3:40ChrisNo, it doesn't. No, I just think what we'll do is, before we leave, we'll just replace it with a standby page, and then when we have it ready, we'll...

3:47WesThat makes perfect sense.

3:48ChrisYeah, there you go. Thank you. Ubuntu has been in the news these last couple of weeks. And Canonical has been giving everybody a lot to talk about. First of all, Ubuntu 26.10, the stonking stingray, was released on October 15th. Very fancy. Got the new GNOME in there and all of that.

4:08WesYeah, we played with a little bit.

4:10ChrisAnd in the past week, John Seeger and, Ubuntu announced plans to restart the Ubuntu core desktop with an aim to make the default Ubuntu desktop in the future eventually use Ubuntu Core one day. No timelines there yet, but that was the stated goal, which is no small thing.

4:31WesI think that means we better pay attention.

4:33ChrisThat could be a huge, huge shift. So we wanted to have John back on the show to help sort through all of this.

4:41JonJohn joined Canonical back in March of 2021. It's been that long. As product lead and became VP of engineering back in February of 2022. He's, of course, a longtime Linux enthusiast, hands-on open source developer. John has also worked extensively with NixOS, Kubernetes, Rust, Snaps, and Cloud Automation. He's currently also helping to drive Ubuntu's modernization, including Rust adoption, security improvements, and that revival of Ubuntu Core

5:14Jondesktop that you mentioned, Chris. That's especially exciting to me because I think I said I would try this for like a month when it came out. I think I remember something vaguely like that. So I'm counting on you, John.

5:31ChrisBack on the show right now, John Seeger returning. John, welcome back to the Unplugged program. We're thrilled to have you.

5:37JonThank you very much for having me. I'm glad to be here.

5:39ChrisWell, your timing is perfect because you've been making a lot of news recently. And there's a couple of projects that we want to dig into. I thought maybe just because I think it's the topic du jour right now, let's start with just all of the security stuff going on. There's been a lot of discussion about the recent IPv4 vulnerability in the Linux kernel and all of that and I know you guys are still as we record now working on a fix how's all that going what's going on there.

6:07JonIt has definitely been a pretty wild ride over the last few months. I think anybody who is shipping software at the moment is dealing with this, not least the kernel project itself, of course, but it is kind of replicated across all of the packages we maintain. The pace at which vulnerabilities are discovered and reported has obviously increased dramatically. The example you're showing there, of course, is one where we didn't cover ourselves in glory, but it is against a backdrop of quite an impressive story, I think. So the news on that one is the fixes are in proposed.

6:40JonThere are some instructions out on a blog on Ubuntu.com that tell you how to get those fixes. One of the interesting stories that I heard from our kernel team, we've all just come back from two weeks in Mexico. We had a product roadmap sprint followed by an engineering sprint where we took, over 1,100 engineers to a place in Mexico, which was kind of wild. The whole company was only 500 when I joined five years ago. But there was an interesting statistic presented to me, which was that in the last six months, our kernel team have shipped more security patches than in

7:11Jonthe prior 20 years of Ubuntu. And so that's not an excuse for the vulnerability that we missed and perhaps were less timely than people would have liked. But it is to try and give some confidence that the machine is being built and the machine is mostly working.

7:26WesYou have taken a lot of steps recently. I mean, just over the past year to sort of advance those timelines. I think maybe people don't appreciate that you're already shipping some of those things even if it may have not applied fully in this case.

7:38JonYes, we are essentially now committed to shipping kernels on a much tighter timeline. And I think often the scale of the maintenance burden that our kernel team have is perhaps underappreciated. So for Ubuntu, we ship over 160 kernels. That is a combination of cloud kernels and OEM kernels and partner kernels. We've obviously got the vert kernels, all of the kernels, essentially. And we have to essentially keep all of them patched and up to date. I think the number I saw was something like 23,000 or 25,000 security patches in the last few months.

8:13JonAnd so the team has had to ramp up really fast. I think a year ago, it would have been fair to say that we were feeling the heat a little bit, but I am increasingly confident that team has built a machine that will serve us well. There will, of course, always be vulnerabilities that we miss. They've committed to a much tighter release cadence. And so they are essentially on a two-week release cadence for kernels, but staggered. So there are essentially new kernels dropping every week, but they kind of lag by a week, essentially. And that's because we run thousands of tests on quite literally thousands of

8:46Jonreal machines, like over 2,000 machines in a lab somewhere that are being driven autonomously through our kernel release and test QA process. And so the idea is that fresh kernels should be arriving near enough every week, which is something that Canonical has never committed to before. So imagine 164 kernels, thousands of tests over 2 000 machines essentially every week and so i don't want to make excuses for the vulnerability that got lots of attention this week that was obviously a miss on our part. And i think it's safe to say we could have communicated more effectively but

9:17Joni you know that's now on the way to fix and i hope that people hearing this will have a bit of faith that, you know going forward they are in safe hands we have some good numbers to support, i think the level of scrutiny that goes into our kernels perhaps we'll do some writing about that shortly.

9:33BrentWith the accelerated pace of the kernels that you mentioned, which is crazy numbers, is that, from a more philosophical perspective, a welcome change? Because all these bugs are being found that presumably existed and were undiscovered. How do you feel about that?

9:51JonHow do I feel about it? I mean, I don't suppose it really matters whether it's welcome or not in reality. It's the reality that we're in, and we have to respond. I think the Kernel team is one of the largest teams at Canonical. And like I said, perhaps a year ago, we were feeling less confident. But with some collaboration from some of our partners, some pressure from some of our partners, some, I think, fair pressure from our partners, it has been very interesting to see that team step up. And I think despite the fact the workload has gone up because of the sheer volume of security vulnerabilities, I think even against that backdrop,

10:26Jonthe machinery they have built has allowed them to be more ambitious, if that makes sense, than they have been before.

10:30WesThere's a lot of times, I think, where sort of having to adapt, move faster, automate more does sort of expose hidden bottlenecks in your process or things you have to change, things you've always wanted to change.

10:40JonThat's right. You'll notice you might have seen that in the run up to 2610, we squeaked in the 7.3 kernel kind of at the last minute. We committed publicly to the 7.2 and we essentially switched fairly last minute to 7.3, which is, I think, a testament to the kernel team's confidence in their process.

10:57ChrisWell, why don't we talk about Ubuntu 26.10 really quick, since that is also the new release. And there's a couple of things in here that are pretty great. We've mentioned GNOME 51 on the show, of course, which has a few really nice things in there, including some nice remote desktop improvements. But there was a couple of items that I think we've been tracking for a while that jump out to us. Number one, obviously, is the transition to Rust Core Utils, I guess, is complete in this release. Is that true?

11:25JonThat's right. So in 2604, it was complete apart from three utilities. And holding those three utilities back was kind of an abundance of caution. I don't think anyone would have got into real trouble, but we were aware of some slightly edge casey time of check, time of use errors. So we held those back. Those issues got fixed by the upstream. And so it felt like 2610 was the time to take the lead off and just commit.

11:49ChrisHow's it been going?

11:50JonIt's been hard. Core Utils has been hard. There's so many utilities. It's so baked into F1 shell scripts, I think it's been successful. I remain committed to it. I was reading an article today from somebody who I think was at the University of Toronto who was pretty upset that sudo-rs had broken one of their workflows. That's a slightly different prospect to core utils because with sudo-rs, the promise was never 100% compatibility. It was a modern reimplementation that skewed some of the choices that were made by the original project. Whereas in new utils, the goal is bug for bug, 100% compatibility.

12:24JonAnd that project has really risen to the challenge. The test coverage, the sort of test parity, if you will, between UUtils and GNU has continued to go up release after release. And what's also really pleasing is I consistently see Sylvester, who's the lead maintainer of UUtils, appearing as one of the top contributors to GNU CoreUtils as he works on his own project and discovers bugs and inconsistencies and maybe emissions in the docks and things like that. So I think it has gone, you know, it's been tough, but it's also been worth it. And I never expected such a transition to be very easy and very fast.

12:57JonIn reality, we were first, and that comes with a responsibility for us to respond quickly to issues. We've also committed to backporting new versions of the Uutils package to 2604, which we wouldn't normally do unless there was a critical kind of security issue. But we're doing that because we recognize that the software was early and some people will be facing issues that we want to resolve for them as a kind of... You know, exception policy in this case.

13:22ChrisSo now that we're here, what is the goal going forward? Or I guess another way for me to phrase that is what are you hoping the dividends that pay off now that you're on a Rust core utils?

13:33JonSo in the long term, it should be less maintenance for a certain class of bug. I'm not exaggerating or making stuff up when I say that the vast, vast majority of security vulnerabilities out in the wild have been up to this date due to memory corruption issues. And I hope is that we will cease to see that with core utils. I also think we're starting to see some nice features. So the latest release has this nice trick where if you type an argument slightly wrong, you put the argument value in the wrong format or you slightly mistype it,

14:03Jonit gives you a Rust compiler style warning and highlights exactly where it was that you typed it wrong and relates to the help. There's some nice performance improvements coming. And so my view is that that is just those sorts of UX changes and improvements are very difficult to land, I think, in the original. There's no appetite to do so, and I respect that. I think that's absolutely fine. But the user experience that I aspire to for Ubuntu is more than that, ultimately. And I think a project like EU Tools will enable us to get there, even if in the short term there's a few rough edges.

14:34ChrisDo you think you're on track for the LTS? Yes.

14:39JonIn what sense?

14:41ChrisIn shipping and making the transition to the Rust utilities in the LTS.

14:45JonI mean, it was done in 2604, apart from three tools.

14:48ChrisOh, it is. So this last three. Okay. So then I'm presuming then the answer is yes. By the next LTS, these last three will make it.

14:56Jon100%. And so by the next LTS, you'll also have NTPDRS by default. And that will be the first time, I think. So it's NTPDRS plus some new stuff, which we're funding the development of. And so that will get you a single binary NTP client server that also speaks NTS and also does PTP without an additional tool, like precision time, without an additional tool. We are dropping BZIP2RS in by default early in 2704. And we're evaluating Z-Standard RS, Z-Lib, and various other compression tools,

15:28Jonwhich are actually very exciting because, that is where the efficiency gains that one might seek from a Rust rewrite could have enormous global scale impact. So imagine all of the Ubuntu machines on the planet suddenly got 5% more efficient at doing Z-Standard or Z-Lib decompression, for example.

15:48ChrisRight, right.

15:49JonQuite a meaningful change.

15:50ChrisA lot of them out there doing that. So is there a transition taking in place with the internal tooling that Canonical is writing? Are you writing that in Rust? What's the direction there?

16:01JonSome. So Canonical has always taken quite a disciplined approach to the languages used internally. So traditionally, it was lots of Python, lots of Go. Now it's lots of Python, lots of Go, and some Rust. And we're increasingly trying to be discerning about where we use those tools. It's not, as some might have suggested, a religious game. There are things for which Go is very good. Go has very powerful primitives for concurrency. If you're writing web services and doing HTTP handling and things like that, Go is a phenomenal choice still. Even some of our systems tools like SnapD are written in Go.

16:32JonBut equally, if you want to write very low-level compositors and database protocols and things like that, Rust is a very compelling choice. So we try to use the right tools for the right job. And so we are definitely seeing more and more Rust projects popping up around Canonical.

16:47ChrisOkay, so you mentioned Snaps. You boys all right if we transition to this one? You ready for this? Okay. So the big news that you made recently, and this is pretty exciting, is that, quote, you are accelerating the Ubuntu Core desktop, the immutable and attestable Linux distribution, and you say specifically the desktop, too. So we all know Ubuntu Core, and it kind of seems to have faded a little bit, so it's, back john what's going on what are the details tell us the story here.

17:19JonYeah so ubuntu core itself has been pretty widely deployed on sort of iot class devices digital signage robots those kind of things, uh and continues to to be deployed as such you'll remember 2023 there was a sort of proof of concept core desktop and then it sort of all went a bit quiet and that coincided with about the time that i took over ubuntu and it's one of it's been one of the things i have been most desperate to get back to. But the sort of cold, hard reality of it was that I didn't feel like we had, the right team or resources to commit to that on that timescale.

17:53JonAnd a combination of some exceptional hiring from my team and kind of growth in the department has meant that I now feel pretty confident that we can tackle it. And honestly, I think this is the most exciting thing I have or maybe will get to announce at Canonical. Because for me, it is the most prolific kind of reimagining of Ubuntu's architecture and makeup in its history. It is replacing the bootloader as a snap, the kernel as a snap, the desktop environment as a snap. And so we have revived that original POC, which still has some architectural

18:26Jondeficiencies, but is now rebased on core 26. And if you know where to look on GitHub, you can get the images and boot them. We're working on a kind of daily release pipeline and now we'll aim to break down some of the more monolithic aspects of that proof of concept into component parts.

18:40ChrisOkay, so there's a lot about this that really bakes our noodle. It's a huge job, and I do want to get into some of the aspects that seem like they're going to be challenging, but... Man, it seems like if swapping out a few CLI tools was controversial, this is really going to rock the boat when people actually wrap their head around what's coming. Because, John, you're coming right at it and you're saying this is going to be the future of Ubuntu one day. And in a way, for a lot of people, I think Ubuntu is kind of like the bastion of a traditional distro where I can install it and I can just run anything I

19:10Chriswould download and et cetera, et cetera. There's got to be at least a portion of the user base and so many of the things that are based off Ubuntu that kind of have this expectation just baked in of how a Linux system works. This is going to be a pretty big transition. Like, are you crazy?

19:25JonA little.

19:27ChrisAll right.

19:27JonBut here's the thing. So traditional Ubuntu, Ubuntu as it sounds today, is not going anywhere. Like, no one should read into this that this signals death for Ubuntu as we know it today. For a start, Ubuntu Core cannot exist without the packages that we build, that we inherit and sync and patch from Debian, right? So Ubuntu, as you know it today, if you are someone who has absolutely no interest in Immutable Linux and Snaps, Ubuntu is going to stick around. You can carry on using it. Don't panic. I hope that there are no plans on any timeline to stop doing Ubuntu as we're doing it today. There is a plan that over the next couple few years, Core will become the default Ubuntu experience.

20:05JonAnd in my opinion, there are very good reasons for that. I know I don't need to sell you folks on immutable Linux distributions, but Ubuntu Core has some really interesting properties, even among the current drop-off Linux distributions. So one of the first criticisms I heard on this post was, well, I've used an immediate Linux system, and every time I update a package, I have to rebuild an image and reboot the system and redeploy the whole thing. Not true on Ubuntu core. Just refresh the snap in place. We can live patch the kernel. So that's not a drawback. It also brings that lovely, I want to switch the desktop

20:36Jonenvironment for a day and see what happens type effect without crapping all over your system. You can just install the KDE snap, boot that for a while remove the KDE snap and go back to GNOME I hope that there will be cosmic snaps and you know NERI snaps and all the rest of it and I think. 2026 has never been a better time to think about what confinement might look like on your workstation. I think it is only a matter of time before we see.

21:06JonVery influential people having their credentials leaked because an agent did something silly. And the way I run my machine is I have very little installed on it. I have a browser, I have a chat client, I have a terminal, and pretty much everything I do is in a LexD container or a VM or something like that. And And there is a story here on the compatibility front that really has come into focus in the last year that I couldn't have told two and a half years ago when I kind of regrettably put a pause on things. And so where I am moving towards is why not use Canonical Workshop,

21:40Jonwhich is this nice little utility that we shipped about six to eight months ago for building little reproducible developer environments. Use that as the sort of compatibility story. So today I have been using Firefox installed in a workshop. I launch it from my GNOME launcher and it pops up on my desktop. It's wired up to D-Bus through D-Bus proxy. It's wired up to Pipewire through Pipewire container. The notifications work. The camera works. Web Serial even worked. And that Firefox is installed as a dev running in a workshop.

22:12ChrisOkay. So, okay, I got it. This is a lot. So first of all, what I caught in there is it doesn't sound like alternative desktops are out. Still going to be able to run Plasma and Cosmic. That's all going to be still a thing.

22:22JonWe're actually working with the KDE folks right from the start. So we are paying a company who's quite heavily involved in the KDE Neon community to work alongside us, in our teams day to day to basically develop the KDE, like cutting edge KDE track alongside what we're doing with the GNOME track. So right from day one, we essentially put ourselves in a position where we can't make decisions that would inhibit us to just one desktop environment or one login manager or something like that.

22:52WesWhat kept me from having a proper website for a long time was it just felt expensive and time-consuming. I mean, for some folks, it means you have to hire someone. For others like me, it means at least a weekend of research to look at a stack of tools. Hostinger's AI Builder flips that. Describe what you want, and it builds a working version of the entire site in minutes. No coding, no design skills required. And you can start from a mock-up before you even make an account. That's the part that actually got me to try it, I'll be honest. What I also like is that it's kind of an all-in-one place for the domain,

23:23Westhe website, the email, the marketing that makes it easy for you to get started or a friend or a family member. One Hostinger account covers it all, so no stitching together five different subscriptions in your domain provider. And if you already have a site somewhere else, Hostinger makes migration quick and easy. You're not locked into whatever you started with years ago. And right now, Black Friday and Cyber Monday pricing is live. A premium plan is just $1.99 a month. Snag Hostinger's service at their lowest prices before the sale ends. Create your website today for just $1.99.

23:56WesGo to hostinger.com slash unplugged and use the code unplugged for an additional 10% off Black Friday and Cyber Monday pricing. Hostinger.com slash unplugged.

24:10ChrisOkay, then the other thing, it sounds like you called it a workshop. This is almost kind of like a separate contained environment where you could, essentially sideload applications that aren't snapped or for some reason you didn't want a snapped version.

24:23JonThat's right. So think, I'm just going to say this, various parts of the internet are about to explode. So strap in. Think Windows subsystem for Linux with the graphics bit. So you can install, like you could even install a Flatpak in a container and the window can appear on your Ubuntu Core desktop as if it was running on your desktop, except it's not, it's running in a container. And that is essentially the, that's the escape patch. So obviously I see this as an escape patch. If we do our job right, all the software vendors in the world are going to want to ship their stuff for snaps because the security story, the update story,

24:57Jonthe kind of usage metric story should be so compelling for them and for their users that they want to ship their stuff for snaps. But it is no good for us to go out in the world when none of the apps are snaps and tell everyone that this is the future of Ubuntu when they boot it up for the first time and can't get the stuff done they need to get done. And so workshops and containers and the appropriate kind of wiring and polish around how you wire those apps into the desktop are kind of the escape hatch. And that's going to be a focus. I mean, I'm running a sort of prototype of it today, in a sense.

25:24ChrisIs this essentially like LXD under the hood, kind of wired up?

25:27JonIt is LXD.

25:28ChrisOkay, yeah, okay.

25:28WesAwesome.

25:29ChrisThat's great.

25:30JonSo workshop, canonical workshop is this super nice little thing that we actually imagined three years ago for basically confining... A board vendor SDK. So often if you're doing IoT development or board bring up enablement stuff, you end up with these kind of large SDKs, lots of files, lots of libraries, lots of utilities. And if you're working on boards from five different vendors, this stuff can end up all over your system and conflict and be a bit of a pain. So Workshops was basically a way of like, how can we wrap up an XD container in a nice declarative kind of YAML format that can be shared among teams?

26:02JonAnd so the idea is in your repo, you have a workshop.yaml, you type workshop launch, and you get the same developer environment as the rest of your teammates, makes much like a nix shell right much like a nix flake shell something like that could be interesting in a desktop.

26:13ChrisEnvironment for sharing those workshop environments around.

26:16JonAnd that's what our teams are doing and and some of our customers and users are starting to do so we've built this product and it sort of only recently occurred to me that we could, extend the set of interfaces and integrations that we have with workshops to make it very easy to say well when desktop apps launch in this workshop the app needs to show up on my desktop and make that very seamless. So that's the kind of, that's the get out of jail free card. But actually what's really important to me is that we get the first class experience, right? Because I don't love the idea of running a totally unconfined browser on my system.

26:47JonI certainly don't love the idea of running codecs or open code on my system where all my credentials are. Like, I think we all agree probably that that's not great hygiene. And the problem is it is very easy for people to do that today on Ubuntu. You have to know what you're doing. You have to know that you don't want to do that in order to not fall into that trap. And I think for me, one of the attractive things about Core Desktop is we don't, give our users the option to put themselves in that position. They will have the flexibility to get done what they need to get done,

27:18Jonand it should feel smooth and seamless, and we shouldn't get in their way, but they can't accidentally install an agent that's going to root their system on the machine. It doesn't work. So that's kind of the thinking. And there's three use cases I outlined. One was the kind of Chromebook use case. Boot to a browser, lots of users. For lots of home users, that's going to be enough, I think. Just a very nicely optimized, accelerated browser that automatically updates in the background without you ever doing anything. Like, we can achieve that today, right now. Like, I could send you an image, you could boot it, it will work.

27:49JonThe other two are interesting for two different reasons. So one of them is a developer workstation. So that's, you run VS Code, you've got workshops, You know, you do all your development in containers. We have a nice VS Code extension for Workshops. It's aware of all the Workshops in your machine. You can remotely attach to them, all that kind of stuff. And then the third one is a bit of a wild card, a bit of an underappreciated part of the world in Ubuntu so far, which is a gaming variant. So this is boot Ubuntu Core straight to Steam or the new Epic Games launcher.

28:17ChrisThis one feels the most ambitious.

28:20JonWell, is it though? I mean, we have a Steam Snap and it works.

28:22ChrisYou know what I mean? Tell me if I'm wrong. John, it feels a little bit like you're saying to Val, like, hey, take a look at this. You could be doing it this way. We got something over here you could use. That's the part that feels a bit ambitious to me.

28:37JonI'm not not saying that, but I'm also not saying that. And so, well, I guess what I'm trying to say is this is not about going after Valve. This is about we have a huge user base.

28:46ChrisYeah, yeah.

28:47JonI would love, so my view is it is on us to make this thing so good that they want it. When they built SteamOS, we didn't have the thing they wanted, clearly, or they would have spoken to us about it, right? And I think if I were Valve, do I want to be maintaining a mirror of Arch Linux and all this kind of AB boot and Atomic update? Do I want to be maintaining that? Is that the USB for Valve? I don't know. Maybe it is. Maybe they are a Linux company. For me, we're going to build Ubuntu Core anyway. And people have desktop machines, increasingly desktop machines on Linux that they play games on.

29:20JonWe already have a Steam Snap. We are going to have a Snap for the new launches that are coming out from other games companies. And so why not also have the ability to make that experience great on a Steam Deck or on a Legion Go or on a whatever, right? I think there's lots of opportunities there.

29:36ChrisIt could make sense, too, from a desktop compatibility story for the non-Steam Deck devices as well. That's the bit that feels a bit ambitious, and I like that you're going for it. It's also going to push all of the edge cases on drivers and all of that stuff.

29:48JonThat's kind of why I'm doing it, should I mean. So these three initial use cases, I've had people say, well, it seems a bit limited. Why just those three? But if you look at the three use cases, they are three use cases that are going to push the implementation in different ways and highlight to us where we still need to do work before we can make this the default. So my goal is that very soon, as in on the order of the next three to four weeks, there will be images that people can get their hands on. They actually are already out there hiding on GitHub if you look somewhere. But daily images that we are a bit more polished.

30:20JonBy 2804, I would like us to be releasing Ubuntu core images alongside the standard Ubuntu. And then we'll see what happens after that.

30:28ChrisSo let's say some guys found those images previously. And they took a look uh i was surprised i think the entire base for the desktop is only 20 snaps right now i thought it'd be like 100 snaps.

30:40JonWell therein lies the problem so that original proof of concept does indeed it's a few things that we have to sort of quite desperately fix uh one is that in that original proof of concept which we have basically taken and just lifted into 2026 so there's a few few patches that needed to be rebased on snap d there's, all of the snaps got lifted from core 22 up to core 26 but what the cheat is it kind of jams the whole of gnome into a single snap yeah okay, and that's obviously not the answer right so the whole desktop not even just the whole of gnome like essentially the whole desktop so our primary kind of

31:14Jonpiece of work over the next few months is going to be figuring out like what is, the correct decomposition of a desktop clearly we're going to have pipe wire and wire plumber do they live in a single snap do they live side by side in the same snap we're going to have things like lib camera we'll have gdm we'll have sddm where do they live also, for a login manager that needs quite a lot more permissions than a standard snap and so we'll have to do custom work in snap d to enable new interfaces and allow login managers right because.

31:40ChrisAs of right now that the login manager actually is not snapped is it or it's not confined at the.

31:45JonMoment uh so that is one of the things on the images we've just made i believe is true so certainly for gdm we have managed to make that confine but with patches to snapd so it's a upstream version.

31:57ChrisOf snapd at the moment to make it work i.

31:59JonSee yes so it's all very custom so the idea is so one of the reasons i haven't widely, distributed those images and the build tooling is because we're just doing some work there's a new, tool that we have been building in alongside all of our other craft tools called image craft image craft much like snap craft and rock craft will be a single yaml format if you know those formats it will look incredibly familiar to you and we have a working image craft for Ubuntu core desktop which is, beautiful like so much nicer than anything i've ever seen to build an ubuntu image and as soon as that is ready,

32:31Jonessentially it will be possible for anybody to cut their own custom image custom ubuntu core desktop image by editing like a hundred line yaml file which is kind of wild that it's it's almost in the kind of almost as easy to cut your own images nixos kind of territory declarative.

32:46ChrisImages i mean i love it all right all.

32:48JonRight yeah yeah.

32:49ChrisOkay so it sounds like you're going to be hitting a lot of the edges that will be pushing probably on snap and snap d itself to get here yeah well that's probably almost worth.

32:59JonIt and i mean so we have work in the kernel to do we have work in app armor to do we have work in snap d to do we likely will have to patch bits of the desktop itself like this is not something that is going to land in six months right and it is going to be quite the challenge but i, i have to believe that this is the future of ubuntu if you look at what is going on in the world right now with security and agents and and confidential compute and verified boot, measured boot, I have to believe that this is the answer. There are so many of those things that we can do to an extent on traditional

33:32JonUbuntu, but we'll always be hampered by the fact that the file system is kind of floppy. Every process can kind of access everything, modular some file system permissions and some AppArmor rules. And I think the way we tidy that up is with Core.

33:43WesI mean, it seems like we started this interview talking about the security story, right? And there is a security story with this change. I think maybe also there's a velocity story because the securities have kind of is also moving faster, right? So does core change what you can do and how you can deliver updates?

34:00JonYeah, I mean, firstly, most of the system will be on autopilot. Like, you know, snaps update in the background automatically. We can choose that cadence. They also have this nice property where you can say, well, I want to be on Ubuntu 2604, but I want Gnome from 2804 or from 2710, for example. And that's quite hard to achieve on a normal Ubuntu system without doing quite messy things with apt that might result in you wedging your system with a snap. You can just say, actually, I want to point the desktop stack at a different track and pull that in. And there's no reason why it shouldn't work.

34:31JonSame with for example lib camera and pipewire so often one of the things that happens is users of the lts as we're getting towards the next lts so 18 months nearly two years after the last one they download the lts that is currently supported and the shiny new camera module that was just, just built six months ago doesn't have support because it's got the wrong version of lib camera, maybe the wrong kernel version whereas with, with core it's much easier to kind of pick and choose bits and and the oems will be able to do the same so dell would be able to say for example

35:01Jonship their new laptop with the 2604 lts but in their image they could pick up the snaps and have the support for that camera without having to do kind of like a hacked up frank and buntu thing that's really hard to support right some custom.

35:11WesRight oh it sounds like a much more composable system that's impressive.

35:15JonThat's exactly the word i would use it's composable.

35:18BrentJohn with this aggressive timeline you mentioned 2804 as being the default.

35:23JonNo not for the default, so i i haven't yet said this publicly but i don't see any harm in doing it other than perhaps raising the cortisol levels of my team slightly so i would like the images to be available for people to play with in 2804, my like stretchy stretchy stretch goal is to have this ready for being the default in 304 but i would like to caveat that with that is an incredibly aggressive timeline given the amount of work we have to do what a great release that.

35:50ChrisWould be what a great number that would be that's so great.

35:53JonI love that ambition right uh and so that's what i'd like to do now, I think the reception has been really interesting, more positive than I thought when you say to someone, hey, you know those snaps that Reddit loves? We're going to put them everywhere. It's the only way you can install stuff. But I think in reality, the narrative is starting to shift there. You know, I think I've said this before, and I'll say it again. Having explored the Boot C ecosystem for some time, I spent a long time running Nix, even while working at Canonical. My first three and a half years at Canonical,

36:27JonI was daily driving NixOS. The things that Snaps are good at, they're very, very good at. And the things they're not good at are not existential limitations. They're things that we have yet to implement. And I think one of the reasons we haven't implemented them is because it hasn't been the goal to make it the default. And now I am going to be orienting the teams internally very much towards this goal. This is absolutely the focus for the Ubuntu Engineering. Org is going to be landing this.

36:58JonSo maybe that changes the calculus slightly. The SnapD team is also in the best shape I've ever seen it. The leads there are doing a phenomenal job. They're very excited. And so I have to believe that this is possible.

37:10BrentOkay. Reframing then, 30.04 as the default, given that still accelerated timeline for something so, let's say, different than current Ubuntu, it's going to be snaps everywhere, as you mentioned. Yeah. What does that mean for the Snap infrastructure and delivering Snaps to users in a reliable way? Is that obviously a backend that's going to need to scale, but do you have ambitions to have that backend be Snaps as well?

37:44JonYes. It's no secret that we have had our fair share of infrastructure difficulties over the last year. Some mostly inflicted by some bad actors hurling traffic at us in all kinds of shapes and sizes at critical points. Some of it is our fault, and we're working hard to fix that. We have quite a lot of new infrastructure coming online over the coming weeks and indeed a plan on how we will do a more comprehensive job of global distribution of snaps if you will so more points of presence in more parts of the world the store itself is already,

38:17Jonquite scalable it's built in such a way that i'm very confident we'll be able to scale it it's already i mean it's grown arms and legs it's no longer really the snap store it's the snap store it's the charm store It's the image store. There's a bunch of store backends that all essentially collect artifacts from RockCraft, CharmCraft, SnapCraft, ImageCraft, SDKCraft, etc. So the underlying engineering there, I think, is strong. How we deploy it and how we scale it to satisfy this is an obvious question,

38:49Jonbut a question to which I think we have an answer.

38:51ChrisSure, sure. I think what I what I always think is absolutely wild about what you do at Canonical as a company is these just absolute insane support timelines, for enterprise customers that absolutely demand it. And it feels like if you're lucky and you get there, then like the 2030 versions, of an Ubuntu LTS could be versions that are truly supportable for a really, really, really long time in a way that's actually sustainable by the company. And that just is clearly a feature enterprises want. So, yeah,

39:23Chriswe've been talking a lot about the desktop, but this has to totally change the game for LTS enterprise support once they start making the transition, right? Right.

39:33JonI think so. And I think the cell for Ubuntu core is kind of more clear in the enterprise setting. So I want to be very clear that it is absolutely my intention to make this great for the tinkerers, the hackers, the Linux enthusiasts. I am one. Do you know what I mean? And all the time I'm doing this job, I'm somewhat committed to using Ubuntu and I don't want to build a thing I don't want to use. We've had nerdy conversations about the weird crimes I used to commit with NixOS and Arch and all kinds of other things. I don't want to build a thing I don't want to use. But I think the cell for enterprise is much clearer a system that where every

40:03Jonapplication is confined there are very clear models about what every binary on the system can and can't do it's easy to roll out updates completely automatically the updates are atomic they're transactional right the, user user data is kind of backed up and snapshotted before apps are updated there's this kind of composability of okay well, my my workstations over there all have nvidia gpus in them because they're the ml people and the servers over here have melanux cards in them and they need something weird in the Enit ID, all of those problems, should be easier to solve with core.

40:35ChrisAnd then there's the application support levels too like all of that too yeah that's.

40:39JonVery right i think especially like legacy in quotes applications yeah you know the the whole point about snaps of course is they bring their runtime the base with them and so it is eminently possible to keep either very old versions of software working on very new linux or very new versions of software working on very old linux and the whole point is that the machinery takes care of that. And we see, like, that is something that is proven. Like, I am happy to be, you know, I'm happy for people to dispute snaps on the desktop where they, you know, I don't want to minimize that people have had bad experience with snaps on the desktop.

41:13JonI've seen apps that don't work. Some apps don't work. It's hard to make them work in the same way that, you know, you NixOS loving folks will recognize that sometimes packaging apps for Nix is hard. Like you have to patch them and they do unexpected things. And we have all of those problems. But I have to be honest with you, I run a basically stock Ubuntu system, and every single app that I interact with on a daily basis for the last year and a half has been a snap, and it has bitten me once or twice, and not for very long. And not only is support for snaps getting better, but the way that we understand

41:45Jonthem and facilitate them is getting better. And I think my hope is that we can build a system that's genuinely very compelling. Like I want an Ubuntu core home assistant.

41:53ChrisYeah. No kidding.

41:54JonThat automatically updates and snapshots the data and takes care of all the backups. Like I want that. And I think we can make something fun for the non-enterprise community too.

42:03ChrisWell, here's what I'm hearing. So we started talking about the recent security issues, and you mentioned the pipeline that you've been building out at Canonical to deliver patches faster. And then we talked about delivering Ubuntu core experience where individual components can be updated and swapped out. And with the rate of security issues we're seeing today, that just seems like a thing that not only enterprise users are going to want, but eventually regular users are going to appreciate the ability to have a stable base where those components can be updated when security issues come out. Like they're really, if you think about it, is a pretty complete security picture

42:35Chrishere from production of patches to deployment to end users that, you know, if you can execute in the right timeline, you may be meeting the market right where they're at when they realize they need something like this. So that's pretty exciting for you guys.

42:49JonYeah. And we do have genuine use cases in the market with real customers, big customers, customers whose names you will definitely know where snaps have proven to be the thing. They were, they were reticent. They lean into it. We help them. And so, you know, I accept that there have been difficulties, but I also, I flatly reject the idea that they are just unsuitable for the desktop and they don't work because I have no evidence to support that in 2026.

43:18ChrisSounds like you're going to try to prove it. You're going to try to prove it works.

43:22JonI'm not going to try, Chris.

43:25ChrisI love it, John. Very good. Well, thank you so much for coming on and updating us on everything. You guys have been super busy. There's been a lot going on. And we're absolutely going to follow the new core work with super interest. We're going to be trying it. We've already pulled down the images and been poking around. So we're already pretty excited. So keep up the great work, John. And thank you for joining us.

43:46JonThank you very much for having me. I appreciate it.

44:00BrentCanonical has introduced Mina, an open source privacy focused speech to text tool that runs entirely on your own Linux desktop without requiring cloud services.

44:12ChrisAnd you can get your hands on it right now.

44:15BrentRight now.

44:16ChrisIt's not installed by default on Ubuntu 26.10, but it is very easy to get up and going.

44:21WesYeah, technical preview, I think, right?

44:23ChrisMm-hmm.

44:23WesTechnology preview.

44:24ChrisIt's a snap, obviously, when you try it on Ubuntu. I think the goal is to try to have this thing in production shape by 2804, which will be the next LTS. And you can also get it installed outside of Ubuntu if you'd like. Like Brent said, it is entirely local. Speech recognition happens on your device. It's offline capable. And it supports multiple different backends, depending on the best fit for your hardware. And hardware accelerators if you got it.

44:50BrentIt feels to me like they've really been listening to their crowd with this one. Going all local is, of course, going to get the, well, dear folks of ours like PJ, who really, really, you know, that counts for a lot of people. So I feel like they're listening to their crowd, or at least our crowd.

45:08ChrisIt feels like baby steps to me. But voice dictation is kind of, of all of the different local models that are pretty easy to run on different, grades of hardware, voice dictation is kind of where we're at. It's like the best showcase for that.

45:23WesYeah, they've got it pretty nicely set up, especially if you're on Ubuntu. There's a wide, you know, the most recent few releases all support it because you can just do it in strictly confined snaps. So each piece, you know, there's Mina, Mina Whisper, Mina Parakeet. So you can have the inference side just attached via Snap. And Snap confinement is useful if you want to make sure that this kind of data is kept where it needs to go.

45:44ChrisNow, you can also install it by building it. You know, it's a Python app. You don't have to use Snaps if you don't want to.

45:49WesYeah, it's got two parts. There's Python on the server. That does the inference part, right? So it's got adapters for Whisper, Parakeet, Fun ASR. You just kind of do pip extra. It's got UV in the repo there. All very standard, supports CPU as the default target. NVIDIA GPUs are also supported. There's plans to do other accelerators down the road.

46:10BrentNice.

46:10WesAnd then there's a client, which is Rust.

46:13ChrisOh, wait for it.

46:14WesYeah, come on.

46:15ChrisSorry, sorry.

46:17WesRust. Yeah, Mina Desktop Demon. No GTK, just Tokyo ZBus GetText, plus a little Mina Testbed CLI. There are GUI crates, too. So there is the pill, I think they call it, that you can get for the GUI. I haven't tried that yet. That needs GTK4 living with you. Kind of the interesting part on text injection, how it actually works, it's got the IBUS injector, which speaks the IBUS-DBUS wire protocol via that ZBUS library.

46:48WesNo FFI. Registers the component and the engine and then becomes the global engine during dictation. So it hooks kind of right into how the system expects this to work and takes over. Yeah. It also has that, because of that, it has the smarts to reject password and pin fields.

47:04BrentOh, that's cool.

47:05WesI don't know if that's configurable. Mm-hmm.

47:07ChrisSo functionally, the way this works is it's kind of like a push-to-talk system. And it sets up a hotkey on your box, and then it inputs the text into whatever is an active input field. And I have an Aurora Silverblue slash system here, and I've installed it on this system. And if I hold down super shift T, it should start working, and then it should transcribe and input whatever I say. So I'll give it a shot here. So what did I say? Super shift T?

47:35WesThat's right.

47:37ChrisOkay i am now speaking and we'll see if it dictates whatever i say into the thingy hopefully it does and i'm going to release the super shift t now nothing happened.

47:47WesI don't see notifications because i was going to say if you don't install the gui part.

47:51ChrisYou don't get the notification.

47:52WesWell no uh the main interface is system notifications which worked pretty well um i got it running on my system in plasma and so just pop up and say hey now transcribing and then.

48:01ChrisWhen it was done it would okay i'm going to try it again i because i do see it in my system tray i'm gonna try it again what i say was shift super shift t super shift t okay i'm trying you again and i do see my system tray icon changing, and i have firefox set as the input window, it didn't work this time.

48:17WesAnd for mine you didn't hold it it was just.

48:19ChrisOh should i do that oh really i don't hold it maybe i'm doing it wrong.

48:22WesYeah like triggers it it's press to talk again.

48:24ChrisOkay so with hex it's push to talk maybe that's what i'm thinking about, Okay, now I've just pressed it once. Oh, yeah. Oh, yeah, that was it. You only press it once. So it was capturing all of that, the entire time we were speaking. It was capturing all of that. So let's try it again. Okay, let's try it one more time. So I'll do super shift T, and then we'll see how it does, right? Super shift T. I am the captain of the USS Linux Unplugged, and I am trying to transcribe my

48:59Christext for future alien races. Then do I press it again to stop? Yep. Okay. Super shift T. Yeah, there it goes. Boom. Make it so it captured the whole thing. I am the captain of USS Linux Unplugged and I am trying to transcribe my text for future alien races. Then do I press it again to stop? Yep. Okay. Super shift.

49:22WesThere we go. Yeah, I tested it over here earlier and from you talking way over there.

49:26ChrisYou picked it up?

49:27WesYeah.

49:27ChrisHuh. It's pretty quick.

49:30WesBy default, it'll just auto-download Whisper Tiny. But you can change that. You can point it at local stuff if you want. It can do parakeet, too.

49:40ChrisI kind of prefer push-to-talk, but I think I could learn that. That seems fine to me. The thing is, I want to do it one more time. Brent, you try it, okay? And I want to watch while you do it. I want to watch what my CPU usage is. So before we start, my CPU usage is at a baseline of around 3.9% at the high end, all right? So I'm going to press it, super shift T, and then you speak now.

50:02BrentHello, I'm here reporting how many tabs I currently have open, but the counter that I'm using doesn't actually go that high, so I have no idea.

50:12ChrisOkay, super shift T again. And I saw it look like, I'm going to run it one more time, but it looked like CPU usage was only 10% there for a second. That doesn't seem right. Maybe that capture was too slow.

50:23WesIt does, I think, use...

50:24ChrisThe GPU?

50:25WesNo, it's aimed at CPU by default. So they have kind of put some thought into that.

50:29ChrisOkay, so they've put thought into...

50:30WesBecause if it's, you know, they're going for some sort of like intent-driven Ubuntu maybe in the future. So they want, if you ever wanted it to be a widely deployed feature, you're going to need it to be able to run ideally right across a lot like whatever people have, even if they don't have a nice.

50:44ChrisOkay. So if you, by the way, if you're dictating and it loses focus, it stops dictating. All right. I'm going to run it one more time. I'm now using a plasma system monitor and I'm going to do, okay, I'm going to do some dictation here, boys. Super shift T. Okay. I'm doing one more real world test here and my CPU usage is at 24%. So we went from about 4% to 24% and I'm going to do super shift T to stop it. Oh, now the, okay. Okay, CPU spiked to 60% after I stopped it for just two seconds

51:15Chrismaybe, maybe at that much, and then it's done. So after you release it, it takes on this system about one to two seconds. And then it gives you the whole everything you said all at once. It's pretty quick.

51:27WesYeah. So I think that's the sort of Rust part running. And then it calls over to the Python server to do the inference.

51:32ChrisI mean, I maybe. So, I mean, if you're being fair, at the very highest end, I paid a 20% CPU tax. But actually watching it, my CPU is bouncing around 17% without this running at the moment.

51:41BrentSo who knows?

51:44ChrisActually pretty darn impressive. That is pretty freaking impressive. Fast too it doesn't feel it doesn't feel like my system's heavy like i don't feel like i'm using a ton of system resources to have that run.

51:55WesIt seemed like you could just have it sitting there.

51:57ChrisYeah which is a real sweet spot.

51:59BrentDo you think it's smooth enough to you know move to a new web page go to a new entry for i don't know sending a message or recording something and just like boop, say it and then boop out because we've been doing like the slow experimenting version but how do you think that would translate to just everyday use.

52:16ChrisI think as long as you have the act as long as a text field is the active input box and it doesn't switch in the middle of dictation you're good to go.

52:22WesYeah i had to work with slack and open code i would say you might eventually want something bigger than whisper tiny i had a.

52:29ChrisFew things.

52:29WesLike it kind of struggled with mina itself a bit and it got my name is west at one point.

52:34ChrisSo the one nice thing about hex that i'd love to see and i don't have the gtk toolkit stuff to install to see if this is an option but with hex you can go in there and there's like a technical database and a common miss, transcribed words and you can you can put in there and say when you pick up this word insert this word instead and when you when it mistakes you know apt get for you know something else replace it with actual app dash get, and so i like that feature i don't know if this has it just because i don't have the gdk4 dependencies to show that but i think the performance of this

53:06Chrismight be better than hex having tried that recently.

53:09WesSo So there's some interesting technical... I mean, so there's like the style, push to talk versus, you know, toggle on to talk or however Brent said it. In a way... They take kind of very different approaches. So Mina is like the Linux desktop, you know, kind of makes sense because it's coming from Canonical. Like it plays inside like the sort of the proper way to do it. Like you would expect it to be done. It's in the input method framework, iBus, Dbus. So the sort of correct desktop citizen approach. Hex bypasses all of it.

53:39ChrisOf course.

53:40WesX test, synthetic key events, clipboard paste on X11. W type on WL roots, Wayland. Remote desktop portal on GNOME and KDE. So sort of cruder, but compositor agnostic.

53:52ChrisYeah.

53:52WesSo I think that makes it, you know, because it also supports Mac and Windows.

53:55ChrisYeah, that's, I guess, Hex's advantage there, but I don't care about that too much. I care about this. I care about it being a better Linux citizen.

54:04WesYeah, Hex might be a little farther along in some ways and a little more, you know, developed.

54:08ChrisEspecially with the annotation features. I think it kind of pulled ahead there, but who cares?

54:12WesAnd I think mine is maybe putting more thought into some of the privacy security architecture. Not that that's a problem in Hex, but just like they've taken this modular snap-based approach. They're really being careful about that kind of thing and integrating with these first-class protocols.

54:30ChrisAnd seemingly picking very good models. I don't know where Hex is on that, but I like the model flexibility with Manure, however you say it.

54:37WesYeah, they both seem to have pretty good model flexibility, which is great.

54:39ChrisYeah, that is true. That is true. Both of them seem to do. Better than I expected.

54:45WesYeah i got i was able to get it running uh yeah it wasn't packaged for nix but yeah just a rust and a python which are both pretty standard and then on kde you just register like a global shortcut and then you can tie it immediately into the rust side of things the rust client and then, away you go i think there is a bug on some older firefoxes that it can interact weirdly with like it just won't show up or i think some people had firefox crash but that was fixed in recent firefoxes so just be updated there which you probably want to do anyway.

55:14ChrisI feel like this is a workflow for you Brad.

55:17BrentI remember about a year and a half of go I after one of the shows said hey Chris I really, am trying to get working a way to just talk at my computer and I tried for a little bit and never got it there well they're doing the work for me so bring it on.

55:37ChrisYou're going to, if you try this, you're going to find the brain dump aspect of it so useful, especially if you're working with a clanker in an open code session, because they're really good at parsing a brain dump into structured information. Like they're really good at that. So you can just blah, and then it cleans it all up. Boom.

55:56BrentI think even for like an everyday task, like journaling or something like that, or capturing some kind of memory or a note, like where you just, you want it in text, but actually you just want a stream of consciousness. This is a perfect tool, especially if you're just sitting at your desk.

56:11ChrisPretty good. I think if this is the direction they're going, it bodes well. You'd really have to find something. You'd really have to dig to find something you don't like about what they're doing here. You really have to dig. And the fact that you don't even have to use snaps if you don't like snaps you can't even complain about that but if you do use a snap it's really easy to get going a gpl3 good work guys good work we're very impressed.

56:43BrentWe received some wonderful feedback recently and there's a bit of a theme here what are you building a little bit of a building report going on here and we've got a first note from pedro, Pager says, hey there, Linux team. Aw, Drew, you're included. After reading back in August about the teenager that got his HP printer working on macOS with CloudCode, I tried something similar with a BioKey side-touch fingerprint reader, which was Windows only, to get it working on Linux.

57:16BrentWorking with CloudCode and Codex, I've now built a native Linux driver that lets me enroll a fingerprint and authenticate through the normal Fedora and KDE desktop tools. It's still experimental. And I'm testing it on my own hardware.

57:31ChrisThat is so neat. I've been wondering if things like fingerprint readers, the other thing unlock would be.

57:36WesYou know, this one doesn't work.

57:38ChrisI know. You know what would be amazing is I have an LTE adapter in one of those Nix books.

57:42BrentOh, yeah.

57:43ChrisBut they don't support no Linux. I know. I love that.

57:46WesGo for them. Go for them.

57:47BrentWe're in like the golden age of fix the little thing that's been bothering me forever.

57:52ChrisGolden age, brother.

57:53WesPedro has a reality check for us.

57:55ChrisOh, no, really? Oh, it was going so well.

57:57BrentPage was continuing. Getting there was a much longer effort than I actually expected. Read, it was brutal. Plenty of false starts and dead ends. This isn't something someone could simply vibe code without a background in software architecture. It's given me enormous respect for the people who brought unsupported hardware to Linux, particularly before these LLMs.

58:19ChrisYeah, I know.

58:21BrentMissing features and incomplete hardware support are still barriers to adoption and I hope projects like this encourage others to help close these gaps. Happy to talk about what worked, where AI fell short and show the driver in use best.

58:36ChrisPedro Man, how great would that be to sit down over a meal with Pedro and chat about it?

58:42BrentThat's so fun You know.

58:43ChrisThat'd be a great convo.

58:44WesWhat can I need to find a meetup location?

58:46ChrisGood idea, Wes, we want to take Anton?

58:49WesYeah, Anton writes in, hey, Linux Unplugged Crew. I set up an Intel N150 mini PC as a home server with Jellyfin, Sonar, Radar, Browler, and a request app.

58:59ChrisOh, nice.

58:59WesTo simplify that stack, I built Torsure.

59:02ChrisWait, this is the developer of Torsure?

59:04WesYeah.

59:04BrentWow.

59:06WesOne Docker container with one database and one interface made for the TV first. Version 1.0 came out on October 26th.

59:12ChrisI've been watching. Well, my clanker found it the day you launched this and alerted me about this and said, this is right up your alley. And it was right.

59:19WesA few details. Playback starts while the file downloads, fetching a few minutes ahead and stopping when you close the player. How much have we been talking about that?

59:26ChrisLove that so much.

59:28WesTV interface for the browser engine of LG TVs from 2020 with LG WebOS app and Magic Remote Pointer. Library Jellyfin plaques Cody can read with hard links so kept files take no extra space. On-the-fly GPU transcoding when a TV can't play a file.

59:53ChrisSo for those that are not quite following, what this is solving is typically with the home media stack, if you really want the full automated, like backing up off the internet stack, you're looking at anywhere from three to five containers. It's a lot. This is doing it all in a single stack and providing the interface as well. Really impressive.

1:00:13WesTurning it into one focus product.

1:00:16ChrisIt was already on my, I should be looking at this and following it.

1:00:18WesThanks for writing in.

1:00:19ChrisYes.

1:00:19WesThat's awesome.

1:00:20ChrisYes. That's really great. Thank you, Anton. Appreciate that. We'll put a link to Torsher, T-O-R-S-H-E-R, in the show notes. Koya writes in, hi, Chris West and Brent. Your Oops All Shells episodes seem like a good reason to mention Koya Shell, a hyperlink shell built by my boss, Stefan. It brings desktop controls, launcher, and notifications together with a shared theme from login to lock screen underneath the JavaScript UI on Koya's own GPU render without a browser underneath.

1:00:51ChrisThat is nice. There's a live ISO to try it from. Great. You can try it out when it boots. The shell is CEC by four, so it's Creative Commons 4.0. Runtime is an open source, but it is free for non-commercial use. Would you be having a play? Interested to hear where it's among the shells you've discussed? Well, I would be interested in taking a look at this.

1:01:10WesYeah, I saw this float through. I just thought, like, maybe we can see what it looks like. I don't know.

1:01:14ChrisI have been pretty happy with the recent shell setup, but I think there's room for improvement in terms of efficiency.

1:01:19WesIt seems very clean.

1:01:21ChrisYeah. I do like the way it looks a lot.

1:01:24WesAnd it just seemed like worth seeing because there are clearly folks out there also experimenting with the shells.

1:01:30ChrisWe're building our own shells.

1:01:31WesIt's a whole new thing.

1:01:32ChrisIt's absolutely crazy. It's absolutely crazy, and I'm here for it. I'm here for that craziness. Thank you very much, Koya. Or not Koya. The project name is Koya. Thank you very much for sending that in. Gentlemen, we did get a couple of uplinks as well, and we have a few that I have to play really we fell behind because I meant to play some more in we're.

1:01:54WesNew at this.

1:01:55ChrisI know we're figuring it out so we'll let the agent pick one and we're going to clank away, All right. Up first is Matthew. Matthew is calling in about an agentic installer, I believe. Let's see.

1:02:09ClipsHey, guys. My name is Matt Mangano.

1:02:11ChrisHey, Matt.

1:02:11ClipsI'm an aerospace engineer by trade, and I work with AI workflows and helping, the company I work with work through that process. And I also am a big open cloth fan, try to contribute a little bit here and there. But I have just built a small little project where I just embedded an agent into a live ISO. So, you know, your girlfriend, your mom, your dad, a lot of people can't install

1:02:46ClipsLinux easily, especially Nix. So when you boot up the live ISO, you can sign in with your ChatGPT subscription or use an API. And basically just tell the agent what you want and the agent will install it. There's also some markdown files that, you know, kind of guide the agent to, make it a little bit easier for the user. So it's called...

1:03:16ClipsAgent-installer on my GitHub repo. So I'll try to send it to you guys. I don't know if I can paste it into this or not. Anyway, it's a great listen to you guys. Have a good one.

1:03:29ChrisThank you, sir. Brent, I want you to build that for me. I'm going to play one more. I'm going to play one more and let the bot pick again. All right. I'm not even going to see who this one is, but I think this is one of the queue we needed to get to.

1:03:42ClipsHey, guys. All right. 741 again.

1:03:44ChrisHey, it's Ari.

1:03:45ClipsOn the launch, you had mentioned that it seemed like your VR episode didn't get as much of a response as you were hoping. And I don't personally know how this will make you feel, but I think that that kind of was more of like a reference episode. And I think you've kind of had these types of reference episodes in the past, like the PCAE pass-through episode that you guys have done around being able to do PCAE pass-through with the Linux kernel. Um...

1:04:16ClipsWhere it might not get very much signal at that time. But over the years, I think people will keep referring back to that type of content, for help and assistance as they end up either acquiring a VR headset or either. Wanting to perform the thing that you're talking about in the reference episode. So personally, I think that makes it a really sticky topic, even though it might not get as much immediate response as you're expecting.

1:04:46ClipsSo I think long term that's really good for the show though but like I said I don't know how to make you feel. I actually ended up using the PCA pass through episode reference links whenever I was working on a project on my old job, that episode really gave me a lot more confidence in the technology and what I was doing because you all did your deep dive which was really helpful so I just wanted to say thank you for all the awesome content you guys have continued to make.

1:05:17ClipsI love being a member of the network and I've been really appreciative of all the new member features as well and even though the bootleg to me was enough in the first place I'm glad you guys have been able to find ways to make the membership more enticing, for new people. So thanks guys. See ya.

1:05:36ChrisThank you sir and thank you again for all your work you did on the website and in the community. It's good to hear from you.

1:05:41WesSupport all kinds of ways.

1:05:43ChrisAbsolutely. And I like the idea of a reference episode. It does make me feel slightly better. I will be honest. So I haven't put this all together yet, but we want more of your uplinks. We'll try to feature them from time to time. And I think there's going to be, Probably some sort of trifecta of HomeLab submission and maybe an uplink to go with that where you can reference your HomeLab submission and then possibly to put in front of us a boost, especially if you're a member, that could be a great idea because you get it for free anyways. And that kind of trifecta could be, you know, that could be a way to get your

1:06:16ChrisHomeLab a little extra juice with a little uplink and a boost to go with it. I'm just saying. I'm just saying something to think about as you're strategizing out there. Because you know the winners of last year's home lab special they're still they're still glowing they're still like they still you know and brent still have a.

1:06:33WesLot to glow about they were impressive.

1:06:35ChrisBrent still sending them their checks you know he's gonna get that yeah writing those checks so they're gonna get that any day behind.

1:06:40BrentBy one but we'll get.

1:06:42ChrisThere yeah yeah we'll get gentlemen we do have some picks and i think this one wes found just for brent This is a tab management extension for the Firefox called List All Tabs. And like tab management tools, it offers a lot of your standard features, but it brings in something kind of new to the mix that caught my attention, Wes. And that's this dashboard, if you will, of your open tabs.

1:07:11WesFormer dashboard guy. That makes sense.

1:07:12ChrisYeah, well, I'm kind of a dashboard guy, and I noticed that this kind of looks like a dashboard of your tabs, And I thought, you must be thinking of Brent when you found this.

1:07:20WesOh, yeah. Well, actually, someone sent it in for Brent. I've lost the name, unfortunately.

1:07:24ChrisIt's almost like he has a reputation. It's just...

1:07:33WesOh, Sudo Satur. That's who I found this from.

1:07:35ChrisThank you, Sudo. Thank you. Appreciate that. List all tabs. This actually is really nice. What do you think of, as a tab connoisseur, Brent, what do you think of something like this?

1:07:43BrentWell, Sudo sent this in and I was a little hesitant because, you know, tabs and precious and all that stuff.

1:07:49ChrisYeah, obviously.

1:07:49WesYou're very special tabs.

1:07:52BrentI don't want to mess with my tabs. But Wes was like, come on, just try it. And so I did.

1:07:58WesWell, you're the QA guy. I send all the new stuff to you.

1:08:01BrentYeah, it turns out this thing's fantastic.

1:08:04ChrisOh ladies and gentlemen ladies and gentlemen oh.

1:08:10BrentIt was super easy to install in Firefox because it's already set up as an extension that you can add. It's got raving reviews already. And also, I guess it's based on an older project, but there's been some modernization. And the interface on here, just to manage your tabs, is really nice. It's got an important thing right at the top. Somehow, that's exactly what I need. You can sort by like tab order or last recently used or least recently used,

1:08:44Brentmost recently used, which is that's just how did you get in my brain to design this thing?

1:08:49ChrisThat is really clever. Yeah. Okay.

1:08:51BrentIt also lists, it can list all of your tabs by URL instead of by window, for instance. So if you have a, let's say a bunch of GitHub links, but open in various windows and tabs.

1:09:06ChrisOr for me, like a bunch of Wikipedia tabs, you know, the Wikipedia rabbit hole thing. Yeah, definitely.

1:09:11BrentCombines them all in one place. So you can reference them all under one little section in the extension. It's fantastic.

1:09:17WesSee, tabs is data. This is what I'm saying.

1:09:19ChrisYou were right.

1:09:20WesThis looks great.

1:09:21BrentBut I'm not done yet. It also has a little, you know, sleepy button, which unloads all of the tabs in that particular section. So for instance, I have github.com here and it's got a little Z beside it. So you hit that and they all get just... Automatically put to sleep instead of the browser trying to guess because like i know what i'm not using and i could just hit a nice little button here it's fantastic.

1:09:45ChrisI'll be dipped that's.

1:09:46BrentGreat right uh there's also an unload all button because if you have too many tabs then you need that which so that speaks to me, um but you can also look at them like listed by window which of course is handy, and there's a saved section so you can also save particular tabs into like a spot where for some reason, these are more important, which is awesome.

1:10:10ChrisI'm installing this right now.

1:10:11BrentOh, I'm telling you, it's great. And I could keep going. Like it has a recently closed part so that in case, you know, you did the wrong thing and you want to bring it back, you can do that. It also has a little trash icon so you can quickly delete tabs. It also has a save, you know, save button beside all the tabs so you can put them in that super great save section.

1:10:28ChrisOh, all right.

1:10:29BrentBut best part is it also detects duplicates so that you're not, you know, saving the same page four times.

1:10:36ChrisYeah, that's not a problem I have, but I could probably understand why you would have that problem.

1:10:39WesIt seems handy.

1:10:40BrentAlso, these are just the features I've found so far. I've only used it for like 12 minutes. So really huge thank you, SayShirt. I will, well, I'm already using this. So great. It works for my brain. I hope it works for other people too.

1:10:53ChrisOkay. All right. Our next pick is to reduce your browser dependence. And I understand it's hard. I get it. There's a lot of good web apps or bad web apps that you have to use every single day. I'm there with you. So what if I could get you to try out Tangram?

1:11:11WesTangram.

1:11:12BrentTangram. Yeah.

1:11:14ChrisYeah. This is another take at the attempt to make one browser for all of the web apps you have to use, where you can kind of have separate, isolated environments, and they are each one individually sandboxed, persistent, independent tabs. You can change their position. It supports smart notifications for desktop integration. So, like, if you've got to be in, like, a work messaging application. That's nice. It's got that. It's got touchpad and touchscreen support.

1:11:46ChrisAnd the idea is that you put your Discord in here, your web telegram, your WhatsApp, if you use Element in the web, whatever it might be, your Mastodon. You put it all in here as individual isolated tabs that this thing then keeps persistent and independently manages them for you. And then gives you an interface to see an overview of those or pop through the different tabs.

1:12:10WesA browser for your pin tabs. I like it.

1:12:13BrentI have a question.

1:12:14ChrisYeah.

1:12:15BrentOh, that little tagline, browser for your pin tabs. I don't really know what a pin tab is. Can you... Do you think this will maybe nudge me into the pin tabs era?

1:12:31ChrisI mean, you could think of it as another way to have a bunch of tabs.

1:12:35WesWell, I was thinking maybe the ones that work for you in here is how you learn what a pin tab is. Like if it works for you in Tangram, then it would have been a pin tab if you ever had learned those.

1:12:44ChrisBut GPL3 and available on FlatHub.

1:12:46WesSo pretty convenient, what's your excuse?

1:12:48ChrisOkay. All right. Now this next one, we're getting a little serious here. Your boy here has been using Pinch Flat. as a way to download YouTube videos offline and then present them as first-class citizens in Jellyfinnerplex. And when I say first-class, what I mean by that is they look like every other TV episode or movie with a poster, metadata, description, even skip intro,

1:13:19Chrisskip commercial, all this.

1:13:20WesIt feels native.

1:13:21ChrisFeels like a native thing. And it's a fantastic way for me to watch YouTube because I kind of got sketchy internet at home from time to time. Sometimes we're on the go and we're on LTE. And so watching YouTube is a chore on those kind of connections. So if you just got things you watch, a few channels you watch on the regular, you throw them into old pinch flat here and it pulls them down, saves them, gets all the NFO and metadata information, saves which ones it's viewed, all that. Beautiful. Unfortunately, YouTube's been making it harder and harder for good old classic YTDLP to work properly.

1:13:54ChrisAnd this has impacted the performance of Pinchflat and related projects. And so more and more, I'm missing videos. And so Pinchflat NGX comes into the picture here. Pinchflat NGX. An enhanced self-hostable version of Pinchflat with a YouTube media manager and metadata built in here. And one of the things it does do differently, besides some UI changes that I think look quite nice, is it uses a different download pipeline. And it uses a tool that I'm not super familiar with, but seems to be producing

1:14:25Chrisbetter results, and they kind of go through all of it in their description of how it all works. And it uses something called bgutil, which integrates with like a token you get from YouTube or something like that, and bypasses some of the issues that YTDLP runs into. Really nice. And it's one of these things that once you set it up, not only do you generally not have to think about it as long as it doesn't break, but it makes the experience of watching YouTube not feel like YouTube, right?

1:14:57ChrisIt just, there's a couple of channels I want to watch and I don't want the rest of that crap. And because it does also integrate things like skip intro or things like that, it feels more like an actual TV series that I watch.

1:15:11WesUh it's you know because i think the original development of pinflat kind of stopped right seemingly um so this is a continuation in a few ways including i mean like original sqlite only this one adds postgres support yes uh it adds oauth and oidc support yeah, before you kind of got a basic worker pool now it has live diagnostics granular worker tuning it also has better support for single video sources one-off direct video downloads if you want to do that too which is pretty convenient so big upgrade.

1:15:41ChrisAnd then you know you just plumb it in you have it saved to a directory that you have jellyfin or plex index you add that as a library so i just have a youtube library and if you were to browse it you'd be like this is great, um but with pinch flat having some issues and kind of not getting updated much it was really great to see pinch flat ngx come along and it's also gpl3 as well, Big recommend, and especially if you need to archive some videos, like if there's a channel that has really helpful instruction videos that you want to archive offline.

1:16:11ChrisThe other reason I started using this is there's some videos that I don't mind at all my kids watch. There's other videos, not a big fan. And so this is a way to kind of get access for the family to YouTube in a way that doesn't leak into the other videos that maybe you're worried about. So it's nice from a parental standpoint too. So it's Pinch Flat NGX.

1:16:29Wes90% Elixir.

1:16:30ChrisNo, that's cool.

1:16:32WesWe need a sound for that one.

1:16:33ChrisSomething cool, like an Excalibur kind of sound or something. But we will have links to all of this in the show notes, Wes. And where do people find those show notes? Well, that's on our website at linuxunplugged.com. And this is episode 692. Or you can find it over at jupiterbroadcasting.com. Either website, you'll have our RSS feed. That's the pro move. Grab that feed, put it in your podcast catcher of choice. Get the episode every week, obviously. But take a look at it, too. There's some useful stuff in there.

1:16:59WesMm-hmm. Cloud chapters. Yeah.

1:17:01ChrisJson, json json json.

1:17:06WesWe also have transcripts vtt and srt srt and you put something special in.

1:17:11ChrisThere's an mp4 in there too i don't know why but there is don't do it.

1:17:15WesDon't even look at it.

1:17:16ChrisDon't don't do it don't do it don't do it although brent's so damn handsome.

1:17:20WesSometimes there's cats.

1:17:21ChrisYeah and dogs there's that too all right that's everything for us guess what and we will be back live at a regular time back from texas linux fest and i'm sure with stories to share, So, why don't you head on over to jblive.tv and make it a Tuesday on a Sunday. We get together over at jblive.tv and jblive.fm if you're on the go at 10 a.m. Pacific. Convert it at your local time at jupiterbroadcasting.com slash calendars. Don't forget our members get the full bootleg, and it's usually a whopper.

1:17:54WesAnd don't forget to get your home labs decorated.

1:17:57ChrisYes.

1:17:57WesPrepare for submission.

1:17:58ChrisThat's right. Coming soon, we'll have more information, and I'll be on the website as well. Thank you so much for joining us on this week's episode of Your Unplugged Program. Thank you to John for joining us. Good luck to all of their endeavors. And with that, we will see you right back here next Tuesday, as in Sunday.

Open the transcript file

Hosts and Guests

Hosts

Guests

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. 693: Two Cities, One Linux

    1 hr 41 minVideo

    From the hallway track at Texas Linux Fest to the builders at Ubuntu Summit in London, we bring you the people and projects shaping our Linux community. Sponsored By: Managed Nebula: Meet Managed Nebula from Defined...

    Listen to the audio

    Credit the people on this episode

  2. 691: Fedora's Hidden Upgrades

    1 hr 21 minVideo

    Fedora 45 has serious upgrades under the hood; we dig in, and may have found our new favorite dashboard. Sponsored By: Managed Nebula: Meet Managed Nebula from Defined Networking. A decentralized VPN built on the...

    Listen to the audio

    Credit the people on this episode

  3. 690: The S in PHP is for Security

    1 hr 12 minVideo

    A major PHP flaw puts servers everywhere at risk; so we hack ours. Meanwhile, open source helps debloat our smart TVs. Sponsored By: Managed Nebula: Meet Managed Nebula from Defined Networking. A decentralized VPN built...

    Listen to the audio

    Credit the people on this episode

  4. 689: First Kontact

    1 hr 19 minVideo

    A new documentary reveals KDE's 30-year underdog story. Plasma pioneer Aaron Seigo joins us to look back, and explore what's next with Plasma 6.8. Sponsored By: Jupiter Party Annual Membership: Put your support on...

    Listen to the audio

    Credit the people on this episode

  5. 688: Eyes on Deck

    1 hr 21 minVideo

    We've turned our Steam Deck into a security monitoring system; Monitor your property, identify people, pets, and vehicles with detailed daily intelligence briefings. All local. Sponsored By: Jupiter Party Annual...

    Listen to the audio

    Credit the people on this episode