Untitled Linux Show 263 Transcript

Please be advised that this transcript is AI-generated and may not be word-for-word. Time codes refer to the approximate times in the ad-free version of the show.
 

Jonathan Bennett [00:00:00]:
This week we're talking about AI in the kernel and a Blender release and the possibility of X going open source. COSMIC has a new big release with Frosted Glass. Flathub defends its ban on AI, and I have a story from Trench on that one and a lot more. You don't want to miss it. So stay tuned.

Rob Campbell [00:00:23]:
Podcasts you love. From people you trust. This is TWiT.

Jonathan Bennett [00:00:32]:
This is The Untitled Linux Show, episode 263, recorded Saturday, July 18th. Getting back with my ex.

Rob Campbell [00:00:39]:
Hello everybody, it's Saturday. And you know what time that means it is. Why am I here? Because Jonathan's having some technical difficulties. He is here, but I'm just opening the show for him to get us started. So here we are in The Untitled Linux Show. Let's, We've got Ken and Jonathan here with us.

Jonathan Bennett [00:01:00]:
Yeah, I am on location at Open Sauce. I am here repping the Meshtastic booth, but I didn't wanna miss ULS. I'm having some technical issues with my Framework laptop talking to my Bluetooth headset. Who would've thought audio, Bluetooth audio on Linux is a problem? I've never heard of that. Why? It's always been perfect. But anyway, I'm here, but I may be mostly color commentary for the show, but I didn't wanna miss it.

Rob Campbell [00:01:28]:
And neither will you.

Ken McDonald [00:01:28]:
And I'm here. I'm always color commentary for the show.

Jonathan Bennett [00:01:36]:
Yeah.

Rob Campbell [00:01:36]:
So to start us off, I guess I have the first story and you've all heard the saying that drove many Linux users crazy. And that one is Microsoft loves Linux. Well, now it's time for a new one. And that one, maybe I'll drive some of you crazier, is Linux loves AI. Okay. And I know, I know that one's going to upset many of the, many of you out there in the Linux community. AI has become one of those topics where everyone has a strong opinion one way or the other. And to many in the Linux community, it's negative.

Ken McDonald [00:02:17]:
Yeah.

Rob Campbell [00:02:17]:
People say it's unethical. It's hype. It writes bad code. Keep it out of the open source. And, you know, even this week I heard it compared to the way a compiler can inject malware into a program, saying that AI can also inject malware into code, apparently hidden so that nobody can see it upon review. I don't know about that, but maybe, maybe not. Anyway, this last one, this last comment I heard was actually in response to this week's story about Linus, Linus Torvalds, that is the great Linus Torvalds and his current position on AI. So Linus Torvalds was, has always, was always, wasn't, was not always an AI cheerleader.

Rob Campbell [00:03:09]:
In fact, back in 2024, he described the industry as roughly 90% marketing and 10% reality. But instead of permanently planting his flag, Linus tried the technology and his opinion has evolved. Sounds a lot like somebody else we know on the show here for those following along. Now, no, no, no, he's not vibe coding the Linux kernel.

Ken McDonald [00:03:38]:
Yet.

Rob Campbell [00:03:39]:
The kernel has guidance for AI-assisted contributions. Code must still meet licensing requirements, and AI cannot sign off on a patch, and a human developer remains responsible for what gets submitted. But Linus has vibecoded an application before. His AudioNoise side project experiments with digital audio effects. Linus wrote much of it himself, but says its Python visualizer was basically written by vibecoding.

Ken McDonald [00:04:07]:
Wow.

Rob Campbell [00:04:09]:
coding, you know, with Google's anti-gravity. That's not one I'm really familiar with, but he joked that rather than Googling examples and copying them, you know, he just cut out the middleman himself and let AI build the visualizer for him. And now Linus has gone further responding to anti-AI sentiment around kernel development. He said Linux is not an anti-AI project. He called AI a useful tool, said nobody is being forced to use it, and told those who cannot accept it, you know, others using it in open source, that open source gives them a choice. You know, fork the project or walk away. That does not give AI a free pass. Bad patches, hallucinated bug reports, licensing questions, and extra work for maintainers, you know, are real problems.

Rob Campbell [00:05:10]:
And Linus himself has criticized low-value AI-assisted submissions. But those are problems to manage, not reasons to pretend the technology does not exist. As I have said before, Linux has always been about taking useful tools, improving them, and placing responsibility on the person using them. Maybe Linux loves AI is a little bit of an exaggeration, but Linux clearly is not afraid of AI. And according to Linus, it will not become an anti-AI project just to justify the loudness, loudest people in the room. So I'm definitely glad to hear this. Based on earlier stories this year about being overwhelmed with AI bugs, it was feeling rather anti-AI to me. So I kind of feared that the Linux project And I made it known on the show, kind of, I think, at least implied it, you know, that the Linux project was going all in on this anti-AI direction.

Rob Campbell [00:06:14]:
Thankfully for Linux, I was wrong.

Jonathan Bennett [00:06:19]:
Yeah, I saw an interesting take on this that compared it to the move from like assembly language and raw C to some of the more modern languages. And I thought that was an interesting take on it, that it's considered to be less of a revolutionary, more of an evolutionary change. That may be a fair way to look at it. It's always interesting to think about these things, these big changes that we go through, that we get to see firsthand. And in 100 years from now, how will history treat it? Or 1,000 years from now? I thought the internet was going to be the big change and that it changed everything. And, you know, nowadays AI has come along and it's like, well, maybe that, maybe the two of them will be talked about in the same breath or considered the same thing. Definitely, definitely changed the way we all work when we write code.

Rob Campbell [00:07:10]:
Yeah, in the long run, they're kind of close together, you know. I mean, in our lifetimes, they're not that close, but when you're looking out 100 years, they almost pop up at the same time.

Ken McDonald [00:07:20]:
I saw how much computers changed everything.

Rob Campbell [00:07:23]:
Yeah, you did.

Jonathan Bennett [00:07:27]:
Yep, absolutely.

Ken McDonald [00:07:29]:
I'm sure my grandparents saw how much phones changed everything.

Rob Campbell [00:07:36]:
Yeah.

Ken McDonald [00:07:36]:
And electricity.

Jonathan Bennett [00:07:38]:
I mean, there's big changes in every generation, right? But then there's some changes that you look back on, like the movable type printing press, right? That was such a revolutionary change. It changed everything. It changed the entire world. It's, you know, when you're in the midst of it, like I said, it's hard to tell exactly where you're at in the timeline and how history will remember it. Well, all right, we are going to take—

Ken McDonald [00:07:59]:
Scary is how many different changes I've actually seen in my lifetime. Computer, internet. I think Ken just called himself old. Yeah.

Jonathan Bennett [00:08:12]:
Yeah, absolutely. You know, I think it really boils down to, you know, the people at the kernel are pragmatists. They care about the code and they care about the code being right. And they see this as another tool to try to achieve that. And, you know, I can't blame them. I've kind of been treated basically the same.

Rob Campbell [00:08:27]:
Glad for that.

Jonathan Bennett [00:08:28]:
Yeah, for sure. I also think that, like, I know there's questions about, like, the legal issues around open source and the ethical issues, but I had a lawyer on Foss Weekly, not Tris, but before that, several months ago. And he made the point that the genie's out of the bottle now, and you're never, ever gonna be able to put it back. And that's kind of become my approach to this. Like, even if, in theory, there are legal problems, The genie is never going back in the bottle. All right, let's take a quick break. I know we've got other stuff coming up. In fact, Ken is going to be up next with news of some release of Blender.

Jonathan Bennett [00:09:04]:
We'll get to that right after this.

Ken McDonald [00:09:08]:
And Jonathan, this week, as you were saying, we've got Michael Larabelle and Markus Nister writing about an update to my favorite open source, free and cross-platform 3D computer graphics software suite. Yes, I am talking about Blender. This time it's version 5.2, and it's a long-term support version. It now has, uh, the Circle, Space Evenly, and Flatten Loop tools built in instead of installing as extensions as we've done in the past. Blender 5.2 also provides a big update to the Geometry node. Including a new physics system that focuses on hair and cloth. A new sound socket type has been added that uses an imported sound to output the amplitude as a float. Blender 5.2 also received some long-awaited features in Draw Mode to include updating thumbnails for brushes.

Ken McDonald [00:10:13]:
And having a texture brush pack available after enabling online access. Now, the rigging and animation updates in this release provide some nice quality of life improvements to include more options when spawning bones and a new head or teller slider for adjusting positioning along a bone when using bone Parenting. Blender 5.2 adds a new allow pre-roll option in the playback popover, allowing Blender to play back frames located before the defined playback start frame. Now, according to Marcus, Blender 5.2 LTS improves memory use and faster startup time on texture-heavy scenes. Add support for snapping to the lattice objects. Support for interpolating vertex custom data when merging vertices— vertices. Well, I don't know if that's BSOF or long I.

Rob Campbell [00:11:20]:
Vertices.

Ken McDonald [00:11:20]:
And support for HDR and wide gamut color spaces to scopes. Now you can find more details in Michael and Markus's article. Or you can watch the 36-minute official overview video that I also have linked in the show notes.

Rob Campbell [00:11:42]:
Which feature are you most excited to try first, Ken?

Ken McDonald [00:11:47]:
It's a tough decision there. There's some that I didn't discuss that I really am looking forward to playing with. But you have to find out about those when you read those articles.

Jonathan Bennett [00:11:58]:
They've got a, uh, a release page here with, uh, multiple like showcase reel and recap video and all kinds of fun just showing off some of the different things they can do. Um, would be really interesting to, uh, look through all of those. Looks really cool. Yep.

Ken McDonald [00:12:14]:
I still got to finish watching the, uh, official overview that they put together. Yeah, I just—

Jonathan Bennett [00:12:20]:
I didn't realize it was that long. That's quite the, uh, It's quite the overview.

Ken McDonald [00:12:25]:
Aren't you glad I didn't go that long into describing it?

Jonathan Bennett [00:12:30]:
Yeah.

Ken McDonald [00:12:30]:
Uh, I suppose that'd have been half the show there.

Rob Campbell [00:12:33]:
Yeah, Blender's not anything I've ever, uh, really played with though.

Jonathan Bennett [00:12:37]:
Definitely. I have not either. I, I certainly want to. I, I like the idea of it. I've just, I've never found the time. Um, I wonder how many people are now starting to do things like, uh, you know, use the AI to create Blender models and Blender scenes. It's got to be a thing people are playing around with. I know, in fact, here at Open Sauce, there is at least one booth where people are using AI to make circuits.

Jonathan Bennett [00:12:58]:
And I was sort of surprised that that was working as well as it was, but he had a physical device. He's like, I had the robot design this thing for me, and then I had it made and it works. So it's taken over all kinds of places.

Rob Campbell [00:13:11]:
Yeah, one of the things I was thinking about is, you know, the time limitation. And first you gotta learn and figure out what you're doing. And, you know, I almost thought that Maybe using AI now might be a way to jump into it fast, just to learn, you know, have it guide you and help you.

Ken McDonald [00:13:28]:
Right.

Jonathan Bennett [00:13:28]:
Like, hey, this is the thing I want to do. Give me a step-by-step guide to do it. Yeah, I think that's a really interesting use for sure.

Ken McDonald [00:13:35]:
Yeah.

Jonathan Bennett [00:13:35]:
All right.

Ken McDonald [00:13:36]:
I've done some playing around with it, basically just following some of the tutorials, one for creating Gus, another one for creating a castle. A shark modeling and a Yellow Submarine.

Jonathan Bennett [00:13:55]:
Do we all live there?

Ken McDonald [00:13:58]:
I think that's what the Beatles said.

Jonathan Bennett [00:14:00]:
Yeah, that's what they said. That's cool. Yeah, Blender is a real open source success story. You know, there's another potentially growing open source success story. Is that the way we want to pitch this? There's a company out there that's talking about open sourcing a lot of things, at least. And to their credit, they have open sourced some of these things. I'm not going to give the punchline away. I'm going to let Rob talk about it.

Jonathan Bennett [00:14:25]:
Who is this that's out there open sourcing stuff?

Rob Campbell [00:14:29]:
Elon Musk. You know, a lot of the other tech shows talk a lot about Elon Musk. I don't. We tend not to too much, but here we are. We got a story with him. So Elon Musk says that Twitter, Or X, I guess. X is going to open source X. And this is one of the stories, you know, I'll believe it when I see it on the GitHub repository, even though, yeah, I guess, you know, they have open sourced some things.

Rob Campbell [00:15:00]:
But according to Musk, once the company finishes reviewing its software for security vulnerabilities, it will release the platform's entire codebase with no exceptions. He also says third-party reviewers will be invited to compare the published code against the software actually running in production. You know, companies sometimes release a carefully selected portion of their software, call it open source, and leave out the pieces people are most interested in examining. I'm pretty sure Microsoft has done some of that. Anyway, Allowing independent reviewers to verify that public code matches the, uh, that matches the production platform would provide a much stronger level of transparency. Icolyte researchers examine how content is ranked, how moderation decisions are implemented, how advertisements are selected, and whether certain accounts or opinions receive preferential treatment. Or that they don't. But again, this is currently a promise, not a release.

Rob Campbell [00:16:07]:
Musk has made open source promises before with mixed results. You know, Twitter released part of its recommendation system back in 2023. Maybe it was X then, whatever. But that repository received relatively few updates and was never a complete, easily runnable copy of the platform. There have also been some legitimate successes. X published a newer version of its For You recommendation algorithm in January 2026 and delivered a significant update in May, including a runnable inference pipeline. xAI also released the Groq-1 model weights and architecture under the Apache 2.0 license in 2024. On the other hand, Musk said Grok 3 would become open source in about 6 months after Grok 2.5 was released.

Rob Campbell [00:17:09]:
That deadline has passed around back in February 2026, and Grok 3 still has not appeared. There's also no public date for this complete X release. You know, after we finish the security review could mean next month, next year. Or never, you know, maybe they'll never finish it. And then when the code appears, we need to examine the license. You know, posting source code where people can read it is not necessarily the same as releasing genuinely open source software that people can modify, redistribute, redistribute, and build upon. Open sourcing X would not automatically fix the platform, restore trust, or reverse every questionable decision ever made since Musk bought Twitter. But It would allow outside developers and security researchers to replace speculation with evidence.

Rob Campbell [00:18:04]:
But my big question is, why? Why would he spend billions of dollars just to give it all away? Well, you know, I have a few ideas. You know, one, he wouldn't actually be giving away the service, just the code they use to provide it, which is a much different story. You know, code doesn't give you followers as, uh, as discovered on many other platforms that have come out since then. Um, though I have quite a few followers comparatively on other accounts. Um, you know, speaking of other accounts that, you know, didn't really become much, anyone remember Threads? I still see ads for it, so I guess it's still out there, but I don't log into it anyway. Musk is always trying to cut costs, especially, you know, with this money pit, uh, called X. And, and if the code goes open source, you know, he could in effect gain thousands of free developers, uh, that contribute to the project. And, you know, maybe he could let go of some of, uh, the ones he's paying today.

Rob Campbell [00:19:10]:
Or, you know, if not that, you know, maybe, maybe he, he could at least have a better Twitter, X. Anyway, anyway, it would be a cool step, uh, and, and maybe, just maybe, if Musk releases the entire thing, uses a real open source license, and proves that it matches production, maybe I'll finally stop calling it Twitter. Maybe it might even earn, uh, X back a tiny little bit of my respect.

Jonathan Bennett [00:19:47]:
Rob is getting back.

Ken McDonald [00:19:51]:
Well, Rob, um, got— that's— I thought, uh, Musk bought that as a tax write-off.

Jonathan Bennett [00:20:03]:
No, getting, getting back with your ex, Rob.

Rob Campbell [00:20:07]:
Getting back with my ex.

Jonathan Bennett [00:20:10]:
There's your show title right there.

Ken McDonald [00:20:12]:
Yeah.

Jonathan Bennett [00:20:14]:
So I went and looked at the Twitter/X GitHub repository and there are like 102 repositories there. Like they have pushed a bunch of code out. Unfortunately, as you noticed, most of it is sort of dead, like last updated in 2023 or 2024. There's a couple of them that are kept up to date. Personally, I think, I think Musk likes the idea of making things open source, but I think he just like doesn't have the bandwidth to stay on top of it. to continue to push people at the companies to keep up with it. So yeah, I don't know. I would love to see Twitter be like a better open source steward and part of the community, grow an actual community around their code.

Jonathan Bennett [00:20:54]:
We'll see if it happens.

Ken McDonald [00:20:55]:
Isn't he trying to get people to volunteer to help improve it for him?

Jonathan Bennett [00:21:00]:
I mean, when you look at the repositories, that's not what they're doing. I don't think anybody's doing pull requests. I don't think they're pulling any of that code back. Honestly, I think it's more that Musk likes the idea of things being transparent and people being able to look into it. And this is a way for him to try to accomplish that.

Rob Campbell [00:21:19]:
I think getting people to help would be a great use case though. I'm sure.

Jonathan Bennett [00:21:28]:
Yeah. I mean, that's part of what makes open source so cool is that people get to help and fix bugs that annoy them personally. You build a community around it. Yeah, absolutely. I would, I would love to see that with Twitter. I think that would be really cool.

Ken McDonald [00:21:41]:
And it definitely gives us the option to fork some of that and go off in another direction from what Musk has been doing.

Rob Campbell [00:21:50]:
You know, you say that, but you know, you're talking about Twitter. There's already clones that are like so close to it. I mean, what's missing is maybe the algorithm used for Stuff, but I mean, what more would they really want in say Mastodon or Bluesky or—

Jonathan Bennett [00:22:08]:
Yeah. I mean, so much of Twitter is the network effect. I, while yes, you could theoretically fork it, like, so, you know, you can imagine the scenario where they release enough of the source code that you could fork it and run your own Twitter, but like, what good is it to have a Twitter when it's just you?

Rob Campbell [00:22:23]:
Yeah.

Jonathan Bennett [00:22:23]:
It's, it's the network effect is what makes it not, not the same thing.

Rob Campbell [00:22:26]:
It's like if Walmart open-sourced the design of their building. All right, everyone else takes the design of their building, but—

Jonathan Bennett [00:22:33]:
That's a great comparison.

Ken McDonald [00:22:36]:
I think several companies have done that.

Rob Campbell [00:22:39]:
Well, yeah, because it's easy to clone something like that, just like it's not too hard to clone something like Twitter.

Ken McDonald [00:22:46]:
Would this having the source code open give those clones the capability of Interfacing with Twitter for that transfer, communicating between the clone and—

Rob Campbell [00:23:01]:
They allowed it.

Jonathan Bennett [00:23:04]:
But yeah, I had a, I had a pipe dream for the longest time that Twitter would introduce the ability to connect to the Fediverse via Mastodon. But I don't think anybody else thought that was a good idea. I was sort of alone on that island.

Ken McDonald [00:23:17]:
Oh well. And they asked you what you had been smoking?

Jonathan Bennett [00:23:21]:
Yeah, something like that.

Rob Campbell [00:23:22]:
Isn't that the hopes or what they say Bluesky is supposed to become? Isn't there something like that?

Jonathan Bennett [00:23:28]:
I don't know. I've not actually looked too deeply into Bluesky yet.

Ken McDonald [00:23:31]:
I think that's the hope for Bluesky, Mastodon.

Rob Campbell [00:23:35]:
I thought—

Ken McDonald [00:23:36]:
What's the one that Facebook came out? Oh, Thread?

Rob Campbell [00:23:38]:
Well, I thought, who is the guy who came out with Bluesky? Which is this—

Jonathan Bennett [00:23:45]:
Was it Jack Dorsey?

Rob Campbell [00:23:46]:
Jack Dorsey, the same guy who started Twitter. I thought that was like something he actually said. Maybe I'm hallucinating. Maybe I'm AI.

Jonathan Bennett [00:23:56]:
Maybe Elon Musk is not the only CEO that makes these pronouncements and then figures out later it was a bad idea and walks it back.

Rob Campbell [00:24:03]:
Maybe.

Jonathan Bennett [00:24:04]:
All right, let's take a quick break and then we're gonna talk about an up-and-coming desktop, something everybody here is excited about. We'll be right back after this.

Ken McDonald [00:24:16]:
And Jonathan, Rob, this week, Bobby Borisov, Michael Larabelle, and Marcus Nester. We have 3 writers writing about Linux hardware vendor System76 releasing the latest stable version of their Rust-based desktop environment for Pop!_OS. And as we all remember, it's called COSMIC. This time it's version 1.3. Now Bobby, Michael, and Marcus all talk about the long-awaited frosted glass effect arriving. So here we're trying to blur some lines, I think. Now according to Bobby, the feature brings a soft blurred transparency to COSMIC's windows and interface components. Instead of making surfaces transparent and letting the wallpaper distractingly show through, It diffuses the background behind them, creating a polished layered appearance while keeping text and controls easy to read.

Ken McDonald [00:25:18]:
According to Michael, Cosmic Epoch 1.3 also brings translation updates and other minor refinements. And according to Marcus, this release improves fractional mouse wheel scrolling in the Cosmic Term Terminal emulator improves frame pacing of embedded applet popups and adds an option to preserve panel or dock style on window maximize in COSMIC Panel. COSMIC 1.3 also updates the new COSMIC Monitor app with support for collecting GPU memory usage for Intel GPUs. The ability to show GPU power and total VRAM in the dashboard. The ability to suspend NVIDIA GPUs and support for calculating power usage for AMD or Intel GPUs. As always, you can get more details by reading the articles by Bobby, Michael, and Marcus. You'll find the links, of course, in our show notes. So Rob, when's the last time you tried Cosmic?

Rob Campbell [00:26:36]:
Oh, I, I use it, uh, regularly.

Jonathan Bennett [00:26:39]:
I do. Are you running Cosmic now? That's cool.

Rob Campbell [00:26:42]:
Yeah, I, I need to upgrade to it though.

Jonathan Bennett [00:26:45]:
Yeah, I, I saw the, uh, people talking about the frosted glass and, uh, a few, a few comments people made about it was one, it sort of resembles another Glass-based interface that someone else did very recently. And then in the Phoronix comments, there are people talking about, you know, it reminds me a lot of Aero on Windows 7. It's kind of a throwback. This is the thing that people have done over the years with desktops, like, oh, we have GPU power, let's do translucency effects. Which, I mean, it's cool, it's fun to look at for a while. You maybe eventually get tired of it. It depends upon like how well it's done. I need to actually pull up a COSMIC install somewhere and take a look at it.

Ken McDonald [00:27:27]:
I'm sure there's somebody where you're at that might be running it.

Jonathan Bennett [00:27:31]:
There might be. I mean, I've seen a lot of Framework laptops around and a lot of people nod at me when I talk about Linux. I've seen people wearing hats with Tux on them and all kinds of fun stuff. Yeah, a lot of Linux nerds here. It's pretty cool.

Rob Campbell [00:27:45]:
Yeah.

Jonathan Bennett [00:27:47]:
What do you guys think? Is COSMIC going to be the new GNOME? I mean that in the best possible way, not the worst possible way.

Rob Campbell [00:27:53]:
Like to replace it and be the better GNOME.

Jonathan Bennett [00:27:55]:
Yeah, yeah. Yeah, and not be a dumpster fire the way GNOME seems to be these days.

Rob Campbell [00:27:59]:
It's very GNOME-like.

Jonathan Bennett [00:28:02]:
Well, wouldn't it? I think it's possible.

Ken McDonald [00:28:06]:
Wasn't GNOME the inspiration for COSMIC?

Jonathan Bennett [00:28:10]:
Oh, absolutely. I mean, COSMIC was at one point just a theme on top of GNOME. And then GNOME sort of decided they wanted to change some things to make the life of the System76 engineers harder doing the COSMIC theming. So they did it not because it was easy, but because they thought it was easy. They decided to roll their own desktop environment.

Rob Campbell [00:28:29]:
You know, I don't think GNOME's going any way. I mean, there's plenty of niche desktop environments out there, but, you know, maybe if COSMIC, the real success of COSMIC, would be if it replaced—

Ken McDonald [00:28:43]:
KDE?

Rob Campbell [00:28:44]:
No, no, no. GNOME. Because like a lot of distros have GNOME as their default desktop. So if COSMIC became a default.

Jonathan Bennett [00:28:55]:
I would love to see COSMIC shipped as the default in certain places. Or maybe KDE is the default and COSMIC is the second option. And then, yeah, if you really want to, you can install GNOME. I'd be okay with that.

Ken McDonald [00:29:05]:
If you want to really slow down your system, install GNOME.

Rob Campbell [00:29:09]:
Oh, yeah.

Jonathan Bennett [00:29:11]:
Yeah, if you want it to look exactly like the GNOME developers think it should, install GNOME. All right. So the Flathub ban on AI, I have some notes on this from the trenches. We're going to let Rob tell the story, and then I will tell you my anecdote because we, My project just recently shipped a Flatpak. So Rob, take it away, and then I'll jump back in and we'll talk about the details of how this actually went down in practice.

Rob Campbell [00:29:43]:
All right. I'd love to see your— or hear your anecdote afterwards, but let me give you my update on this and my strong opinion on it. So a few weeks ago, Here's the backstory. A few weeks ago, we talked about Flathub effectively banning newly submitted applications created with AI. Now, Joey Sneddon at OMG Ubuntu has followed up with an article titled Flathub's AI Slop Ban Looks Like It Was the Right Call. The argument is based on research by a Linux developer, Evangelos, Paterakis, who, uh, examined 120 projects that had received Flathub's AI slop label. According to his review, only 32 still showed signs of active development, while 88 appeared to have been abandoned. But I don't think that proves Flathub was right.

Rob Campbell [00:30:48]:
Um, you know, in fact, Flathub may have helped produce that very result now, you know, being used to justify its decision. You know, how many— also, I'd like to know how many non-AI applications were also abandoned in that time period, or maybe mistakenly rejected, or who knows. But, you know, I feel like there's, you know, more metrics to this than what's led on here. You know, and also, you know, for independent developers, getting an application onto Flathub provides exposure your users and feedback. You know, when you remove one of those prominent ways for Linux users to discover an application, you know, fewer people are going to install it. When nobody is using your software, reporting bugs, or suggesting improvements, it becomes much easier to lose motivation and move on. I know I've been there. You know, developers are often motivated by knowing that somebody is Actually using what, what, what they created.

Rob Campbell [00:31:55]:
And, you know, and having users to file bug reports, feature requests, uh, you know, lets the dev know that updates are needed and, and to keep working on it. Um, you know, I've thrown things out there and like, oh, nobody's using this, I'm bored, and I've abandoned them. So it has nothing to do with AI abandoning it. Um, you know, also, you know, so, so Flathub.

Jonathan Bennett [00:32:16]:
Yeah.

Rob Campbell [00:32:17]:
closes that door. And now people are pointing at the developers who walked away as evidence that the door should have been closed. Um, I don't know. There's also questionable assumptions that software without commits for like a 2 or 3 month timeframe is abandoned. Programs don't need constant updates simply to prove that they are alive. Some applications are simple utilities designed to do one specific job. Once the program works, uh, has no known vulnerabilities, and, and has no outstanding bugs, development may just be finished. You know, that's not abandonware.

Rob Campbell [00:32:58]:
That's, that's just completed software. The, the developers, you know, even like Twitter, are all those— is all that, all that code that, uh, they put out there that hasn't got updates, is it abandoned? I mean, I don't know, maybe it is. But anyway. developers may just be working on other projects, you know, too, while this release simmers out there. You know, they may be waiting for some reports to come in, uh, or, or some free time to come back, or waiting for someone to actually use it and see, is anyone going to use my stuff? Should I keep going at it or, or just move on? I don't know. So fortunately, even, uh, Pederikus acknowledges that commit activity cannot definitively determine whether a project is maintained and that some software simply does not require further changes. Flathub has, you know, sure, they have the right to stop automated submissions, reject developers who refuse to communicate, and protect its voluntary reviewers from spam. They can do whatever they want, but, you know, I think those problems should just be addressed directly rather than this very indirect report approach, which I've said much of this before, you know, judge applications by their security functionality and whether a real person is willing to support them, not by whether that person used AI to help them create, create it.

Rob Campbell [00:34:26]:
You know, a lack of GitHub activity after being denied access to a major audience does not prove the application was slop. And may simply prove that developers need users too. I have an application on GitHub that I released several months ago. I've had no bug reports or reports of any kind. It just sat there out there for a while. It wasn't abandoned. I probably have no users, I don't know. But anyway, it wasn't abandoned.

Rob Campbell [00:34:55]:
I just worked on some other things for a while, and then just this week I updated that project. Uh, with some new features. Things don't need constant updating. It's like we have, you know, our ADHD with screen time has transferred over into our need for updates into, uh, every project out there. So anyway, let's not pretend inactivity in such a short time span means literally anything at all. Come back to me in, you know, a year or so. And tell me where we're at with these stats, you know, and bring some non-AI numbers with to compare. You know, I, I just need more.

Jonathan Bennett [00:35:36]:
I think that's an interesting point, asking about, uh, what the non-AI, uh, numbers would be, how many of those are, you know, inactive in the start— in the same period and therefore would be considered abandoned.

Rob Campbell [00:35:47]:
Yeah, I've created many non-AI applications that, uh, you know, I just got no activity. I was like, yeah, whatever, no one else wants to use it. I'll just—

Jonathan Bennett [00:35:54]:
I'm done. Yeah, I sort of think, and I think this bears it out, that really what Flathub is doing here is they want to have a sign to tap. And, you know, they'll make exceptions. I know firsthand that they'll make exceptions, but they want to have a sign to tap when one of these projects comes in and it's like, I really don't think that's a good fit. Or, you know, we— Let's be honest, that is what happens, right? People vibe code something and it's really Interesting. It's really exciting. The price on their preferred model goes up or they lose access to the free trial or whatever. And then yeah, some of those things do get abandoned.

Rob Campbell [00:36:33]:
And by a flag to tap, you mean an excuse, right?

Jonathan Bennett [00:36:40]:
It's the whole don't make me tap the sign meme, right? I think that originally came from The Simpsons or something and the bus driver had the sign that he tapped anyway.

Ken McDonald [00:36:48]:
Oh yeah.

Jonathan Bennett [00:36:48]:
Yeah. It's—

Rob Campbell [00:36:49]:
okay. Gotcha.

Jonathan Bennett [00:36:49]:
Not, not necessarily an excuse, but just a simple way to make the explanation that, look, you know, some of these things we reject the right to refuse, and this is why.

Rob Campbell [00:36:57]:
You know, but I, I say, you know, you know, like you said, they make exceptions. And you're— yeah, you're telling me that they're gonna— they make exceptions. Okay, so you're also telling me that there's other times when they just tap the sign and say, AI, get out of here. If they're looking at that project and there's a problem with it, why don't they address the actual problem with it? You know, this is bad code.

Jonathan Bennett [00:37:24]:
Well, because the actual problem you see is— well, not even that it's indecipherable, but like it has all of the signs of the AI wrote it and we don't see any human fingerprints on it. We don't necessarily know who the person is that's submitting it and we don't want to have something in FlatHub that's going to be abandoned. And, you know, 6 months from now, we're going to have to have the conversation about pulling it back out. I think that's what they're trying to avoid. So I have the story here that we worked on the Meshtastic Android app, and James, our developer there, just did a knockout job. He made it Kotlin Multiplatform. He turned it into a KMP app. And so therefore it runs on the desktop.

Jonathan Bennett [00:38:07]:
And one of the things that we wanted to do was get it out as a Flatpak. Made that work, and then we want to get it out on Flathub. So we made the application on Flathub. Here's the thing, the Mesostastic desktop, it works. You know, we've got our ducks in a row. And they did what the Flathub guys do. They came back with a whole bunch of nitpicky changes that they wanted us to make. You know, and I mean, this is pretty typical for any time you want to get software uploaded somewhere.

Jonathan Bennett [00:38:32]:
They've got their, you know, they've got the way that they want things to work. It's like, well, this name should be capitalize this way, or, you know, you should make this change upstream. You know, a lot of things that to us seemed really, really picky, but to them, it was pretty normal. And then we went through all that, and then they put the tag on it. It's like, blocked because AI, because our Android guys used AI a lot to do the conversion from the old-school Java stuff over to Kotlin Multiplatform. And the comment on it was, blocked because AI. However, I'm certainly willing to do the, uh, I forget what he called it, not a workaround, um, not exclusion. Basically, I'll give you a pass because you guys are an established project and obviously there's work continuing here.

Jonathan Bennett [00:39:17]:
And, you know, that sat there for a couple of days and then somebody went in and hit the button and we got uploaded to Flathub. And you can, you can go get Mitchtastic Desktop now on Flathub. And it's a few little problems because we've not run on the desktop before now. So there's some things that will still need work, as one would expect. But it was a reasonably painless process. And they obviously want good software to get uploaded when they know that somebody's going to maintain it. And I legitimately just think that what they're trying to avoid is to have dozens and dozens of abandoned projects where somebody got real excited and made the thing work and then ran out of tokens.

Ken McDonald [00:39:56]:
Well, I've got to agree with Joy and Tom Haruga, who, uh, Joy, uh, references, uh, in his article, that yeah, what we're seeing is not dedicated, uh, coders going in there and playing with this. It's people saying, oh, I've got this idea. They have the AI put something together for that idea, then something else catches their mind. So they turn and start looking at that and forget about that one.

Rob Campbell [00:40:30]:
I mean, that's fair, but that doesn't mean that every AI application, obviously, and Jonathan's point, but so I just wonder, you know, where's the cutoff?

Jonathan Bennett [00:40:42]:
I just want to take a second and stop and say, Isn't it so cool that you could have an idea and use an agent to build something as sort of a, you know, somewhere between a mockup and a prototype? You can essentially get a working prototype just by asking the agent to do it. Like, I understand that that makes some people salty, and I get that, but at the same time, it's so cool that we live in a time where you could do this.

Rob Campbell [00:41:03]:
You know what? Just this week alone, this week alone, there is a tool, I guess I'll call it, without going too deep, that I always wish I had at work. And I've looked for stuff and I found open source that were, was really close, but I just couldn't get it to work. I couldn't get it to work. And I'm like, this week I'm like, what can I do? I'm like, oh, I've always wanted this. And it took me a good, probably 8 hours playing around with, maybe not even 4 to 6 hours, probably playing around, getting exactly how I wanted it. And, you know, I've been looking for this tool for probably years. And it's like, you know what, here it is. And I'm actually not even in the department anymore that would utilize it or tie it in, but I made it and I tested it, it works.

Rob Campbell [00:41:49]:
And then I sent it over to the guy. I'm like, hey, uh, back when I used to run your department, uh, I always wish we had this, so here you go. I gave him the code and then I'm like, wait a minute, I'm gonna put this on GitHub. And my name I used was maybe a little— initially was, uh, could have had some copyright issues. So I'm like, but let me come up with a name for it first because it's very generic and whatever. And so I came up with the name of like an hour later. I'm like, here's my GitHub link. So I mean, something I wanted forever that I could not find.

Rob Campbell [00:42:23]:
I tried to make, I just, I couldn't get it to work right in the past. And, you know, an afternoon, I got it. It was awesome.

Jonathan Bennett [00:42:29]:
Is this afternoon project going to show up again after we close the stories for the day?

Rob Campbell [00:42:34]:
No, this project is, it's, it's, it's not a Windows, it's not a Linux application actually.

Ken McDonald [00:42:40]:
Okay.

Rob Campbell [00:42:40]:
Okay. It was more for the MSP side for, for a lot of people running Windows. So it is a Windows app.

Ken McDonald [00:42:46]:
Yeah.

Rob Campbell [00:42:47]:
Yeah.

Jonathan Bennett [00:42:47]:
Yeah. I got it.

Ken McDonald [00:42:48]:
What I've been enjoying with the AI being able to write scripts, preferably Bash, is I can throw it the What I would take hours to get all the logic worked out of, find the different options I need for certain commands to specifically do what I want.

Jonathan Bennett [00:43:15]:
Just, yeah.

Ken McDonald [00:43:17]:
Say I want to back up this group of subdirectories onto this external drive, and in a few minutes it's Got a working prototype. That's what I'm going to call it. Yeah, I can then dig through and personalize even more. And, oh, so, but I remember where it would take me days playing around trying to get something to— and where I could sometimes make painful mistakes while playing with.

Rob Campbell [00:43:50]:
So to get back to this Flathub thing, Um, and no one's gonna have this answer, but, you know, let's just say I got more time. I know programming. I'm not an expert in many languages. I'm, I'm a novice. But with AI, you know, I can, I can make stuff and I can go and I can understand the code, um, and make manual adjustments and know what's going on. But, uh, so I just wonder, like, at what point, if I made something and something I really want to maintain and get out there, what would I need to do to be considered established enough to get their good graces? I'm just—

Ken McDonald [00:44:25]:
start putting features that you don't really need into it.

Jonathan Bennett [00:44:30]:
That's an interesting question. I think that's a question for the FlatHub guys. It would be very interesting to ask them. I would say that it probably is going to boil down to continuing to support the program. Now, how long do they want to see that support before they consider it to be like a well-maintained, established I don't know. I don't know what the answer is going to be on that.

Rob Campbell [00:44:50]:
And, you know, I'd also like, I mean, you say continue to support it, you know, right now, let's just say I make a small thing. Maybe I have 5 people using it. Continue to support it. Does that mean I just have commits going on it for, you know, a few years or whatever that timeframe is? Or does it have to be something that people have actually heard of like Meshtastic? I mean, people, maybe not everybody, but a lot of people Obviously, that's how we made it in, right?

Jonathan Bennett [00:45:15]:
We have that name recognition. I get that. There's something to be said for, like, you shouldn't be judged just by whether you have name recognition, because there's lots of excellent projects out there that nobody's heard of, or that very few people have heard of, because they're new, or they're a niche, or they've just not gotten that lightning in the jar yet to sort of enter the public consciousness.

Rob Campbell [00:45:37]:
So I partly wonder how much of yours was name recognition.

Jonathan Bennett [00:45:40]:
Oh, I'm sure, I'm sure that's what it was. They've heard of us.

Ken McDonald [00:45:43]:
75% name recognition. Yeah, probably 25%, at least 25% perseverance. And I think that's probably going to be the— what's going to be the tiebreaker is whether or not you keep coming back and say, how about now?

Jonathan Bennett [00:46:00]:
Yeah, I think that would probably be the way to approach this, really. It's like, you know, you, you do the submission to Flathub, you take care of all the things they say, and they're going to say, well, it's an AI application. But first off, first off, make that clear up front. If you've, if you've written an application almost completely using AI, like, just, just put it in there. Like, hey, I understand this may get denied because of this, but I would eventually like to make it into Flathub. So I'm making the application now. I intend to keep, you know, updating it. Uh, give me some guidance on what we need to do, right? And, uh, that would be the first thing I would say.

Rob Campbell [00:46:32]:
Is there like an application then where you fill out like Stuff and like you put that in.

Jonathan Bennett [00:46:36]:
I think it's a, uh, I think it's actually a pull request inside GitHub.

Rob Campbell [00:46:39]:
Oh, okay.

Jonathan Bennett [00:46:40]:
It's how that's actually done. So yeah, there's, there's a field there where you can put notes and that's, that's the way that I would approach it. Like, hey, this, yes, this is vibe-coded, but yes, I intend to keep it updated. I would like to do this for the long haul and I would like to see it on Flatpak. Would love some guidance on what we need to do to make that happen.

Rob Campbell [00:46:54]:
That's my new goal. Have something in there.

Jonathan Bennett [00:46:57]:
Yeah, there you go. And that's a good goal.

Ken McDonald [00:47:00]:
You're like, I can see Flathub saying, Well, we'll look at it in 6 months.

Jonathan Bennett [00:47:04]:
That may be exactly my video. I don't know.

Rob Campbell [00:47:06]:
Years ago, I had an application in the Lindows Click-to-Run App Store. Obviously not 5-code. It was a very basic, very basic tool.

Jonathan Bennett [00:47:18]:
I mean, that's Rob's claim to fame right there. World-renowned by people who run Lindows and look through all of the applications. World-renowned. In that group of people. All right. Let's talk about UEFI here in just a second. We're going to let Ken talk about a new boot manager. But first, we're going to take one more quick break.

Rob Campbell [00:47:41]:
We'll be right back.

Ken McDonald [00:47:44]:
And Jonathan, this week, Bobby Borisov wrote about another entry to the growing list of Linux-supporting boot managers and bootloaders. It is called Visor and is a new open-source UEFI boot manager for Linux that combines the speed and efficiency of GRUB with the polished icon-driven interface of Refind. Now it's written primarily in C now and is capable of booting Linux, either EFI stub kernels, unified kernel images, Or even chain loading other EFI executables, including Windows Boot Manager. Now, according to Bobby, it is set apart from similar offerings by not using a scripting engine or numerous generated configuration files. That sounds like GRUB he's comparing it to there. Instead, Vysor relies on a single boot.conf file stored on the EFI system partition. Now, if your system firmware provides a compatible pointer device, which I believe mine does because I'm able to use my mouse when I'm in my BIOS, then you aren't limited to the keyboard but can select and launch entries with a mouse, touchpad, or even a touchscreen. There is also a text-based menu available as a fallback when graphical rendering is unavailable.

Ken McDonald [00:49:20]:
Visor even supports Secure Boot. When launched through Shim, Visor uses Shim's verification protocol to check Linux kernels, the UKIs, Windows Boot Manager, or other PE format EFI images before execution. Then of course, if verification fails while Secure Boot is active, Vysor will not start the image. Now, for more information, I'm going to recommend reading Bobby's article, and it does have a link to the, uh, Vysor GitHub where you can look at that. And if you want to dig through the code itself, see exactly how it's doing it. So Jonathan or Rob, are either one of y'all going to play around with this?

Rob Campbell [00:50:10]:
So I don't know, it looks pretty interesting.

Jonathan Bennett [00:50:14]:
You know, I would have to look into doing UEFI inside of like a KVM. Last time I tried to do that, it was a very complicated procedure to try to make that work. Hopefully it's easier now.

Ken McDonald [00:50:27]:
Reading through a Reddit subreddit that dealt with this, the author of the program said He actually copied his EFI partition into a VM so he could get the images that he was displaying on his GitHub.

Rob Campbell [00:50:51]:
I like the graphicalness. I saw the headline this week and I did not read the article. Is it like it's configurable? Like, so can you like, do you know how configurable? Can you like really make your own graphical stuff any way you want or is it limited to that? to that almost Apple look that I see on the screenshot.

Ken McDonald [00:51:10]:
I'm thinking this may be what I use that HP laptop that I was gifted for playing with.

Rob Campbell [00:51:23]:
Yeah.

Jonathan Bennett [00:51:24]:
You know, I see people complaining that it's got the anime-style backgrounds and the response to that is you can change the background. So I know you can at least change that. If you don't want the anime-style eyeball looking at you from the middle of your screen.

Ken McDonald [00:51:38]:
Yeah, basically you're just putting a group of icons or images that you want.

Rob Campbell [00:51:45]:
I'd love to see something like as configurable as a web page or like a GUI IDE for application where you could just, if I wanted to put Windows icon up here and And, you know, uh, CacheOS down here and something else here and just be all goofy and weird and stupid, you know, to my heart's desire. I don't know. I'd love to see something like highly configurable there.

Ken McDonald [00:52:15]:
But yeah, I don't know.

Rob Campbell [00:52:17]:
It looks like it's possible.

Ken McDonald [00:52:18]:
That's probably what I'll end up doing is installing it on that HP and playing around with it there.

Jonathan Bennett [00:52:25]:
Yeah, that'll be fun. Let us know how that goes. It's always nice to have more options. And you're right, it is prettier than GRUB.

Ken McDonald [00:52:34]:
Have either of y'all ever used Refind?

Jonathan Bennett [00:52:38]:
Once or twice years ago, back when we first started doing UEFI on real hardware. I tried to support a laptop that was just buggy, buggy under GRUB. And I think Refind was what we ended up using that made it boot.

Rob Campbell [00:52:53]:
You know, I'm pretty sure you can make GRUB prettier than it, than the current default. Well, I used to customize it more back in the old days.

Ken McDonald [00:53:01]:
Probably.

Jonathan Bennett [00:53:02]:
Yeah, you probably can. But I would lose my hacker cred if it was something other than white text on a black background.

Rob Campbell [00:53:09]:
I mean, you got GRUB customizer, which some say don't work.

Ken McDonald [00:53:12]:
And it might be easier now that you've got an assistant to help you.

Jonathan Bennett [00:53:16]:
True. Rice my GRUB.

Rob Campbell [00:53:19]:
Next week, I'm going to have my own bootloader for y'all.

Jonathan Bennett [00:53:24]:
That's a bad idea, Rob. That's not a good idea. No, that's AI slop. Get out of here with that. Oh goodness. All right, let's talk VR. This is something that I've been eagerly awaiting. What's new with the Valve VR headset, Rob?

Rob Campbell [00:53:41]:
Yeah, so we've all been excited about all the new Valve hardware. hardware, uh, the whole hardware lineup in general, uh, for a while now. And we finally have another tangible sign that, uh, that one of those products is getting closer to being ready. Valve's upcoming wireless VR headset, the Steam Frame, um, not to be confused with anything that, uh, Jeff was confused and thinking the frame was, uh, a few weeks ago. It's, it's their VR headset. Anyway, has just had an important piece of its Linux foundation released publicly. Valve has been working with the open source consulting company Collabora on something called HoloCore. This is a pure ARM64 port of Arch Linux that will serve as the foundation of the operating system running on the Steam Frame.

Rob Campbell [00:54:42]:
Now, for the first time, Collabora has published the project source packages, compiled binaries, and development container for people to explore. This is necessary because the Steam Frame will run on a Qualcomm ARM processor instead of the traditional x86 processor used by the Steam Deck and Most PC gaming. Standard Arch Linux does not officially support ARM64. So Valve and Collabora have had to build much of that infrastructure themselves. Even though Holocore currently contains only the software needed for the Steam Frame development and image creation, that limited collection still includes several thousand packages and their dependencies. Collabora also has created a repeatable build system capable of following Arch Linux constantly changing rolling release repositories. You know, now this, now this is, it's not a complete Steamframe operating system. And it certainly is not a downloadable image that will magically turn your random ARM device into a Valve head.

Rob Campbell [00:56:01]:
But it is kind of a behind-the-scenes infrastructure that has to exist before Valve can ship the real product. It also gives the open source community an early look at Valve's work to bring SteamOS beyond the x86 hardware. That could eventually matter for far more than VR. The same foundation could help SteamOS reach ARM-powered handhelds and laptops or other gaming devices in the future. Most importantly, code is harder to dismiss than rumors. Valve still has not given us a final Steam release— Steam Frame release date or price, but this project is clearly moving beyond prototype and private development. The operating system foundation is being built, tested, and now open for outside experimentation. We still do not know exactly when we will be able to buy the Steam Frames, but every time Valve releases another real piece of platform, it feels a little less like future hardware and a little more like hardware that is actually being prepared to ship.

Rob Campbell [00:57:14]:
I believe they were saying 2026, rough timeframe. So, We're halfway through if they keep on that. Maybe I could be wrong on that date, but 6-ish more months to go. So hopefully we'll see something this year. I know this is something you're maybe even more excited for, Jonathan.

Jonathan Bennett [00:57:36]:
Yeah, I've been watching this. I think it's a pretty cool idea. I'm going to blow your mind and tell you that we don't have 6 months left of the year. We have like 4 and a half.

Ken McDonald [00:57:46]:
January, February, March, April, May, June, July already.

Rob Campbell [00:57:51]:
July, 5, 6, 5 and a half, I think.

Jonathan Bennett [00:57:55]:
I mean, get ready, get your Christmas spirit ready.

Rob Campbell [00:57:59]:
Yeah, they'll be out in stores before you know it.

Ken McDonald [00:58:03]:
Yeah.

Jonathan Bennett [00:58:03]:
Go ahead, Ken.

Ken McDonald [00:58:05]:
But what I think that this is for primarily is so that before they even get that Frame Visor launched, they can give developers the option, opportunity to play with that environment and start maybe possibly writing programs or games specifically for that environment.

Jonathan Bennett [00:58:29]:
Yeah, because I'm sure that there will be some dev kits that they send out, and I don't know about— yeah, you know, you talk about— I'm sure there will be games written specifically for it. Because it's such a different thing than anything else has ever done. I guess it's similar in some ways to the Quest, right? Like the Meta Quest, I think they call it. It was kind of a standalone headset too. It was just so underpowered was the problem. Whereas this is going to be much more of a real sort of Linux experience. And so lots of things are going to just work. I look forward to it.

Jonathan Bennett [00:59:00]:
I think it's going to be a neat—

Ken McDonald [00:59:01]:
Meta Quest, formerly called the Quest? Before Meta bought it?

Rob Campbell [00:59:06]:
No, it had a different name, uh, before Meta. I can't remember what it used to be. What was it? Okay, come on, audience, tell us. It was the, uh, Oculus. Oculus Quest. Yeah, too slow, audience.

Ken McDonald [00:59:22]:
Yes.

Rob Campbell [00:59:23]:
Yep, yep.

Jonathan Bennett [00:59:24]:
You're right, you're right. Yeah, yeah. Okay, I thought those were 2 different companies.

Ken McDonald [00:59:30]:
The article I read about this by Markus Nester actually thinks that you'll see curious developers spinning it up in it, pulling it to put in a container in it, spin it up with Distrobox.

Jonathan Bennett [00:59:47]:
Oh, for sure.

Ken McDonald [00:59:48]:
And plus using QEMU emulation on the x86-64 host. And then try building packages themselves using the published preview repository that Valve's got.

Rob Campbell [01:00:02]:
Just put it on a Raspberry Pi 5.

Jonathan Bennett [01:00:05]:
You probably could. It probably would run there, actually. It would be a really interesting development environment.

Rob Campbell [01:00:10]:
Put 2 monitors there for the 2 lenses, and I'll just put those monitors close to your face.

Jonathan Bennett [01:00:15]:
Yeah, that's gonna look great. It's perfect. Let's do that. Let's get right on that, Rob. Right on that.

Ken McDonald [01:00:23]:
Or just go with a single monitor, um, so you can see how that 2D game's gonna feel.

Jonathan Bennett [01:00:31]:
Yeah, yeah, absolutely.

Rob Campbell [01:00:34]:
Well, even— hold on— even a 2D— a 2D— hold on, he says a 2D game though is going to have to be offset because you're going to have 2 screens. So if you have a 2D game played in a, in a, in a, a VR headset, you, you're gonna have to offset that, right?

Jonathan Bennett [01:00:54]:
I mean, you project it, you project it onto a virtual screen.

Rob Campbell [01:00:58]:
Yeah, yeah, the system, I guess, would, would—

Jonathan Bennett [01:01:02]:
Yeah, the system's gonna do it, but it's gonna make it look like a— what it does is it'll take it and it'll make it look like it's a big screen TV floating in front of your face, essentially.

Rob Campbell [01:01:13]:
Sure. That's how like, uh, I've, I've, I've, I've never actually watched a show or a movie or anything in the, uh, in the Meta Quest, but you can watch Netflix in there and it just makes it look like a 2D screen in front of you.

Jonathan Bennett [01:01:28]:
Now what, what could be really interesting, and I know that some projects have done this, there's been some work done on this, but like, uh, uh, a game that runs on a known game engine. there is actually depth information in there. And I've seen people do crazy things like hack the game engine and pull that depth information out real time. And so when you play the game, instead of just projecting it flat, it turns it into a 3D experience. And that's pretty cool when that works well. I've seen people do that all the way back—

Ken McDonald [01:02:00]:
Is that—

Jonathan Bennett [01:02:00]:
NES and Super Nintendo games, that when you project it, it'll do that. It'll make different layers. Which, you know, it's, it's a clever hack and it works really well on some things.

Ken McDonald [01:02:10]:
Like the 3DS?

Jonathan Bennett [01:02:12]:
I mean, the 3DS is kind of cheating. It's already, it's already doing that. Uh, all right, Ken, you want to tell us about Clonezilla? What is coming up there?

Ken McDonald [01:02:22]:
I'll be more than happy to, especially since Bobby Borisov and Markus Nächster took the time to write about the latest version of— and yes, it's still my favorite disk imaging and cloning utility. As Jonathan said, we're talking about Clonezilla. Live. This time it's version 3.3.3. This latest version is powered by Linux kernel 7.0.14 and based on the Debian— so that's the unstable repository— as of July 5th, 2023. Fairly recently. Now, according to Bobby, a key addition is reverse connection network cloning support in the OCS-on-the-fly utility. In this mode, the remote system initiates the connection rather than the controlling system, which is useful in environments where firewalls, NAT, or other policies restrict incoming connections.

Ken McDonald [01:03:25]:
It allows administrators to perform remote cloning without adding inbound firewall rules. I'm sure, Rob, you've had experience with that. Now, according to Marcus, the OCS-on-the-fly tool now has the ability to handle setups with multiple disks that have existing partitions. And this update also adds the network-manager-tui package to the live system. so you can use Network Manager's text mode interface. Clonezilla Live 3.3.3 also introduces 2 new tools, namely cnvt-ocsiso-qcow2 and ocs-check-initrd-module, and improves the disable _sudo_use_pty function in the ocs-live-hook-functions tool and updates the SUPP_BOOT_PARAM_OCS Live Extra variable for use with Netboot clients.

Rob Campbell [01:04:46]:
Oh.

Ken McDonald [01:04:52]:
Now, the Memtest Plus memory testing tool is updated to version 8.10 in this latest release. So that alone might be the reason to get it. Now, even though Marcus thinks this is a pretty small issue, I'm still going to recommend reading both articles for those details I didn't cover.

Jonathan Bennett [01:05:13]:
Yeah, absolutely. I like that reverse connection thing. So like, that means you can set a server up in the cloud somewhere with the images you want to be able to put out. Then when you get to your computer, you pop your Clonezilla key in, you give it that URL, and it'll reach out and grab it and pull it down. Like, that's super cool for doing, you know, on the fly, on like in situ cloning of machines. That'd be great. Rob, you guys should use that.

Ken McDonald [01:05:37]:
Or for an enterprise, they can have all the images sitting on a central server, get to the remote system and long as it's connected to the local network, boot up and have it automatically just download the image.

Rob Campbell [01:05:53]:
Yeah, I've used Clonezilla many times, but it's been a few years since I've really needed it for much. Maybe it's been a couple of years since I've cloned a system. But yeah, I don't remember any of those remote network cloning things in there. I don't know if they were not there or I just didn't need them. But that sounds like—

Jonathan Bennett [01:06:14]:
That's new in this release.

Rob Campbell [01:06:16]:
Oh, that's all they were talking about.

Jonathan Bennett [01:06:17]:
It's a new feature.

Rob Campbell [01:06:18]:
Oh, that— yeah, that sounds really cool. I can see a lot of uses for that.

Ken McDonald [01:06:23]:
So that way, if you've got, say, you, uh, back up your, uh, system or even just one of your bootable partitions to, say, a file server you got, you could then just use the, uh, Clonezilla Live, boot up, pull that straight back down to that system.

Jonathan Bennett [01:06:44]:
Yeah, I think it's gonna be useful for people, admins that do like a golden image.

Rob Campbell [01:06:48]:
Yeah, that's what I was gonna say.

Jonathan Bennett [01:06:50]:
I've got my Linux or my Windows or whatever, whatever operating system you wanna run, you do your configuration, and then usually there's a step where you sort of make it a generic install rather than for that machine. And then you can capture that and throw it up on a server somewhere and then just run through your machines and clone them all in record time. I think that is, I think that's pretty interesting.

Rob Campbell [01:07:12]:
Yeah, I was gonna say, I mean, my backups, I don't use Clonezilla for backups because I usually have things with like versioning in it and stuff like that. But golden image sounds like a really good use case.

Jonathan Bennett [01:07:24]:
Although there is something to be said for using Clonezilla as a backup. That way, if you're, you know, if it's an important machine and it goes down, if you get a Clonezilla image, you can, you know, you can spin up a VM of it in no time.

Rob Campbell [01:07:33]:
I mean, you can do that. You could do a the right backup tools too.

Jonathan Bennett [01:07:37]:
Yeah, yeah, indeed.

Ken McDonald [01:07:40]:
But that's one of the main reasons I use Clonezilla is I'll back up the partition before I do any kind of upgrade, system upgrade on it.

Jonathan Bennett [01:07:50]:
Yeah, I mean, that's kind of the paranoid approach, but sometimes paranoia pays.

Ken McDonald [01:07:56]:
It kept me being able to go roll back to Ubuntu 20 when I tried upgrading to up to Ubuntu 22.04.

Rob Campbell [01:08:05]:
You just need a real backup appliance system like your backup.

Jonathan Bennett [01:08:09]:
Go back to episode, I have no idea what episode, but whatever it was, ask the agent, ask Google Chat what it was. It'll tell you. All right. Let's take a quick break and then we're going to roll into our command line tips. So don't go anywhere. We'll be right back. All right, Rob, what you got for us? All right.

Rob Campbell [01:08:27]:
This week for your command line tip, I have some AI slop for you. So anyway, a little backstory here. One of the tools I used to always like to use when I was buying new memory is, you know, rather than open it on my computer and look at it like, what do I have in here again? I would go to crucial.com and they had a scanner that would tell me what I had in there, what What type and what my max is and what I could use and all that. And then sell it to you. Well, I'm assuming now that they— you can still buy RAM there. I looked, but I could not find that scanner. And I'm assuming maybe they don't have it since they aren't focused on consumer memory anymore. I don't know.

Rob Campbell [01:09:16]:
But, and one other downside to that is that it was Windows only. So I had to If I ever did that, I had to be booted up into Windows. Well, now I decided to create something. Yes, this is my GitHub here. I'm calling this MemReady. It's a very simple tool that will basically tell you these things. I'm going to run this MemReady. It outputs So this is one of my Proxmox systems.

Rob Campbell [01:09:53]:
So it outputs how much RAM is installed, what's visible by the OS, max supported RAM in the system, 64 gigs. So it says I have 32 gigs of upgrade headroom. Total slots is 4, used slots are 4. And, you know, when you go down further, it says, you know, DIMM slot is 8 gigs of this kind of memory in it. You know, 1, 2, 3, 4, they all have, they all have the same. Yeah, you got manufacturer, a bunch of details there. Also points out there's a warning in the notes, uh, maximum supported memory is reported by firmware and SMB BIOS and should be confirmed with the computer or motherboard manufacturer before quoting an upgrade. So there's a little disclaimer there.

Rob Campbell [01:10:37]:
It says the data sources, the tools that you use, if Um, and then there's some actions. You can refresh, you can do T to save the text, H to save as HTML, uh, J to save this JSON, C to save a CSV, and well, Q to quit. So just a little, uh, AI slop that I created this week to, uh, fill a need that, uh, I had. Um, okay, maybe it's not neat, there's ways around it, but it's replacing a tool that I used to like that I, I I just, I couldn't find an alternative to this, I suppose, besides using those, uh, built-in tools the way that it does it. But it's nicely formatted for you. And, and this is just the, uh, the initial release, so I'll see what I can do to update, upgrade it. And if you, if you use it, uh, and you got some ideas, go ahead and put an issue request. I'll try to get it, uh, Updated and fixed features, whatever it is.

Rob Campbell [01:11:38]:
Cool.

Jonathan Bennett [01:11:39]:
Is it a Bash script? Did you write it in Go? Is it written in Rust? You should have written it in Rust. Have you heard about Rust? It's Go.

Ken McDonald [01:11:47]:
Okay.

Jonathan Bennett [01:11:49]:
You wouldn't have all those memory problems if you'd written it in Rust.

Rob Campbell [01:11:52]:
Memory problems with memory.

Ken McDonald [01:11:55]:
There you go.

Jonathan Bennett [01:11:57]:
Rust solves all your memory problems, obviously. All right, uh, Ken, what do you have for us?

Ken McDonald [01:12:02]:
Well, this week I will demonstrate a Linux command for checking if a specific directory or file is a mount point for a separate storage device or drive. I find it handy for verifying that I have already mounted my USB drive before backing up to it via script. The command, of course, is mountpoint. And it does just this one function. And the— probably the most important one to remember is mountpoints followed by -h or --help. Because that would give you how the usage information for it, as well as describing the different options, such as how to keep it quiet, how to not follow any symlinks. And to get the major and minor device number of the file system or block device. But for those of you that are listening, I have actually just typed mountpoint -h, and it is displaying all that information.

Ken McDonald [01:13:13]:
Now, depending on your needs, the -d or the --fs-dev number can be used to print the file systems Major and minor device number. Let me go ahead and give you a demonstration of that. I can't remember if I've got this mounted or not. I'll find out here in a minute. No, it is not mounted. So let's go ahead and mount that. I've created a .img file that I can use to mount. And I found out that with sudo rs, Tab will tell it not to echo anything.

Ken McDonald [01:14:07]:
So there's another tip for you.

Rob Campbell [01:14:09]:
Absolutely.

Ken McDonald [01:14:10]:
So let's try that again. So as you can see, for those of y'all listening, I just did mount point -d ./demo folder, and it came back with a major number of, uh, 7 with a colon, then a minor number of 30 for that device.

Jonathan Bennett [01:14:34]:
Now for sure there's something mounted there.

Ken McDonald [01:14:37]:
Yep. And You can also use the -x for the device number. Now, without that in there, it tells you that that is not a mount point. But let's go back to the demo folder. And what it did with that one, it says, ./demo folder is a mount point. Okay.

Jonathan Bennett [01:15:10]:
That actually makes sense. Your NVMe is a block device. It's not where something is mounted, it's the thing that gets mounted, right?

Ken McDonald [01:15:18]:
But, uh, where it can be really handy is if you want to test to make sure something's, uh, mounted before you, uh, do another command. This time I'm going to use, uh, Mount point, uh, ./demo folder followed by the double ampersand. Jonathan, you remember what that does?

Jonathan Bennett [01:15:44]:
Yeah, it's, it's a, it's a small way to do like an if statement.

Ken McDonald [01:15:50]:
And then I'm following it with the ls command, and it's going to ls any files or folders that's in the ./demo folder.

Jonathan Bennett [01:15:59]:
Yeah, if mount point— if the mount point command returns true, then it will continue to follow the other side of the ampersands.

Ken McDonald [01:16:06]:
And in this case, it is true. So it comes— gives us 2 bits of information. It says that the ./demo folder is a mount point, and then it lists the lost+found directory that's inside demo. So if you are doing it in a script but you don't want to have anything cluttering it up, That's where you want to hit enter the -q flag. And this time it just does the lost plus found. Very cool. And of course, what do you think is going to happen if I unmount it and then run it?

Jonathan Bennett [01:16:49]:
It's probably not going to do anything.

Ken McDonald [01:16:52]:
That's exactly what happens.

Jonathan Bennett [01:16:54]:
Yeah, super cool. I was, I was musing as I was listening to this, is there a way to like lock a folder so that you can't write directly to the folder, but if you mount something inside of it, you can write into that? That's yet another way to kind of work around that, that potential case where you script something.

Ken McDonald [01:17:12]:
That's where you can use the mount options that we've covered in the past. One of those is to mount read-only, remember?

Jonathan Bennett [01:17:21]:
Yeah, I don't mean mounting something read-only. I mean the folder where you're going to mount it into before you mount it. You can, you oftentimes you can write files into there. You would, you would probably have to set your permissions on that folder to where your normal user can't write into it and then mount this root, something like that.

Ken McDonald [01:17:38]:
You'd have to be careful with that because then when you mount, you may find you can't write into the mount.

Jonathan Bennett [01:17:46]:
Yeah, if it inherits the permissions.

Ken McDonald [01:17:48]:
I don't know.

Jonathan Bennett [01:17:48]:
I'd have to play around with that. That's actually a really interesting question. I like using the mount point command to check that though. I've never done that in any of my scripts and I really should. They're all a little bit fragile because of that. So that's definitely something to think about.

Ken McDonald [01:18:05]:
But what you can do is go back to the episode that I did the find mount command. It's a bit more advanced. So you can actually look at what the permissions are.

Jonathan Bennett [01:18:17]:
There you go.

Ken McDonald [01:18:18]:
And if you find that the permissions allow you to write into it and you don't want that capability, then you can use the mount command to modify them.

Rob Campbell [01:18:30]:
Yeah.

Jonathan Bennett [01:18:32]:
All right. Well, I have taken a page out of Rob's handbook and I'm going to pitch something that I'm involved in in my show tip. And I don't have it pulled up or anything, but I'll just tell you that the Meshtastic desktop is now on Flathub. They were kind enough to look at us and go, oh, you're an established product, we'll let you in even though you use some AI. And that is the MeshTastic app as a Kotlin Multiplatform program and now available on Android and the Linux desktop and coming to other places as well. So if you want to get that, it is now out there and easy to go grab. I won't continue on pitching my own stuff though. I'll leave it there.

Jonathan Bennett [01:19:13]:
I appreciate Rob and Ken being here. I'm gonna give each of them the opportunity if they want to, to plug something. If they come back, if they've not abandoned me, I'm sort of solo here on the screen. There we go. Rob's back. We'll let Rob go first. What do you got to plug, sir?

Rob Campbell [01:19:30]:
Just my normal, for those who wanna connect with me, find out more. Maybe I should add a link to my GitHub on here. But for now, it's not on here. Uh, anyway, you can find me, robertpcampbell.com, which is also always at the bottom of the screen here for anyone watching.

Jonathan Bennett [01:19:48]:
Uh, over here.

Rob Campbell [01:19:48]:
There we go. Anyway, you get to that website, you'll find links to my LinkedIn, my Twitter. I'm still calling Twitter. Bluesky, Mastodon, and a place to donate a coffee to me in $5 increments. You know what, uh, I'm thinking with, um, Inflation. I should probably move that up to $6 at least, huh? For a coffee.

Ken McDonald [01:20:09]:
I should move that up.

Jonathan Bennett [01:20:10]:
I remember when the joke was that Starbucks was more like $6. That joke's not very funny anymore.

Rob Campbell [01:20:17]:
No, it's not.

Jonathan Bennett [01:20:19]:
No. All right, Ken.

Ken McDonald [01:20:21]:
Well, in the show note, I've got a link to a, Article that Abhishek Pavik wrote. And I was going to start off by asking, would you pay $5 for, uh, the book Understanding the Linux Kernel by Daniel B. Bovet? If you would, then definitely check out this link because about how you can get not only this book but 14 others for under $3 each.

Rob Campbell [01:20:57]:
Yeah.

Jonathan Bennett [01:20:57]:
Very cool. All right. Thank you guys for being here. I appreciate it. It's good to have the guys with us and good. I'm glad that I made it. It was, it was real uncertain there for a while if I was going to make it to the show today, but I got, I got in. We did it.

Jonathan Bennett [01:21:10]:
I found a nice quiet place to record from. The battery didn't die. The Wi-Fi mostly held out. It's a good day. Everything worked. All right. Appreciate you guys being here. If you want to find more of me, normally I would point over my shoulder here at the monitor showing Hackaday.

Jonathan Bennett [01:21:24]:
I'll just have to tell you about it. You can find more of my stuff at hackaday.com. Specifically, that's where FOSS Weekly lives, which, you know, we are continuing on with that and had a lot of fun. We've got some more fun stuff coming. I sort of poke fun at Rust. We've got a Rust project, Rust program coming up, and some other really neat stuff. I would say we've had some really fun shows here recently too. The one with Tris about open source legal issues was a lot of fun.

Jonathan Bennett [01:21:49]:
And then in the couple of weeks, we've also done the, uh, uh, the, the president of the Risk 5 Foundation, which was a really interesting conversation too. So go and check those out. Uh, other than that, we appreciate everybody that's here. I thank you whether you watch or listen, whether you get us live and on the download. And we'll be back— not on location, but we'll be back next week on the Untitled Linux Show.