Untitled Linux Show 264 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 Canonical and some enterprise offerings that they're bringing. Then we talk about Jellyfin and it's a bit of a leadership crisis. Not all is bleak there though. MPEG-4 Layer 2 is finally patent-free and a whole lot more. You don't want to miss it. So stay tuned.
Rob Campbell [00:00:19]:
Podcasts you love. From people you trust.
Jonathan Bennett [00:00:23]:
This is TWiT. This is The Untitled Linux Show, episode 264, recorded It's Saturday, July 25th. We have the worst analogies. Hey folks, it is Saturday and you all know what that means. It's time for the Untitled Linux Show and I am back. I am no longer in California. I am back in my comfortable home office here in the wonderful flyover state of Oklahoma. And we've got the guys back with us.
Jonathan Bennett [00:00:54]:
It's not just me. We've got Rob, we've got Jeff, we've got Ken, not necessarily in that order, but they are here.
Rob Campbell [00:01:00]:
Hello.
Jonathan Bennett [00:01:00]:
And we've got some Linux stuff to talk about. We've got some hardware to talk about, some software to talk about, some enterprise stuff to talk about. In fact, Rob is here to lead us off with an enterprise story from Canonical. Rob, take it away and tell us what in the world are those guys up to?
Rob Campbell [00:01:20]:
All right. So Canonical has launched something called Enterprise Store. But despite the name, this is not another graphical app store competing against Software Center, Snaps, or Flatpak or any of the others. Instead, the Enterprise Store is designed for companies that need tighter control over their Ubuntu software or where their Ubuntu software comes from, which versions are installed, and when updates are allowed to roll out. The Enterprise Store runs inside the organization's own network, and acts as a local proxy between its Ubuntu systems and Canonical software stores. Rather than every server, workstation, or device connecting directly to the internet, they can retrieve approved software through the center or central internal store. This is especially useful for corporate networks with strict firewall rules, regulated industries with formal change control requirements, Air gap systems that may not have internet access at all. And for fully disconnected environments, updates can be downloaded elsewhere, manually transferred into the network, and then distributed through the enterprise store.
Rob Campbell [00:02:41]:
It also caches download files locally, which reduces bandwidth usage and makes repeated installations faster. Administrators can also pin specific software versions. That means a company can test and approve one version, keep its system on that version, and prevent an automatic update from unexpectedly changing something in a production environment. Canonical is also supporting high availability deployments. Multiple Enterprise Store servers can operate behind a reverse proxy, reducing the chance that one failed server will prevent an entire organization from accessing software or updates. The store currently provides a single control point for both Snaps and Charms. Snaps handle packaged applications while Charms are used to deploy and manage applications and services across environments such as clouds and clusters. Of course, this also is also another sign that Canonical remains completely committed to Snaps and enterprises.
Rob Campbell [00:03:53]:
People who already dislike Snaps probably will not see this announcement as a reason to change their minds. But in an enterprise environment, the argument is a little different. Businesses are genuinely less concerned about package formats. updates and more concerned about security, reliability, audit records, and making sure an update does not break 500 machines on a Monday morning. Or maybe worse is a Friday afternoon.
Jonathan Bennett [00:04:23]:
Friday afternoon is when that's way worse.
Rob Campbell [00:04:26]:
The Enterprise Store, you know, but on the other hand, you got time to fix it before most of the, you know, everyone blows up. But anyway, the Enterprise Store is included with Ubuntu Pro. And because Ubuntu Pro is free for personal use up to 5 machines, Administrators can even test it in a home lab before considering a rollout. This probably will not be exciting news for the average Ubuntu desktop user, but organizations trying to run Ubuntu inside restricted, highly controlled, or completely disconnected networks, Canonical may have just made software management considerably easier for them.
Jonathan Bennett [00:05:07]:
Yeah, it's super interesting to me to see this. Something else that could be really interesting to do with this is if you have an app that either is open source or for that matter is closed source, but like there's some sort of sourcing or licensing problem that you've got with it, but you're cleared to roll it out inside your organization, you just can't make it public. It would be super useful for that too. Or even like a, a testing version of an app or, you know, something like that, it would be really interesting to be able to roll it out on your own sort of local copy of the store.
Rob Campbell [00:05:42]:
So yeah, I mean, today you could set up your own repositories. I've done that years ago, mostly playing around. But so, but this probably provides some advantages over that.
Ken McDonald [00:05:56]:
I think where this would really be handy for the home lab would— is the fact that you could have your server act as the repository for your snaps, and then any other Ubuntu Pro systems that you've got on your local network could then pull from there instead of having to pull over the internet.
Rob Campbell [00:06:24]:
Yeah, I mean, that's, that's, I know, fairly common in the Windows management world using WSUS, it's called. But whereas instead of having 500 computers all go update the same program from the internet and use all the bandwidth, you know, it can, one can use the internet bandwidth and everyone else can get it from there.
Jonathan Bennett [00:06:45]:
I'm not sure if that shows a real lack of self-awareness or a really great sense of humor that Microsoft named that WSUS. One or the other.
Jeff Massie [00:06:58]:
But I agree.
Rob Campbell [00:06:59]:
It's an acronym. I don't remember what it stands for. I think it's an acronym.
Ken McDonald [00:07:04]:
Yeah, it's pretty soft. Windows something.
Rob Campbell [00:07:07]:
Yeah.
Jonathan Bennett [00:07:08]:
All right. Yeah, interesting to see the Enterprise Store. All right, we have a lot more coming up right after this. We are going to take a very short break and we'll be right back. So don't go anywhere. All right, Jeff, tell us what we should be using instead of Plex in the Well, not such a problem, but that's why I wanted to cover this story.
Jeff Massie [00:07:32]:
So talking about Jellyfin. Jellyfin is the open source media server often viewed as the leading alternative to Plex. It's going through some significant leadership transitions. Within a span of a few days, all 3 of the project's most prominent leaders announced their departures from official leadership roles.
Rob Campbell [00:07:50]:
Wow.
Jeff Massie [00:07:51]:
So founder Joshua Boniface— Boniface? Sorry if I messed that up— stepped down as project leader. Core team member Anthony Lavato announced his departure, and co-founder Andrew Rabbitt also left the organization. Despite the sudden nature of these announcements, all parties involved have emphasized that the transition is amicable and that there's no indication Indications of a hostile, hostile fork or organizational split. Joshua Boniface explained that his departure was primarily driven by burnout. He stated he could no longer provide the mental energy or time commitment required by the role and felt continuing would negatively impact his mental health. According to Boniface, stepping aside was necessary because he was no longer able to perform the job at the level he expected of himself. He described the transition process as cooperative and expressed confidence in the remaining team members, many of whom have been responsible for key areas of the project for years. Boniface also reflected how dramatically Jellyfin has grown since its creation roughly 7.5 years ago.
Jeff Massie [00:08:59]:
He said the original expectation was that a small project— that it was for a small project, perhaps serving a few hundred or a few thousand users. Instead, Jellyfin's evolved into what he described as the leading free and open source media server platform, serving millions of administrators and many more end users. He expressed pride in proving that a community-driven open source alternative could successfully compete with commercial platforms. Anthony Lovato's departure stems from different reasons. While he's been less involved in writing code recently, he remained heavily involved in operational tasks such as managing app store relationships and backend administration. Lovato stated that changes in his personal life have reduced the amount of free time he can commit to the project. Unlike an immediate departure, however, he plans to remain involved throughout, and potentially, it's potentially a year-long transition period to help ensure stability and continuity. The story becomes more interesting when examining the departure of co-founder Andrew Rabbitt, Earlier this year, Rabbert was leading a major rewrite of Jellyfin Desktop, the project's desktop client application.
Jeff Massie [00:10:15]:
And in May of 2026, State of the Fin update, it's kind of a blog that they put out to give a little information what Jellyfin has going on under the hood. Rabbert described an ambitious modernization effort. The rewrite replaced Qt and Qt Web Engine with a Chromium embedded framework, upgrading the media playback pipeline, improving responsiveness— uh, responsiveness, excuse me— for significant future improvements. At the time, the work was presented as one of the project's major ongoing development efforts. However, Rabbitt later indicated that disagreements arose within the organization regarding how he was approaching the project. According to accounts referenced in the reporting, he encountered resistance from some members of the Jellyfin organization. The resulting friction eventually reduced his motivation and contributed to his decision to leave. Evidence of that separation can be seen in the GitHub issue opened on July 17th, where Rabbert announced that the desktop application was no longer an official Jellyfin organizational project and would require new branding and a new name.
Jeff Massie [00:11:28]:
One reason these departures attracted attention is that they come shortly after what appeared to be a positive status report for the project. The May 26th Jellyfin, that's the blog part, published its second state of the Fin update and report showed healthy development activity with hundreds of pull requests being merged each month, hundreds of issues being closed, and contributor counts remaining relatively stable. Development on the upcoming Jellyfish 12.0 release was progressing, performing, and the improvements have been merged in multiple client applications across Android, Android TV, Roku, Kodi, you know, ton of different platforms. Everything was getting updated. Everything is going wonderful. Sunshine, lollipops, and rainbows.
Ken McDonald [00:12:15]:
Mm-hmm.
Jeff Massie [00:12:16]:
Yet hidden within the optimistic update was a warning sign. Now, the Jellyfin team devoted an entire section of the report to burnout. They explained that growing popularity brought increased support demands, more bug reports, more feature requests, and more administrative overhead. The team specifically noted that AI-generated pull requests had become a major burden. Project maintainers reported that receiving large numbers of AI-assisted code submissions of varying quality and we've talked about that before on the show, created additional review work rather than reducing effort. The project ultimately published an official AI and LLM contribution policy in an attempt to manage that growing workload. And I think this was a good policy, personally. The policy does not completely prohibit AI-generated contribution, but it requires contributors to understand the code they submit and prohibits using an AI chatbot as an intermediary during technical discussions.
Jeff Massie [00:13:17]:
The maintainers stressed that contributors must be able to personally explain and defend their work, which I personally think is very reasonable to do. I think it gives a reasonable level or bar that somebody has to cross before they commit code to the project. The burnout section of the May report also reported another important issue. The team reminded users that Jellyfin is built entirely by volunteers. Maintainers noted that abusive behavior from users Repeated demands for support and negative reactions when contributors or when contributions are rejected all take a personal toll on developers. The team warned that this pressure was already contributing to delayed improvements and reduced motivation among both developers and administrators. And the recent leadership departures appear to validate concerns that key contributors were operating at or beyond their limits. Now, this is no indication that Jellyfin itself is in immediate danger.
Jeff Massie [00:14:18]:
You know, it's just the events illustrate a broader challenge facing open source software. You know, successful projects often depend on a relatively small number of volunteers. And as popularity grows, the workload can increase much faster than the number of people available to manage it. The main takeaway is that this does not appear to be a crisis of technology or project viability. You know, because like I said, it remains active, development continues, the remaining team members are expected to carry the project going forward. It's basically, the story is kind of about the human side of the open source development or the project. And the departures highlight the effects of long-term maintenance burnout, the increased pressure caused by a rapidly growing user base, and the new challenges introduced by AI-generated contributions. So the project's founders are stepping away, but they're doing so while expressing confidence that Jellyfin will continue to thrive under the new leadership.
Jeff Massie [00:15:18]:
But don't just take my word for it. Take a look at the article linked in the show notes for more details and to follow all the supporting links in the article. So happy viewing.
Jonathan Bennett [00:15:29]:
Yeah, interesting to see. I mean, hopefully for them, they can come out the other side of this and more people from the community will step up and take over that role. The way that I put it is that AI is exhausting, and I think it's true. It used to be that there was, we've talked about this before, like there's a certain amount of friction that you want in your project. You want the right people sending you pull requests, and you want the right amount of friction so that the right number of those pull requests come in. And one of the things that AI has done is it has drastically cut down on that friction. And that's great, except somebody's got to be there to do the, essentially the management on the other side to deal with all of those incoming pull requests. And not all projects are in a good spot for that.
Jeff Massie [00:16:14]:
Well, and you have to double-check AI. It will do wonderful things. And sometimes it just is like, what the heck happened here?
Jonathan Bennett [00:16:23]:
I mean, that's true, but I don't even know that that's the primary problem, right? And as AI gets better and better, you'll have less of that problem, but you'll still have a different problem, of essentially project management. Like you asked AI to do this thing and it did it perfectly, but that thing that you asked it to do, is it still a good idea for the project? And sometimes the answer is no. And if you just, if you let everyone that sends you a, you know, a good AI-written pull request in, you end up with just this massive monstrosity that's, you know, got everything and the kitchen sink bolted onto it. And that's—
Ken McDonald [00:16:59]:
Feature bloat.
Jonathan Bennett [00:17:00]:
Yeah, well, you run into, yeah, you run into code bloat really, really quickly because again, it's too, it's so easy to say, oh, hey, it's a cool idea. Let's do it. Hey, Claude, do this thing for me.
Rob Campbell [00:17:10]:
It's definitely changed, improved a lot the last few years. I mean, we don't see the pictures with the different amount of fingers and arms and limbs like we used to. And just this week, I found a new artist. On a Facebook reel. I'm like, that's pretty good. I'm listening to this stuff. I like this. So I went and Googled this artist.
Rob Campbell [00:17:35]:
And this was an AI, completely AI-generated artist. Everything, the song, the music, the person, nothing was— I'm like, no way. I'm like, this is good. I liked it.
Ken McDonald [00:17:48]:
Yep.
Rob Campbell [00:17:48]:
I was digging it.
Jonathan Bennett [00:17:49]:
What they did, and I've done, I'm sure what they did is they said, hey, Write me this song, you know, go to, I forget the name of it, Sono or whatever the name of the app is that'll do music for you. Write me a song, here's the title of it, but make it in the style of this artist. And it's somebody you already liked. And so you went and you listened to it. Yeah, I got one of those a couple of months ago. It came up on Pandora, I think, but it was, you know, it was kind of like in the sound of Sabaton, but it was a, you know, very different sort of stuff than what they already, as far as the lyrics go. And I'm like, I really like this. And I listened to it a few times.
Jonathan Bennett [00:18:22]:
I'm like, I need to find out more about this band.
Ken McDonald [00:18:24]:
Yeah.
Jonathan Bennett [00:18:24]:
And couldn't find anything about them. And it's like, wait a second, I know what's going on.
Rob Campbell [00:18:29]:
I was listening to song after song and they were all, they were all really good. They're also somewhat similar in ways, very similar in ways. But it was easy to find when I searched up the artist. It was like my autocomplete was, is AI? I'm like, that wasn't going to be my question. What's going on here?
Jeff Massie [00:18:52]:
I've had it review my week. You know, I'll say, hey, review my week at work, give me a summary, and put it in a comic strip style like Calvin and Hobbes, but replace Calvin with myself, with a short description of me. And it will spit it out. And it's like, wow, that looks amazing. Just like it was the original comic author wrote it.
Jonathan Bennett [00:19:16]:
Mm-hmm.
Ken McDonald [00:19:19]:
Going with, uh, one of the articles that I came across about this by, uh, Ernestus Napras writing for cybernews.com, my takeaway is not so much that AI contributed this. Life is what really mainly contributed to it.
Jeff Massie [00:19:36]:
And that's part of it too, for sure. It grew in size because remember, the main thing that people were using for the longest time was Plex. Well, then probably 2 years ago, Plex took a turn and they stopped really focusing on serving your own files and started trying to be this central—
Ken McDonald [00:19:54]:
Commercial.
Jeff Massie [00:19:55]:
Commercial media repository. So like, oh, if I want to watch Netflix, I'll just— or say I've got Netflix and Prime and something else. Oh, I'll just go to Plex and it'll find it for me and it'll route me to where I need to go. And it was trying to be this central entertainment repository. And they've kind of just left the home media streaming part behind. And that's not what people wanted.
Ken McDonald [00:20:20]:
They started doing more ad-supported streaming.
Jeff Massie [00:20:23]:
Well, that too. And trying to host their own movie channels. And so a lot of people went, you know what? I just want Jellyfin. I've got this media that's on my computer. I just want to serve it to myself. I don't need all this other cruft.
Rob Campbell [00:20:39]:
I really want Jellyfin to become every bit as good as Plex, but hopefully with a limited feature set, right?
Ken McDonald [00:20:47]:
Yeah.
Rob Campbell [00:20:48]:
I don't want to be like Plex. I want to be as good as Plex though.
Jonathan Bennett [00:20:52]:
Yeah.
Ken McDonald [00:20:54]:
Do you want the feature that allows you to use your HDHomeRun as a tuner for For it and act as a DVR?
Jeff Massie [00:21:05]:
Yeah, he just doesn't want it tied into the Netflix and Hulus and all the other stuff. It doesn't need its own independent streaming channels and movie channels and all the other stuff. It just serves what I have in my house.
Jonathan Bennett [00:21:21]:
So there's a story coming up that is probably going to make it a lot easier for Jellyfin to achieve that. Because Ken, do some of those things still use MPEG-4 across the airwaves?
Ken McDonald [00:21:34]:
Possibly.
Jonathan Bennett [00:21:37]:
I out-segued Ken.
Jeff Massie [00:21:41]:
Ken got segued.
Jonathan Bennett [00:21:43]:
Take it away, Ken. Tell us about the patent situation.
Ken McDonald [00:21:47]:
Well, Jonathan, this week we heard from, and it's a group of people here, Bobby Barsalf, Michael Larabelle, and Saurav Rudra. About the last active MPEG-4 Part 2 patent expiring. Now you ask, what is MPEG-4 Part 2? Well, it's also known as MPEG-4 Visual, a relatively older video compression standard released sometime in 1999 by the Moving Picture Experts Group. That's— I think that's where we get the name MPEG from. That forms the core technical basis for popular early 2000s video formats like DivX, Xvid, and early versions of— ready for it— Nero Digital. Now, according to Surov, the MPEG-4 Visual Patent Portfolio license has been in effect since January 1st, 2000. Covering essential patents behind the MPEG-4 Part 2 visual standard from more than 30 licensors, including companies such as Canon, LG, Microsoft, Panasonic, Sony, and Toshiba. I recognize some of those names.
Ken McDonald [00:23:08]:
Now, the very last patent standing belonged to Siemens AG And it was active only in Brazil. According to Bobby, the patent expiring means developers, Linux distributions, multimedia projects, and device manufacturers can now support MPEG-4 Part 2 without paying the patent royalties previously required by the MPEG-4 visual licensing portfolio. That means we should now see Linux distributions with strict policies on patented multimedia technologies have less reason to restrict MPEG-4 Part 2 components or place them in separate repositories. Digital archivists will also be happy that large collections of videos from the 2000s encoded with DivX, Xvid, or related MPEG-4 Part 2 formats can now be supported without uncertainty about the former patent pool. Saroff points out that this doesn't mean every possible MPEG-4 visual patent anywhere is gone, just the ones in this specific pool. Other essential patents could exist outside it, but for the 30+ companies that did join, the licensing obligation is now over. I also want to recommend don't confuse MPEG-4 Part 2 Visual with the modern MP4 container format that can hold video compressed with various codecs. One codec that is formally known as MPEG-4 Part 10, also called AVC, or even more commonly known as H.264, still remains covered by its own active patent licensing program until November 2030.
Ken McDonald [00:25:01]:
So we've still got over 4 years on that one. But if you do want even more details about this patent expiration, Then check out the links in our show notes. I've got 3 of them for you to look at, read into.
Jonathan Bennett [00:25:14]:
Yeah, man. Software patents on audio and video files. They just, they have been a huge pain for Linux users for decades now.
Ken McDonald [00:25:25]:
They've been a bane to expansion.
Rob Campbell [00:25:28]:
Linux loves patent expirations.
Jeff Massie [00:25:30]:
Yes.
Jonathan Bennett [00:25:31]:
Yes, we do.
Ken McDonald [00:25:33]:
Make sure you put that expiration in there.
Jeff Massie [00:25:37]:
Yeah, actually, I would argue that a lot of times we shouldn't have software patents. It's just math.
Ken McDonald [00:25:42]:
It's—
Jonathan Bennett [00:25:43]:
yeah, it's, it's math on machines, right?
Ken McDonald [00:25:45]:
Yeah. Yeah. And you can almost argue, at least in the US, you're patenting free speech.
Jeff Massie [00:25:59]:
Yeah, only if you put it on a shirt.
Jonathan Bennett [00:26:04]:
That one was about a copyright violation and the DMCA, not necessarily patents. But yeah, it's a challenge. I would love to see software patents go away, but I don't think that's happening anytime soon. So the nice thing is that we do have some companies out there, particularly Google, and Google is one of the big ones, but they go out of their way to design design like audio and video codecs that don't use any of these existing patents. And then they will get their own patents on it, but then, you know, release them under some not so restrictive license that's actually compatible with the Linux desktop. And so that's, you know, I mean, that's how we can now download Chrome and Firefox and be able to play a good portion of the things on YouTube, for instance. It just works now.
Ken McDonald [00:26:54]:
AV1 is one of those, isn't it?
Jonathan Bennett [00:26:56]:
I believe so, yeah. And of course, you've had other, you've had other companies and places doing that for a long time. Like there's been OGG and FLAC for forever, and those are both great audio codecs.
Ken McDonald [00:27:06]:
That's my preferred one for when I archive my CD collection.
Jonathan Bennett [00:27:13]:
Yeah, FLAC is great for that, for lossless. So, but yeah, it's been, it's been an issue for, for years and years. Sometimes things work so well now, I kind of forget that that's still an issue, but it totally is. Yeah, every once in a while you run up against that. It's like, oh, they sent me this file. Why does this file not play? Or, you know, when you're running the door.
Ken McDonald [00:27:35]:
I got this new Blu-ray. Why won't it play in my old Blu-ray player?
Jonathan Bennett [00:27:38]:
Yeah, well, a lot of that is actually DMCA and not patents, but, you know, DRM.
Ken McDonald [00:27:44]:
A combination. Yeah. Because some of the newer Blu-rays are using newer codecs.
Jonathan Bennett [00:27:50]:
Well, that makes sense. Makes sense. All right. I guess we're getting back on the AI train, Rob. Is this going to become the new AI show? Is that where we're destined to?
Rob Campbell [00:28:01]:
I think everything is destined to become the new AI show.
Jonathan Bennett [00:28:07]:
So, all right.
Rob Campbell [00:28:09]:
You know, and Open source, open source itself, open source software has always been built around choice. Developers choose their tools that may be AI in many cases, their programming languages, their licenses, and the platforms where, where they're hosted. Some people prefer large commercial platforms while others intentionally choose smaller community-run alternatives. That competition is healthy and a good thing. Different platforms can have different values and developers can decide which ones best support the way they work. But lately, it seems like part of the open source community, you know, they're spending less time on asking how AI can be used responsibly as a tool to improve the ecosystem and more time looking for ways to keep it entirely out. I'm glad Mr. Torvalds at least iterated last week or whatever that the Linux kernel was not one of them.
Rob Campbell [00:29:16]:
So anyway, that brings us to the Codeberg story. Members of the nonprofit Git hosting platform have voted to ban projects that are mostly made up of AI-generated code. The proposal passed by a fairly wide margin, and Codeberg plans to update its terms to prohibit what it calls vibe-coded projects, or what everybody calls vibe-coded projects these days. So this is not necessarily a total ban on every use of AI. Projects may still be allowed to include smaller AI-assisted contributions when the overall work remains human-driven. Codeberg has also argued that large language models are fundamentally incompatible with responsible open source development and that they threaten the free software ecosystem. And that is where, that's kind of where they lose me. I mean, by nature, AI code can't be copyrighted anyway.
Rob Campbell [00:30:20]:
So it's kind of free anyway, maybe, maybe not anyway. Besides that, you know, there is plenty of bad generated code out there. There are abandoned projects, insecure projects, and people releasing software that they don't understand. But all of those problems existed before AI. There's been abandoned projects that were not AI, insecure projects that were not AI. Wow, I know I had many. insecure ones back in the day. I learned about SQL injections the hard way.
Rob Campbell [00:30:55]:
Anyway, as I have said many times on the show, the solution should be to judge the software by its quality, security, documentation, maintenance, and usefulness. Not that AI was involved. The person releasing the software should also be responsible for reviewing it and standing behind it regardless of which tools helped create it. GitHub has taken kind of the opposite approach. Instead of pretending AI is going away, uh, anytime soon, it continues adding AI tools for writing code, reviewing changes, fixing vulnerabilities, and helping developers manage projects. So I completely understand Codeberg's policy that includes exceptions and does not completely prohibit every use of AI, but That does not change the overall direction they're choosing to go. At a time when moving forward with new technology, I think it's critical. I have a hard time supporting this kind of anti-AI sentiment.
Rob Campbell [00:32:04]:
The answer is not, I don't think, is to push AI away, is to learn how to use it responsibly. and hold people accountable for the quality of the work they produce. Otherwise, we risk being the people who insist on sticking with their horse and plow while everyone else moved on to tractors. Codeberg is free to take any position they want, but it's not the right direction for me. And it gives me one more reason to stick with the systems like GitHub or GitLab or Anybody who's not gonna completely block my vibe codes from being on there.
Jonathan Bennett [00:32:40]:
Yeah, so I'm gonna play devil's advocate here for just a moment, and I'm gonna hone in on one particular thing that they talked about, and that is that one of their considerations is the price of SSDs and the fact that they only have the budget to host so many different projects. And You make the point, which I agree with, that a project should be, instead of judged on whether it was used, whether AI was used or not, it should be judged on its code quality. I agree with that in theory, but I will say that that's asking a lot of the people running CodeBerg for them to be the ones to go in and make that judgment on all of those vibe-coded projects that get submitted to them. And so like, there's, there's kind of a, uh, I don't know if it's a chicken and egg problem, but it's just a, it's, it's really sort of a problem of scale. But it, it, again, it comes down to the assumptions that we've made in sort of this open source. And even in the, you know, we'll, we'll host this thing online for free. Um, the assumptions that we've made in building these, they assume a certain amount of friction. That there's, it's, it's a certain amount of, it's difficult to do this thing, and that's going to keep the, the demand low.
Jonathan Bennett [00:33:59]:
And now that we've got AI, that friction is essentially gone, and the demand for all of these products have gone up. And the business case at Codebird apparently does not work with that increased demand. And so there's a lot of these sort of business calculations that, you know, GitHub has made it, GitLab has made it, Codebird has made it, and they change with AI being there. And, you know, they gotta respond to it differently, in a way that makes sense for them. And this is the way that Codeburg thinks made sense for, uh, for, for their site, for their service. Um, but yeah, not everybody's gonna like it.
Rob Campbell [00:34:36]:
I, I think there's, I think there's other ways around that. I mean, some could say AI tooling to review it. Um, you could say make it a freemium product. You know, if, if it's the scale of things, then only make it free for up to so much and then Charge people.
Ken McDonald [00:34:54]:
And you only want to make it free if you actually can copyright it yourself.
Rob Campbell [00:35:03]:
Or charge for AI projects.
Ken McDonald [00:35:07]:
There's—
Jeff Massie [00:35:07]:
then you run into how do you tell it apart?
Ken McDonald [00:35:10]:
And how do you identify that it's an AI project? Most—
Jeff Massie [00:35:15]:
but I get the friction thing because, and I don't think we covered it, or maybe you guys covered it last week, where Flathub was talking about 70% of the projects that they've had, that they've, they at least are saying they were AI contributions. Nobody ever looked at the code again. It's like somebody put something out there, said, oh yeah, it should be on Flathub. And then people walked away.
Rob Campbell [00:35:39]:
I have a very strong rebuttal to the data in that. And you should go check out last week's.
Jonathan Bennett [00:35:45]:
Rob had questions. You should go back and listen to last week's.
Rob Campbell [00:35:48]:
Yeah. I said that was a bunch of, BS.
Jonathan Bennett [00:35:53]:
There's actually one more really interesting thought that I have with this. And that is that there is a distinct advantage to not allowing AI-generated stuff on your repository. And that is that then your site is much more valuable for training AI.
Rob Campbell [00:36:13]:
I thought you were going to say that as soon as you said, I don't know how I knew you were going there.
Ken McDonald [00:36:18]:
Yeah.
Jonathan Bennett [00:36:19]:
This is something that's actually been in the news recently. It's like, you know, once we have sort of sucked all of the internet, all of the human-generated stuff off the internet into the AI models, like, where do you go from there to continue getting better? And if it's just the AI producing stuff, does that make it any better to retrain on that? And the answer, it seems, is no. I saw one story where some bookseller was like selling these old books, out of print in some cases books, but like, you know, before everything became digital. And it's like, train your AI on these. They're guaranteed not to be made by LLM because they were written before the year 2000.
Ken McDonald [00:36:59]:
And we'll have tons of misprints and typos in them.
Rob Campbell [00:37:03]:
You know, maybe in all honesty, you know, I, You know, I use AI to assist a lot of my coding now these days and projects I have up there. I've been doing a lot of that. So my stuff is not going to be suitable for CodeBerg. But in full transparency, I don't know if I've ever even heard of CodeBerg before. So I probably wasn't going to go there anyway.
Jonathan Bennett [00:37:29]:
That's interesting.
Ken McDonald [00:37:30]:
Probably because they didn't host any of your projects.
Jonathan Bennett [00:37:34]:
I have heard of CodeBerg and I've gone there. think once for a project where its code was actually on CodeBerg to go and look at. But yeah, it's definitely not one of the— it's distinctively behind GitHub and GitLab in popularity.
Rob Campbell [00:37:52]:
I may have downloaded a project from there before, but I just didn't really pay attention to it. It's like, well, here's some random hosting site.
Ken McDonald [00:38:03]:
But Jonathan, I think you touched on it, uh, with one of your earlier comments. It— I think it's literally the cost of having to host it all.
Jonathan Bennett [00:38:12]:
Yeah, that's— they said in their, in their, in their post about it, that's a big part of the, uh, big part of the calculation for them is just—
Ken McDonald [00:38:20]:
Probably 75 to 95% of the calculation.
Rob Campbell [00:38:25]:
I could accept that.
Jonathan Bennett [00:38:26]:
But yeah, yeah, possibly.
Rob Campbell [00:38:28]:
It's a reasonable reason.
Jeff Massie [00:38:31]:
Yeah, chip prices going up, SSDs and memory is only getting worse.
Rob Campbell [00:38:35]:
I thought you said they're going to come back down.
Jonathan Bennett [00:38:38]:
They eventually will.
Ken McDonald [00:38:38]:
Yeah, in 10 years.
Rob Campbell [00:38:40]:
This year.
Jeff Massie [00:38:41]:
Well, it kind of surprised me that they've taken off like they have. I didn't— I thought they were going to kind of level off because prices kind of leveled off, and then, then we hear that, uh, the next—
Jonathan Bennett [00:38:54]:
a lot of that stuff's Yeah, the next model got released and then everything goes back up again. That's how that seems to go.
Ken McDonald [00:39:01]:
And some of the memory and storage providers are sold out through 2027, I've heard. I mean, or is it 2028 now?
Jonathan Bennett [00:39:13]:
The fabs are definitely all booked out that far for sure.
Jeff Massie [00:39:17]:
Yeah, every fab that can make memories cranking full speed, And you can't build them that fast. And even if you could, one of the biggest things is the tooling. Say you get somebody like Elon Musk and just go, I'm going to throw money like never before to get this built and fab built in 6 months. Well, you still got to get on the list for machines from ASML and, you know, places like that.
Jonathan Bennett [00:39:42]:
I think actually the interesting thing about this example, Jeff, is that is 100% happened. And it is taking them a while to build all of that out.
Jeff Massie [00:39:51]:
Yeah.
Ken McDonald [00:39:51]:
They've got the shell.
Jeff Massie [00:39:53]:
Yeah.
Ken McDonald [00:39:54]:
Now we just gotta put everything in it.
Jonathan Bennett [00:39:56]:
They have absolutely thrown as much money as they can at that. And yet they are still buying NVIDIA hardware.
Jeff Massie [00:40:03]:
Yeah. I mean, you could, and there's bottlenecks in a lot of that fab equipment and you, I don't know what it is because I'm not involved with that like I used to be, at least for that equipment, but it's not uncommon for, for semiconductor equipment to have 6 months to a year lead time, if not more. And I could totally— I could totally— pure speculation on my part— I could see that if everybody's trying to get on this bandwagon, that it could be 2 years out before your order gets fulfilled.
Ken McDonald [00:40:36]:
How about for AMD cards versus Nvidia?
Jonathan Bennett [00:40:40]:
I mean, the demand is a little bit lower for AMD, so it's going to be a little bit sooner. But I think we're talking about like the backend stuff, not necessarily Yeah, we're—
Jeff Massie [00:40:49]:
I'm talking like photolithography tools, CVD and PVD tools, diffusion tools, uh, big iron testers.
Rob Campbell [00:40:57]:
That's where Elon needs to invest his money, in the places to make those things. He needs to make more of those places.
Jeff Massie [00:41:04]:
I mean, yeah, except the, the research involved with a lot of that stuff.
Rob Campbell [00:41:09]:
Have you ever played like a Satisfactory or Factorio? You just keep finding where the bottleneck is and you fill it in there until you're actually using all of the resources efficiently and it's going through the pipeline as fast as possible. Yep.
Ken McDonald [00:41:25]:
And then you run out of resources.
Jeff Massie [00:41:26]:
Just start that fab, Rob, and I'll help you run it.
Jonathan Bennett [00:41:28]:
Our RPMs rocket per minute. All right, we're going to take a quick break. And when we come back, we're going to talk about something a little more Linux desktop friendly. We'll be right back. Don't go anywhere.
Jeff Massie [00:41:41]:
So now Jonathan said this little more Linux friendly. But I'm going to be talking Wayland versus X11. That's probably one of the most unfriendly topics in the Linux world there.
Jonathan Bennett [00:41:57]:
It has become our generation's holy war.
Jeff Massie [00:42:00]:
Yes, yes, it has. But so looking at Phoronix, and this Phoronix article compares 3 Linux desktop environments on CacheOS using very high-end NVIDIA graphics hardware. The goal was to see whether the desktop environment itself has any meaningful impact on gaming and graphics performance. Now, the desktops tested were KDE Plasma 6.7.2 running on Wayland, GNOME Shell 50.3 running on Wayland, and XFCE 4.20 running on X11. Now, the testing was performed on a Razer Blade 18 laptop equipped with an NVIDIA GeForce RTX 5090. laptop GPU using NVIDIA's R610 Linux driver. And all desktops were tested using their default settings. So nobody, nobody got optimized.
Jeff Massie [00:42:52]:
Now, I should mention that Michael Larrable said that he had limited time with his laptop before he had to send it back. So that's why there's not a greater number of tests and there aren't other things that he tried. It was kind of like, here's what I can fit under the wire before I have to send this back.
Ken McDonald [00:43:09]:
So.
Rob Campbell [00:43:10]:
Yeah.
Jeff Massie [00:43:10]:
That's why it's a little limited. You know, he didn't try COSMIC or didn't try other desktop environments. So now the background for this article is in an earlier Phoronix comparison, which we covered, they examined KDE Plasma Wayland versus X11 sessions. And in those earlier tests, Wayland generally showed advantages for modern NVIDIA hardware, which Michael Arable said led to readers requesting a broader comparison across entirely different desktop environments. So this article is essentially a follow-up to answer that question. Now, one interesting detail is that XFCE was tested using X11, even though experimental Wayland support exists. Michael does mention that he attempted to use the Wayland version, but it only produced a black screen during the login. So the X11 session presented— Xfce in the comparison.
Jeff Massie [00:44:07]:
The first notable benchmark discussed is GravityMark, a demanding Vulkan graphical benchmark, and in this test GNOME Shell delivered the best results. It outperformed KDE Plasma, and both Wayland desktops were noticeably faster than Xfce running X11. The differences were large enough that the desktop environment clearly influenced performance Despite a powerful GPU. The trend continued in 3DMark Wildlife Extreme, another graphics-focused benchmark that is native to Linux. And this suggested that GNOME's graphics stack was extracting slightly more performance from the hardware in several synthetic workloads. Now, when you actually went to games, the gaps became smaller. Cyberpunk 2077 showed GNOME maintaining a slight advantage over KDE Plasma, slight advantage, meaning like, you know, 2, 3 frames per second. So nothing earth-shattering, but the difference was much narrower, narrower than the synthetic benchmarks.
Jeff Massie [00:45:10]:
XFCE generally remained in the last place, though the performance deficit was less dramatic than some of the previous Vulkan, you know, synthetic tests. In Hitman, GNOME and KDE performed very similar—
Ken McDonald [00:45:23]:
similarly.
Jeff Massie [00:45:24]:
However, frame time measurements revealed A useful detail. XFCE consistently showed higher frame times, which can translate into less smooth gameplay even when raw frame rates appear respectable. The smoother frame time behavior of KDE and GNOME is arguably more important than a small difference in the average FPS. A lot of times you'll see other reviewers arguing this where they focus more on, for example, the 1% frames versus the average frames per second, because those little stutters impact your perception of a game a lot more than just the average frame time. In OpenGL testing using unique superposition, the results shifted slightly. KDE Plasma managed to take a small lead over GNOME, GNOME Shell in this workload. The advantage was not huge, but it demonstrated that no single desktop environment dominated every benchmark. Even so, XFCE remained the slowest option in the comparison.
Jeff Massie [00:46:22]:
And the pattern appeared repeatedly throughout the article. GNOME and KDE traded small wins, you know, depending on the benchmark, and XFCE basically typically finished last. Basically, the overall results from the testing is that modern Wayland-based desktop environments appear to be the strongest choice for gaming and graphics workloads on modern NVIDIA hardware. Now, he didn't test AMD, but I bet it's pretty much the same. GNOME Shell frequently posted the highest scores, particularly in Vulkan-focused workloads. KDE Plasma remained very competitive and occasionally took the lead, especially, like I said, OpenGL. It seemed to favor KDE a little better. XFCE's lower resource usage did not translate into better gaming performance.
Jeff Massie [00:47:12]:
In fact, it produced the weakest results of the group. And that's kind of, uh, an argument that it was always in the past that, oh, it uses less resources, therefore this— these smaller, uh, desktops are going to be faster. This data shows that that's not the case in, in, in, uh, this specific example. It would have been nice if we could have had the Wayland results from XFCE, but it's still in experimental stage. So maybe in 6 months or a year, this test can be repeated and then we could get a better idea of how does a very small desktop environment compare to something like GNOME and KDE, though they're not as bloated as they were years ago.
Ken McDonald [00:47:56]:
Mm-hmm.
Jeff Massie [00:47:56]:
But I digress. For users of modern Linux distributions, the result reinforces a broader trend that has emerged over the last few years. Basically, Wayland has matured significantly. NVIDIA's Linux driver support has improved. And like I said, the lighter weight desktop environments automatically deliver better gaming performance is, it's no longer true, or at the very least, not always. I mean, there could be an exception out there, but as of right now, doesn't seem to be a thing. So on cutting-edge hardware, the more modern desktop stacks now offer now are providing the best overall gaming experience. So take a look at the article in the show notes for more details and all the benchmarks that I didn't cover.
Jeff Massie [00:48:42]:
And happy gaming.
Rob Campbell [00:48:44]:
So I'm going to maybe defend, I don't know about defend XFC a little bit here, but suggest a possible scenario for it. Now, this was cutting-edge hardware. Now, I think a system, if you have the RAM, A system that can use it is going to be more performant, which is what you kind of saw here. A system that's more performant. But I wonder how they would compare on an underpowered system, say a 4 gig laptop, you know, 4 gigs of memory or less or more, you know, something much more restrictive with memory, how they would compare if the lighter weight would then excel under that scenario. Because a lot of the people I see online defending this, they're installing Linux on an old system. And that's part of their reason for using a lightweight desktop.
Jonathan Bennett [00:49:33]:
How does it perform on a potato?
Ken McDonald [00:49:35]:
Yeah.
Jeff Massie [00:49:36]:
Well, the thing is though, see, KDE and GNOME aren't, especially KDE always had the, previously had a reputation of being a resource hog. It hasn't for a long time. And there's times when it comes in pretty close to some of these lighter desktops.
Rob Campbell [00:49:51]:
Yeah.
Jeff Massie [00:49:51]:
Maybe slightly bigger, but I mean, it's much more potato compatible now.
Rob Campbell [00:49:59]:
It would be an interesting test, I think.
Jeff Massie [00:50:03]:
Oh yeah.
Ken McDonald [00:50:03]:
What I would like to see is a test with an older video graphics device.
Rob Campbell [00:50:12]:
Why don't you just do it, Ken?
Jeff Massie [00:50:13]:
You got it.
Jonathan Bennett [00:50:14]:
Yeah.
Jeff Massie [00:50:15]:
You got the hardware, Ken. Why don't you run that test and report back? it back to us because you can download the OpenBenchmark stuff that he ran. You can do the exact same tests that he ran and you can include your results in with what Michael had so you can directly compare.
Jonathan Bennett [00:50:36]:
That'd be very interesting.
Ken McDonald [00:50:37]:
Yeah, older hardware and Ubuntu Studio to CacheOS.
Jonathan Bennett [00:50:44]:
Yeah, it wouldn't be apples to apples. You would be changing multiple things at the same time. What would be really interesting is to do an install of Caché and compare it there. But maybe that's—
Ken McDonald [00:50:55]:
Maybe wipe out that older Ubuntu 25.10 install that I've got. There you go.
Jonathan Bennett [00:51:01]:
Yeah, that'd be interesting to see.
Rob Campbell [00:51:02]:
That's in the life now, right?
Jonathan Bennett [00:51:06]:
It probably is. It's beginning to be a little creaky. All right. Yeah, very interesting to see that. There is, There's something else that's not end of life. In fact, it's going to be around for quite a while. And that's Firefox 153. And Ken has this story.
Jonathan Bennett [00:51:24]:
153 is the extended support release, isn't it?
Ken McDonald [00:51:27]:
Yes, it is.
Jonathan Bennett [00:51:29]:
What is new there?
Ken McDonald [00:51:30]:
In fact, George Whitaker even talks about that in his article. But let me go ahead and give credit where credit's due. The articles I'm pulling from were written by Bobby Borisov. Michael Larabelle, Marcus Nestor, and George Riddiker. And of course, we're talking about them writing about Mozilla Firefox 153. And all of them agree that it's bringing improvements in video playback, PDF editing, sharing, privacy indicators, Firefox Labs, something I haven't really played with much myself, and web platform features for developers. In fact, Bobby thought one improvement was even worth writing a separate article about. That feature, according to Bobby, is native containers that Mozilla introduced in Firefox 153.
Ken McDonald [00:52:24]:
Containers allow users to divide their browsing into separate spaces such as personal work, banking, personal or work, banking and/or shopping. Now each container keeps its own cookies and login sessions even though all of them are open inside the same Firefox window. Now I'm assuming that he— since he doesn't mention it, but they're in different tabs with the contain— each tab representing a different container. Now Michael found the Vulkan video decoding support interesting. In fact, he says there is initial Vulkan video decode support for Firefox 153. According to Marcus, Firefox 153 rolls out the beta version of the Smart Window feature, feature that lets you see and choose AI models. I tried keeping this one out directly from the new Smart Window assistant with a familiar address bar for typing websites and searching the web. It also introduces support for sharing any open page with a QR code and support for warning users whenever a website has access to their location by highlighting the location permission icon in red, of all things.
Ken McDonald [00:53:46]:
Now, according to George, Mozilla continues expanding Firefox's built-in PDF editor, editor, by introducing the ability to merge multiple PDF documents, insert images as new PDF pages, and continue using existing editing tools such as annotations, page organization, and text editing. Now, they all wrote about several enhancements designed to give users more visibility and control over website permissions, to include local area network restrictions enabled by default for all users, more restrictive default permissions for browser extensions accessing local files, and a visual indicator when a website is actively accessing your location. Now, I have only touched on some of the many improvements that will impact Linux users. Of course, you can find more details about all the improvements that impact Linux and other OS users by reading the articles written by Bobby, Michael, Marcus, and George. Jonathan or Rob, can either one of y'all name one improvement that I did not touch on?
Jonathan Bennett [00:55:05]:
Oh, I'm sure I could if I went and looked. One that stuck out to me was the HDR for Windows. And I had to do a double take because I thought that Firefox got HDR support a long time ago. And Firefox did get HDR support for Linux like a year ago in like release 130-something, 137 or 138. So it's one of those weird times where Linux got support for the cool new toys way before Windows did.
Rob Campbell [00:55:32]:
Another one. We had a story like that a few weeks ago.
Ken McDonald [00:55:35]:
Mm-hmm.
Rob Campbell [00:55:37]:
Yeah, the only one I noticed is the number's bigger.
Jonathan Bennett [00:55:42]:
Firefox++.
Rob Campbell [00:55:43]:
Bigger is better.
Jonathan Bennett [00:55:44]:
Yes.
Ken McDonald [00:55:45]:
I think Mark has even mentioned about the 154 beta coming out soon.
Jonathan Bennett [00:55:52]:
Yeah, that's—
Ken McDonald [00:55:52]:
If you want to really live on the edge.
Jonathan Bennett [00:55:55]:
They push that. And, you know, Google Chrome does the exact same thing. It's pretty much always one version for the release, the next version for the beta, and then the next version after that is alpha. And they all sort of march along at the same speed.
Rob Campbell [00:56:08]:
So—
Jeff Massie [00:56:08]:
I can tell you I tested 153. And for me personally, I still can't get it to work with OBS virtual camera and Restream.
Jonathan Bennett [00:56:18]:
Just not there yet. Yeah, one of these days, one of these days we'll get there.
Ken McDonald [00:56:24]:
Well, I am actually coming to you in Firefox 153.
Jonathan Bennett [00:56:28]:
Well, there you go. Apparently it works for him.
Ken McDonald [00:56:31]:
Yeah.
Jeff Massie [00:56:33]:
And I've tried all the fixes and the turn this on and reset that. It just doesn't work for me.
Rob Campbell [00:56:38]:
Have you tried AI?
Ken McDonald [00:56:41]:
In Firefox?
Jeff Massie [00:56:41]:
I bet you have.
Jonathan Bennett [00:56:43]:
Hey, Claude, help me fix this.
Ken McDonald [00:56:45]:
Yeah.
Jeff Massie [00:56:46]:
And it gave me a bunch of stuff to try and I tried it and it didn't help.
Jonathan Bennett [00:56:52]:
You need to give it some way to agentically try the things itself and then just like leave it running overnight. Like, all right, Claude, I'm going to bed. Make sure it's fixed by the morning. Don't make any mistakes.
Rob Campbell [00:57:02]:
Or instead of having it just try things, ask, is there something I can test or a command I can run and then feed you to help you diagnose this problem?
Ken McDonald [00:57:12]:
Jeff, I'm going to recommend a site for you to check out after the show, openrouter.ai.
Jonathan Bennett [00:57:22]:
Oh, this sounds interesting.
Ken McDonald [00:57:23]:
I think I picked that one up from one of the Intelligent Machine episodes. Interesting.
Jonathan Bennett [00:57:31]:
Okay, so we're going to take a quick break. We're going to talk about OBS a little bit here later. We're going to take a quick break and then Rob's going to come back and Another enterprise sort of story, Rob. Enterprise and Ubuntu. That's just your theme for today, I guess. But he's going to talk about Certified Ubuntu right after this.
Rob Campbell [00:57:49]:
All right. So I don't know if this— I want to say this is enterprisey except for it's involving Ubuntu. I think this one can fit for anybody. So, you know, whenever I hear somebody say they tried Linux, And nothing worked or things didn't work. I'm always curious about what hardware they are using. Quite often, you know, it turns out they're trying to use some obscure wireless adapter, fingerprint reader, docking station, some other device that was designed specifically for a different OS. And, you know, the manufacturer provides Windows drivers and gives little or no attention To Linux. And then Linux gets blamed when the hardware doesn't work correctly.
Rob Campbell [00:58:37]:
Linux isn't— I don't think Linux is the one to blame in these situations. You know, I'd say these are failures of the manufacturer. Linux supports an incredible amount of hardware, especially considering that developers sometimes have to build that support on their own without help from the companies, the manufacturers that made it. But no operating system can guarantee a good experience on every random combination of components someone puts together. Try running Windows on a system designed for macOS, or even worse, running macOS on a system designed for Windows. So why do we shift the blame to Linux when something designed for a different OS doesn't work on Linux? Well, One of the best ways to avoid these problems is to choose hardware that has actually been tested and confirmed to work with Linux. And Ubuntu is about to make that much easier to identify. Ubuntu 26.10 is adding a new option under Settings, System, and About that will tell users whether their computer is Ubuntu certified.
Rob Campbell [00:59:47]:
It can also show whether the computer was certified for another Ubuntu version, whether it is similar to a certified model, or whether it is just not certified at all. Ubuntu's certification is more than just checking whether the computer boots. Canonical says it performs more than 500 hardware compatibility tests covering areas such as audio, networking, Bluetooth, USB, firmware, power management, memory, and processor. Certified systems also continue to be tested as Ubuntu receives updates. Of course, certified hardware may— uncertified hardware may still work perfectly fine. Certification is not a requirement for running Ubuntu, and it does not mean every uncertified computer is a bad choice. But it does give users a much clearer idea of whether the hardware manufacturer and Canonical have made an effort to ensure everything works properly. Rather than buying whatever computer happens to be available and blaming Linux when one unsupported component causes trouble, look for hardware that was designed, tested, and certified to run.
Rob Campbell [01:00:56]:
Ubuntu put that information directly into settings. It's a small change, but it could help users avoid a lot of unnecessary frustration and perhaps prevent Linux from being blamed for another manufacturer's lack of support. I think the next step is to have a certified Ubuntu badge that manufacturers can put on their products to make it even easier for consumers to choose when looking to buy parts.
Jonathan Bennett [01:01:28]:
Yeah, sounds like a plan. I've never gone out of my way. Well, I guess maybe once or twice, but not often have I gone out of my way here recently to buy like any of the certified equipment because almost everything just works. at this point.
Rob Campbell [01:01:44]:
Yeah, but there is definitely older stuff, or like the really cheap, uh, really cheap stuff that you get on the Walmart shelf that, uh, I guess, I guess I just have another problem for people.
Jonathan Bennett [01:01:59]:
I just have another rule that I don't buy computers from Walmart, and that keeps me out of trouble.
Jeff Massie [01:02:03]:
Or Temu.
Rob Campbell [01:02:04]:
Yeah, I, I know, I know Wi-Fi, not from personal experience, but I know Wi-Fi Wi-Fi cards are a very common problem for a lot of people.
Jonathan Bennett [01:02:12]:
Yeah, that's true.
Ken McDonald [01:02:13]:
That's fair.
Jeff Massie [01:02:16]:
But yeah, it's like when you go, hey, look, I can get this card for a nickel. Yeah, it might not work.
Ken McDonald [01:02:21]:
But if you're building your own system, is it going to be certified anyways?
Jonathan Bennett [01:02:26]:
You could pick the same parts that are certified.
Ken McDonald [01:02:28]:
Yeah.
Rob Campbell [01:02:28]:
Yeah.
Jeff Massie [01:02:30]:
You could have the motherboard and whatnot that—
Ken McDonald [01:02:33]:
Motherboard, hard drive, uh, Memory, hard drives.
Jeff Massie [01:02:38]:
A lot of that's going to be— it's like hard drive, CPU, memory, it's all going to be in there. I mean, unless you're getting something weird, yeah, uh, CPU, it's all going to be certified because a lot of those go to a very exact standard. It's going to be more stuff that's— that needs a specific driver, you know. It's going to be more like your mouse, your printer, your— I was going to say your keyboard, but even a lot of that now is pretty standardized.
Jonathan Bennett [01:03:04]:
Yeah, your keyboard is probably going to work because it's probably HID, just the human interface device, but you may not be able to get your LEDs to turn on. Uh, the stuff I've had problems with over the years have been motherboards, and a lot of times they will come up but just not all of the equipment on there. Like, uh, one motherboard in particular had, um, it was NVIDIA branded, like the— I figured it was the north bridge or the south bridge, whatever gave you like USB and Ethernet. And the Southbridge, I do believe. Yes. The NVIDIA Southbridge just did not play well with Linux. And then I have had trouble with Wi-Fi drivers over the years. Although those pretty much work these days too.
Jeff Massie [01:03:46]:
Well, I was going to say that tells you the age of the hardware. So people that don't know, we haven't had Southbridges in computers for what, a decade now?
Jonathan Bennett [01:03:54]:
Yeah, probably. Everything is essentially a system on a chip.
Rob Campbell [01:03:59]:
No. Yep.
Jeff Massie [01:04:01]:
That's it.
Jonathan Bennett [01:04:01]:
Switch all SSDs. Absolutely.
Ken McDonald [01:04:05]:
All right.
Jonathan Bennett [01:04:05]:
Let's move on and let's talk OBS. Jeff, are you running 32.2 of OBS yet?
Jeff Massie [01:04:12]:
I am.
Jonathan Bennett [01:04:13]:
Did it break? Did it break on you when you updated?
Jeff Massie [01:04:17]:
I, oh, I'm sorry. I do not. I have 32.1.2. So I have not been rolled to the new one yet. Okay.
Ken McDonald [01:04:26]:
And I'm at 32.1.
Jeff Massie [01:04:29]:
So we will see what happens there. So OBS Studio 32.2 has been released as the second maintenance update in the OBS Studio 32 series. And OBS, for those unfamiliar with it, which probably would be somebody new in our audience because we talk about it quite often, is, you know, it's one of the most widely used open-source applications for video recording, screen capture, and live streaming. Now this release provides a little more than— is, is arrives a little more than 4 months after version 32.1 and focuses on a combination of new functionality, workflow improvements, platform support updates, and bug fixes. A headline feature in OBS is a new filter that can compose SDR or standard dynamic range content into an HDR or high dynamic range workflow. Workflow. HDR is becoming increasingly important, important as more displays, cameras, and streaming platforms support wider color ranges and higher brightness levels. In practical terms, this feature helps users integrate older SDR sources into projects that are being produced or streamed in HDR without requiring an entirely different HDR-native production chain.
Jeff Massie [01:05:43]:
Another significant addition is dynamic bitrate support for multi-track video. So multitrack video is a relatively new OBS capability intended to provide greater flexibility when sending video streams to supported services. This allows you to capture different camera feeds, you know, clean gameplay, then maybe a webcam layout simultaneously, and then swap between them in post-production. So you can also— and you can also record up to 6 independent audio tracks in the same recording. So it gives you a lot of flexibility if you're, say, streaming a game and you might have some friends with you, or you're talking to the camera more. You can dynamically switch even though you've already recorded this and maybe it's not a live stream. Dynamic bitrate support allows OBS to better adapt to changing network conditions, helping maintain stream quality and stability. Excuse me.
Jeff Massie [01:06:40]:
Alongside that, OBS 32.2 adds missing file— missing file support in filters, allows plugins to define custom icons for new source types, and adds new copy and paste functionality to the front-end API. The user interface has also received several updates. The traditional add source dropdown menu has been replaced with a new dialog designed to improve source selection and organization. Users of the image slideshow source can now include WebP image files when adding directories, and there are numerous under-the-hood improvements aimed at performance and reliability. OBS developers report improved OpenGL performance on low-end systems, and the release also enhances audio mixer behavior, updates descriptions related to multitrack video, And Linux users running NVIDIA graphics cards should see improved Pipewire support through squashing a bug. Maybe that's what I need. So take a look at the article in the show notes for full details and a link to the GitHub page, which lists out all the changes made, which many of them I didn't cover today. So happy streaming.
Jonathan Bennett [01:07:52]:
Yeah.
Jeff Massie [01:07:52]:
And I will apologize for my voice breaking up a little bit. I'm, I'm up in Idaho and we got a A lot of smoke up here. It's really thick. There's a lot of fires in like Oregon and a little bit from Canada.
Jonathan Bennett [01:08:03]:
So, and here I thought you were just getting, getting emotional talking about OBS.
Jeff Massie [01:08:08]:
Well, I, well, I was trying to save face, Jonathan.
Jonathan Bennett [01:08:10]:
Just come on.
Jeff Massie [01:08:11]:
Sorry.
Ken McDonald [01:08:12]:
It's the smoke.
Rob Campbell [01:08:13]:
This, I'm not sad.
Jonathan Bennett [01:08:15]:
Sometimes smoke gets me.
Ken McDonald [01:08:16]:
I thought you were going to say it's because you'd been smoking pork.
Jonathan Bennett [01:08:19]:
Yeah. Well, that could be too.
Jeff Massie [01:08:22]:
Not, not this week. But I have the smoker out.
Jonathan Bennett [01:08:26]:
Hmm.
Jeff Massie [01:08:27]:
I did meatloaf. I did meatloaf, uh, just a few days ago.
Rob Campbell [01:08:31]:
Smoker's not always out?
Jeff Massie [01:08:34]:
Uh, sometimes I put it undercover and I don't leave it sitting out ready all the time.
Ken McDonald [01:08:40]:
Jonathan, with OBS Studio's latest version, which do you think you're gonna be using better, the option for adding sources Or the SDR to HDR conversion?
Jonathan Bennett [01:08:52]:
I will probably not be using the SDR HDR conversions because the computer that I run OBS on is not HDR capable and I don't have any HDR capable video equipment. So that one's, that one's not terribly useful for me at this point.
Ken McDonald [01:09:10]:
Well, do you need HDR capable video capturing equipment? Or would it be able to convert it so that, say for example, uploading to YouTube?
Jonathan Bennett [01:09:24]:
See, I am not at all sold on the idea that I could take an SDR capture and somehow magically upscale it to HDR. That sounds terrible.
Rob Campbell [01:09:37]:
John, isn't that what they did with The Wizard of Oz?
Jonathan Bennett [01:09:41]:
Like I said, it sounds terrible. That reminds me of when 3D video was the fad and they took all of those these movies that had been recorded and they added the dorky 3D effects. It's the exact same thing. Uh, no, no, that's, that's a terrible idea.
Ken McDonald [01:09:56]:
Well, you could—
Jeff Massie [01:09:57]:
so you could do it, you could convert it to HDR, but that's the HDR format, but you're not gaining any actual information, right? Because all the actual data was recorded in SDR. So it's kind of like the old, uh, going from an old analog tape to a CD. You can make that conversion, but the original information is still in that analog format, and it's not improving the quality just because you put it on a better platform, higher resolution platform.
Jonathan Bennett [01:10:32]:
Let me tell you something that I've seen on social media here in the last few days that is hilarious. seriously applicable to this. So Make Magazine, which I really like Make Magazine, they've featured some of my stuff before. They have advertisements now on Facebook and it is a magazine cover and the big text looks normal, but when you zoom into it, the small, the fine print text is obviously AI generated. Like it's got terrible, you can't read any of it. It looks like—
Jeff Massie [01:11:06]:
Yeah.
Jonathan Bennett [01:11:07]:
you know, symbols from a dream sort of deal. And, you know, people were roasting them from like, why can't you— you're a magazine, why can't you do a real cover? What's wrong with you? And I looked at that and I went, no, no, this is a real Make magazine cover. I have a physical copy of this one because it's the one that covered our stuff. I have a copy of this magazine still. In fact, it's right up there. I have it still in the plastic wrap because I bought like 3 of them and one I didn't open. And it's like, what's going on? And then I realized, oh, They had a low-resolution picture and they used AI to upscale it. And the AI didn't know what the text said.
Jonathan Bennett [01:11:42]:
And so it just made it up. And it's like, this is the problem with AI upscaling. And you would run into the exact same thing trying to do like AI HDR upscaling. I was thinking this is one of the ways you do it.
Ken McDonald [01:11:54]:
You didn't run it through an OCR scanner.
Jonathan Bennett [01:11:56]:
I guess.
Ken McDonald [01:11:56]:
Yeah.
Jeff Massie [01:11:57]:
Well, you got to keep in mind, I mean, HDR is kind of a different format than SDR, but it doesn't mean it's not upscaling necessarily. It's, it kind of relates to that analog to CD analogy I use.
Jonathan Bennett [01:12:10]:
It's upscaling, it's upscaling the bit depth as opposed to the resolution.
Jeff Massie [01:12:16]:
And so you're gonna get more bits of information.
Jonathan Bennett [01:12:21]:
Well, there, there is, it just doesn't have any real bits to pull from. So you're, you would, a simple, a simple mapping, you would take your, what is it, 8 bits of SDR and you would map it out to the 12 bits of HDR. It's something like that. And so you would just have a curve that you would do this mapping, and you're right, you wouldn't gain any more information.
Rob Campbell [01:12:40]:
Well, so, so I, I, I like the SDR. If you had a solid blue screen, you're still— if you upscaled that to HDR, it would still just be that same blue screen. You'd have room for those more colors, but you don't have those colors there.
Jonathan Bennett [01:12:56]:
You have room for brighter colors is essentially what it is.
Jeff Massie [01:12:59]:
Yeah.
Ken McDonald [01:12:59]:
Yeah.
Jonathan Bennett [01:13:00]:
And so all of this to say, my thought is if you were to do an AI upscale, a bit depth upscale of HDR, would it like hallucinate extra details to put into the shadows? Would it hallucinate parts of your scene that it thinks should be brighter? Like there's some real potential for hilarity here and I kind of want to see it.
Rob Campbell [01:13:24]:
It's not AI though, is it?
Jonathan Bennett [01:13:26]:
This is probably not. I was just trying to imagine, like, what would it look like to do that?
Jeff Massie [01:13:32]:
Yeah, this filter is just putting that 8-bit information into that 12-bit or 16-bit format, whatever it is. It's not adding anything. It's just—
Rob Campbell [01:13:43]:
It's like taking my $5 bill and putting it into a wallet that will hold $1 million. Still only $5 in there.
Jeff Massie [01:13:50]:
Yeah, there's still only $5 in there.
Jonathan Bennett [01:13:51]:
in there.
Jeff Massie [01:13:52]:
It's only as good as the source material.
Jonathan Bennett [01:13:56]:
Oh my goodness, we have the worst analogies.
Ken McDonald [01:13:58]:
Here is a project, if you wish to accept it. Take a black and white video that you have and ask AI to convert it to color.
Jonathan Bennett [01:14:13]:
I've seen that done before, actually. Sometimes that's not terrible, particularly with like movie clips. It's pretty interesting to see.
Ken McDonald [01:14:19]:
It depends on if you have a reference, uh, reference.
Jonathan Bennett [01:14:24]:
Yes.
Jeff Massie [01:14:24]:
Yeah. Well, in the old colorized movies used to look terrible. They look like somebody filled them in with crayons, but they've gotten a lot better now. I mean, it— I think, I think some of you, well, you probably all remember the old— was it Turner that went back and was colorizing a bunch of old black and white classics? And some of them looked—
Jonathan Bennett [01:14:43]:
the colors were just weird.
Ken McDonald [01:14:44]:
Oh, you mean the ones that looked like cartoons afterwards? Yeah. Yeah.
Jeff Massie [01:14:49]:
I mean, this was pre-AI, pre— I mean, this was, you know, 15 years ago, 20 years ago.
Ken McDonald [01:14:54]:
When you were paying creators to go in and colorize it by hand.
Jonathan Bennett [01:15:02]:
All right. I'm not going to make the joke that comes to mind. I'll tell you guys after the show. Ken is going to tell us about something new from Raspberry Pi. What is— what are they cooking?
Ken McDonald [01:15:14]:
Well, they're cooking a display just for you, though you may find this display too big. Though Bobby Borisov, Michael Larabelle, Giorgio Mendoza, and Marcus Nestor all wrote about sometimes bigger is better. Now, as Jonathan stated, we are talking about the latest Raspberry Pi touch display called the 10-inch Raspberry Pi Touch Display 2. It's priced at $80 before shipping, regional taxes, and if applicable, tariffs. As the name implies, it is a 10-inch display for the Raspberry Pi 5 and the Raspberry Pi Compute Module I/O boards. This is the latest ones. Now this variant features an IPS TFT true multi-touch capacitive panel that supports 10-finger touch, 24-bit RGB, and a 1200 by 1920 pixel resolution.
Rob Campbell [01:16:20]:
Thank you.
Ken McDonald [01:16:21]:
Yeah, for a minute there. It also features a more generous 85-degree viewing angle and comes with all the necessary cables, connectors, and mounting hardware. The display receives power directly from the connected Raspberry Pi, and requires 2 connectors, a GPIO power cable, and a 22-way mini-to-mini ribbon cable for the DSI interface. It also requires a 4-lane MIPI DSI connection to support its higher resolution. The capacitive touchscreen is supported out of the box by Raspberry Pi OS. And requires no manual calibration. Once connected, it should work immediately with the operating system's existing display and touchscreen drivers. Since it requires the full 4-lane DSI interface, it is not compatible with the Raspberry Pi 4, earlier Raspberry Pi boards, the Raspberry Pi Zero series, or the older Compute Module I/O boards.
Ken McDonald [01:17:32]:
As always, more information is available from Bobby, Michael, Giorgio, and Marcus's articles. And Jonathan, I'm assuming you don't even need— have a need for buying this.
Jonathan Bennett [01:17:46]:
No, see, I might pick one of these up. So like, I don't like the big 10-inch portable tablets, right? I like, I like the 7-inch tablet. But this is not a portable tablet. This is something you sit on a desk or hang on a wall. And there, bigger might be better. So I, and I like the higher resolution too.
Rob Campbell [01:18:05]:
You know what I like to see for Raspberry Pi that would go well with something like this?
Jonathan Bennett [01:18:10]:
What's that?
Rob Campbell [01:18:11]:
A thinner board. So instead of the big HDMI, the micro HDMI, and you probably have to get rid of the Ethernet, but whatever you can do to make it thinner so it— so you can mount that on the back without sticking out as far. Because I have had a, uh, had a display in the past and had a case and it just stuck out far.
Ken McDonald [01:18:31]:
And one of the articles, I want to say Giorgio's, actually talks about, uh, having the, uh, display panel, then, uh, M.2, uh, HAT, then the, uh, Raspberry Pi underneath.
Rob Campbell [01:18:47]:
Yeah, but if you're gonna have that behind, if I'm gonna mount that on the wall, I'm either gonna have to cut a hole in the wall to recess that, or it's gonna be sticking out.
Ken McDonald [01:18:54]:
Put it all in a case that you can mount on the wall.
Rob Campbell [01:18:57]:
Well, then it's gonna be sticking out like, uh, probably 6 inches. I want it to be almost as flush against the wall as possible, not sticking way out.
Jonathan Bennett [01:19:08]:
I will tell you that the Raspberry Pi fits very nicely inside a gang box. 3-gang box.
Rob Campbell [01:19:17]:
Yeah, 3-gang.
Jonathan Bennett [01:19:18]:
That's the, that's the size where it fits really well. And once upon a time there, and I'm sure there still are, some 3D prints where, you know, you, you have a sort of a collar that you screw onto that box and then a second piece that goes onto the Pi and you just hang it. And, uh, it was a, it was a fair— it was a fairly nice deal once you got it all put together.
Ken McDonald [01:19:38]:
Let's go back. You're basically having it go into the wall?
Jonathan Bennett [01:19:41]:
Yeah, so the Pi itself sits inside the electrical box inside the wall, and then the screen can flush mount. It actually, it ended up looking pretty nice.
Rob Campbell [01:19:50]:
That would be acceptable, except for I still, you still have to cut in the wall and put a gang box there if you don't have it, which right now for like my Home Assistant, I just have an old Samsung tablet on the wall. And I did drill a hole for the power to go through, which is just powered in the other room.
Jonathan Bennett [01:20:07]:
Just the cable hanging out the wall and going down and plug it in.
Rob Campbell [01:20:10]:
No one goes in the other room, so it doesn't matter.
Ken McDonald [01:20:12]:
Yeah.
Jonathan Bennett [01:20:14]:
It's the kids' room.
Jeff Massie [01:20:14]:
Nobody sees that.
Ken McDonald [01:20:15]:
It doesn't matter until the power goes out in that room for sure.
Rob Campbell [01:20:18]:
I can hide it better.
Ken McDonald [01:20:19]:
Yeah.
Jonathan Bennett [01:20:20]:
Well, it's a tablet. It's got built-in battery backup, right?
Rob Campbell [01:20:22]:
It does. Yep.
Jeff Massie [01:20:23]:
There you go.
Rob Campbell [01:20:24]:
Power going out, it's not a problem.
Jonathan Bennett [01:20:25]:
Yeah, for a while.
Jeff Massie [01:20:27]:
Little battery backup.
Rob Campbell [01:20:28]:
Little rhubarb pie.
Ken McDonald [01:20:31]:
Blueberry pie?
Jeff Massie [01:20:32]:
Rhubarb.
Jonathan Bennett [01:20:35]:
All right. So that is it for the stories. We're going to end on that absolute high note of rhubarb pie. And we're going to come back after just a minute. We've got some command line tips. They're all command line tips this time, right? We have Linux tips for you. I have been informed that we do more than just command line tips here. But anyway, we'll be right back.
Rob Campbell [01:20:56]:
We'll have some tips. Many times.
Jonathan Bennett [01:20:57]:
Yeah, many times. They try to tell me this all the time. Anyway, Don't go anywhere. I'll be right back.
Rob Campbell [01:21:03]:
All right. My tip for today is Keyword. Yes, this is a command line tip. It is a TUI. So I'm not showing you this on a production because I don't want you to see my SSH production SSH keys. So the data is not going to be great. Not going to be a whole lot of data for you to see. But Keyword is a command line TUI application to manage your SSH keys and many other aspects of that.
Rob Campbell [01:21:30]:
So those watching, you have a tab for your SSH keys where if I had any on here, I could manage those. I could import them right now if I wanted to, but I'm not doing that right now. There is an audit where it can give you a grade on your the security of your SSH setup. Not a whole lot going on here, so it doesn't say a whole lot. It will tell you the— show you the SSH config, which I should have at least set SSH up for you. Yeah, I didn't get to that. A tab where you can generate an SSH key. Another tab that'll show you known hosts.
Rob Campbell [01:22:18]:
Backup and restore stuff. And there's a setting. So there's a password to get into this that you set to get into this application in the first place and the vault directory for it. But a simple application, 2-way application to manage your SSH keys on your system.
Ken McDonald [01:22:35]:
That's it.
Rob Campbell [01:22:36]:
Very cool. Nice and simple, short, unlike whatever Ken's going to bring us.
Jonathan Bennett [01:22:43]:
Well, it's not time for Ken yet. It's time for Jeff's tip. And Jeff, what is UVX?
Jeff Massie [01:22:50]:
Well, it's a command line tip and it's not even the TUI. So I'm not even cheating that way. This is pure command line, just back the old days. So UVX is, you know, let me back up a little bit here. So a couple of weeks ago, I covered a command line tip called whichllm. Now the purpose of it was to show based on what hardware you have on your system, what AI model best Fits. UVX comes from that command, from that command line tip, because those of you who ran that command probably ran it with the UVX command. So what is UVX? UVX is shorthand for uv tool run, and that executes Python command line tools in temporary isolated environments without permanently installing them.
Jeff Massie [01:23:40]:
It's similar to pipe x run but leverages uv's Rust-based resolver for dramatically faster execution. Each invocation creates a very short-lasting virtual environment, installs the requested tool and its dependencies, runs the command, and then discards it all. So this ensures that tools don't conflict with each other or with project dependencies. uvx is part of the UV Python package manager ecosystem and is useful for running one-off tools like linters, formatters, and code generators without cluttering the system with permanent installations. Now I'm not going to go into the options because unless you really need to get into the weeds, just typing uvx space, you know, program, and then maybe if there's any program options, that's enough. The uvx options Or if you're really trying to test different versions of libraries or segregate other parts of the system or code when you're trying to run your program, basically, unless you're really kind of trying to do developer stuff, you're not going to use any options for UVX. So take a look at the link in the show notes for more details and happy coding.
Ken McDonald [01:24:56]:
Very cool.
Jonathan Bennett [01:24:57]:
I will have to use that one. That's actually really slick. All right, Ken, what do you have for us?
Ken McDonald [01:25:03]:
Well, this week I've got a way that you can script your mouse and keyboard inputs so that you can automate them with that script. This week I'm introducing YDOTool. It is a command line device input that lets you programmatically or manually simulate keyboard input and mouse activity or any other human interface device. Now, do want to preface this that the YDO Tool D daemon must be running. Let me go ahead and bring up my screen. Now, behind me you'll see that I've got, uh, a console and Kate, uh, text editor up. Now, this is in, uh, OpenSUSE Tumbleweed. Now you can get more helpful information about the command ydo-tool by following it with --help.
Ken McDonald [01:26:06]:
And that tells you how you use it. Basically, you've got subcommands that you can add to it. So you put ydo-tool followed by the command in question and then the arguments that you'd use for that command. And you also would— could you specify an environment variable, YDO_TOOL_SOCKET, to specify the daemon socket? I've already got that set up in openSUSE Tumbleweed, so I can go to this command, which actually demonstrates several of the commands that you see listed. The first one's going to be YDO mouse move followed with a -X to indicate the X orientation, and I'm telling it to move the mouse in the X orientation minus 99,999, and then the same for the Y orientation. After it does that, I didn't tell it to move them to do a mouse move relative from that. to 800, uh, by 200. Then I follow it with a YDO tool click, 0xC0.
Ken McDonald [01:27:28]:
That's going to indicate that I'm using the left mouse button. The C tells it to down and then back up. Then I'm demonstrating the YDO tool type. by following that with in quotes, hello world exclamation. Now you can do it without it being in quotes and then type exactly what you have on there. But I did that and it took the space out. In fact, I'll demonstrate that here in a minute. But what I'm going to do now Just to make sure you realize where my mouse is, for those of y'all listening, I've just in the terminal double-clicked on where the, uh, ydo2-help had printed mouse move.
Ken McDonald [01:28:22]:
Now I'm going to hit enter. And for those of y'all listening, it just moved the mouse up to the far left, then over and clicked in the Kate text editor that I had open, and in the tab that was already open, it typed hello space world exclamation, no quotes, and then it opened a new tab by just typing the, uh, keys Ctrl+N. And that's where the YDO tool key 29:1, 49:1R, Control and then down, and then 29:0, 49:0, Control and then up. And just to give you an idea of what I was talking about, I'm going to copy that YDO2 hello world out and put it up here. And Jonathan, you noticed how it just immediately typed out hello world?
Rob Campbell [01:29:38]:
Yeah.
Ken McDonald [01:29:38]:
And then did it again at the prompt. So I have to Ctrl+C, otherwise you know what error I'd get.
Jonathan Bennett [01:29:46]:
Hello world is not a, uh, uh, oh, I don't know. See, I would immediately think it would tell you that hello world is not a valid, uh, application to run. But you've also got the exclamation mark in there, which is going to try to do something from like the previous command.
Ken McDonald [01:30:01]:
Yep. Which is why I did the Ctrl+C. Just exit out. But it's a great way to automate, like, if you want to select another application and then do something within it. Or you could have a— in a script, have a command that would open that application, followed with the, uh, single ampersand to move it to the background, and then run your command, your YDUO tool command on the next line to do whatever you wanted to in that command.
Jonathan Bennett [01:30:45]:
Very cool. All right. I like it. I'll have to keep that one in mind for scripting and such. Super useful.
Rob Campbell [01:30:53]:
I think, Ken, I think that's my favorite tool that I've seen you share on here.
Ken McDonald [01:30:58]:
Is it because it's the shortest description?
Rob Campbell [01:31:02]:
You could have done it shorter, but no, actually, I wrote that one down. That's something I think I could definitely find some uses. I like automating stuff. So now you can automate GUI things.
Jonathan Bennett [01:31:12]:
Yeah, absolutely.
Jeff Massie [01:31:12]:
It kind of reminds me of the old, on the Amiga, it was A-REC language, or there was an old AT&T language that was based off called REXX, R-E-X-X, that was designed for automating and controlling programs.
Jonathan Bennett [01:31:26]:
Yeah. And then you had XDO for the X11 stuff.
Rob Campbell [01:31:30]:
Yeah.
Ken McDonald [01:31:31]:
Now, this will work with X11 or Wayland.
Jonathan Bennett [01:31:34]:
That's super cool. Yeah, that's neat.
Ken McDonald [01:31:38]:
All right.
Jonathan Bennett [01:31:38]:
Well, as everybody knows, Rob and I in particular, we've sort of embraced and Well, Jeff has too. And I think Ken has to some extent too, but we've all sort of embraced the AI programming. We've got the bug. It makes things so much easier.
Rob Campbell [01:31:50]:
Yeah. How many listeners, Linux listeners have we lost because of that?
Jonathan Bennett [01:31:55]:
I don't know. Probably a few, but you know, it is the future. And you know, I'm not against people writing code by hand. And if somebody wants code written by hand, that's fine too. That's cool. I've found it to be so much more useful to be able to pull the AI in and to get help from it.
Ken McDonald [01:32:14]:
Where I find it helpful is when I get an error back, asking it what the error means.
Jonathan Bennett [01:32:19]:
Yeah, it's true.
Ken McDonald [01:32:20]:
Instead of trying to do the Google search that would take you through 10,000.
Jonathan Bennett [01:32:26]:
So I have been using Claude on Copilot as part of VS Code through Copilot, right? And you can get the Claude extension there. And so it plugs in and everything. works, which was great until I started working on something that was Android related. It's like, well, I've got to do this through Android Studio. And so I went to looking like, how do you do Android Studio with Claude? And the response I got back was you have to install the Claude command line tool. Oh no, no, I've looked at this before. That's one of those bash to curl isms that I really try to avoid running on my computer. I don't want to curl your dumb bash script down and run it without looking at it.
Jonathan Bennett [01:33:07]:
And I really don't want to have to go and try to do an audit. Yes, I know I could ask Claude to look at the script, but that's a little bit of— that's too navel-gazing for me. And then, so I was looking at this and weighing my options. I went looking for like a Snap or a Docker image or an app, anything to try to containerize Claude to get it onto my desktop. And they don't make it easy to find, but they actually have Claude in apps, in RPMs, and in APK. So I've got the link to it. It's in their setup docs under install with Linux package managers. You can apt install it.
Jonathan Bennett [01:33:50]:
You can DNF install it. And you can use APK and add Claude code. So If you can also install it with npm. Oh my goodness. So if you have not used Claude because you're like me and you really don't like the curl to bash-isms, then there is an alternative. There is a way to do it where it's all nicely packaged up inside a package that you can, at the end of the day or whenever you're done with it, you can do an apt uninstall or a dnf uninstall and get it off your system and have some ability, some thought, some hope that maybe it's not spread its tendrils all over your system and you can eventually claim it back. I was actually pretty excited to find this and I've been using it quite a bit since then. So very, very useful.
Jonathan Bennett [01:34:39]:
And they make it so hard to find too. Like I went Googling for this and it was, it took me a while to find it again where it's at. You know, it's like down under administration and in the advanced setup stuff. And finally you find it. So. Putting a pin in it. This is where it's at. So there you go.
Ken McDonald [01:34:59]:
Nice.
Jonathan Bennett [01:35:00]:
All right. I think that is it for the show. We've gotten through the tips. I'm going to let each of the guys get in the last word on something. We'll let Ken go first. I think he's got one more story that we didn't get to. Ken, what do you have for us?
Ken McDonald [01:35:14]:
Well, I'm going to recommend that everyone follow the link I have in the show notes. But pertaining to LibreOffice 26.2.5, it's an article by Markus Nester that I enjoyed reading, but I didn't think we'd want to add another story for me to give. So I'm going to let y'all read it if you want to.
Jonathan Bennett [01:35:39]:
Yeah, absolutely. All right. And Rob?
Rob Campbell [01:35:44]:
I got nothing special this week, so just my usual. If you want to, uh, get more to me, get more of me, connect with me, um, or maybe you just want to show off to all your friends that you are social media friends with me, because that is something to brag about, all you gotta do is go to my website, robertpcampbell.com. And from there you could find links to my LinkedIn, Twitter, Bluesky, Mastodon. And if you're really happy with what I do, Which it appears nobody's really happy with what I do lately. Probably the AI or maybe the anti-X sentiment or anti-Mint. I don't know what it is. If you are happy with what I do, here's where you can donate a coffee. And they're cheap coffees.
Rob Campbell [01:36:33]:
They're only $5 coffees. None of these $6, $7, $8 coffees like the rich people drink.
Jonathan Bennett [01:36:40]:
Oh goodness. All right. And Jeff, what do you got to plug?
Jeff Massie [01:36:45]:
I'm going to need your help on this one, Jonathan.
Ken McDonald [01:36:47]:
Okay.
Jeff Massie [01:36:48]:
I need you to play my son. So.
Jonathan Bennett [01:36:52]:
All right.
Jeff Massie [01:36:54]:
Knock, knock.
Jonathan Bennett [01:36:55]:
Who's there?
Jeff Massie [01:36:57]:
Hike.
Jonathan Bennett [01:36:58]:
Hike who?
Jeff Massie [01:37:00]:
Unsuspecting son. Dad waiting with bated breath sets the perfect trap. Have a great week, everybody.
Jonathan Bennett [01:37:11]:
Bravo. I'll remember that one.
Jeff Massie [01:37:13]:
Thank you, Steve, for that one. I got that submitted to me.
Jonathan Bennett [01:37:16]:
Absolutely.
Rob Campbell [01:37:19]:
All right.
Jonathan Bennett [01:37:20]:
Yeah, appreciate everybody being here. As for me, I'll plug the stuff over at Hackaday. Speaking of office suites, open source office suites, we talked with Michael Meeks of Collabora this past week. And that was a really interesting conversation to try to figure out what sets Collabora aside and where it came from. Did you know that Collabora was spun out of open source? SUSE? I had no idea. Part of one of those, part of a novella. Really interesting backstory there. So anyway, go and check that out, hackaday.com/floss.
Jonathan Bennett [01:37:47]:
Other than that, we just, we appreciate everybody being here. We have such a good time with you and we appreciate you coming along, whether you watch or whether you listen, whether you get us live or on the download. Thank you very much. And we'll see you next week on the Untitled Linux Show.
