Untitled Linux Show 242 Transcript
Please be advised this transcript is AI-generated and may not be word for word. Time codes refer to the approximate times in the ad-supported version of the show.
Jonathan Bennett [00:00:00]:
This week we are concerned about Linux Mint, but we're celebrating Pear OS. River is bringing modular window managers to Wayland. The kernel 6.19 is out and it's time to talk about 7.0. Mesa 26 is out, although it'll probably be a couple of months before any of us really get to play with it. And a whole lot more. You don't want to miss it. So stay tuned.
Rob Campbell [00:00:24]:
Podcasts you love from people you trust.
Jonathan Bennett [00:00:28]:
This is TWiT. This is the Untitled Linux Show, episode 242, recorded Saturday, February 14th. Syntactical Sugar. Hey folks, it is Saturday, and on top of that, it is Valentine's Day, at least on the day that we're recording this. So you know what that means, it's time for us geeks to not spend the day with our significant others, but instead to get geeky about Linux and open source software and hardware. I am joined by the Lonely Hearts Club. No, actually, all of our significant others were just kind and let us come and do the show even though it is Valentine's Day. We got Rob, we got Jeff, and we got Ken.
Jonathan Bennett [00:01:09]:
Welcome to each other.
Rob Campbell [00:01:10]:
Yeah, we're here.
Jeff Massie [00:01:12]:
Thank you.
Ken McDonald [00:01:12]:
At least one of us brought their significant other along.
Jonathan Bennett [00:01:15]:
Oh, really?
Ken McDonald [00:01:17]:
Isn't Jonathan's wife in the audience?
Jonathan Bennett [00:01:20]:
Well, she probably is, but she's still also in the other room with the kids.
Rob Campbell [00:01:23]:
I thought there was an announcement I didn't— it wasn't appropriate or something.
Jonathan Bennett [00:01:29]:
I'm looking for Ken's wife sitting back there on the shelf. Yeah. Is she ever back there? I can't see her.
Jeff Massie [00:01:40]:
Oh, we are there. I'll say we're— we are with the ones we love. We're with all of you.
Jonathan Bennett [00:01:44]:
Analytics. Analytics. We got tux.
Rob Campbell [00:01:47]:
Oh, it can't be with the one.
Ken McDonald [00:01:48]:
My wife sent me to school today.
Rob Campbell [00:01:50]:
Are with. Got.
Jonathan Bennett [00:01:52]:
Yep. All right, well, let's stop with the cheesy Valentine's Day jokes and puns because they're getting painful. We are all exceptionally good at dad jokes. Come by naturally.
Jeff Massie [00:02:03]:
Thank you.
Ken McDonald [00:02:04]:
Yeah, let's just go straight to the mint.
Jonathan Bennett [00:02:06]:
Yeah, let's, let's, let's talk Mint. Rob's got this story, and I've seen this in a couple of places. People are worried about Linux Mint. Rob, what's, uh, what's going on there?
Rob Campbell [00:02:16]:
Yeah, so, you know, as listeners of the show know, one of my common criticisms of Linux Mint is how far behind and outdated it is compared to most other mainstream distros, uh, most notably, um, in their delay to move away from X11 onto Wayland., and I like to criticize it.
Jonathan Bennett [00:02:40]:
Often.
Rob Campbell [00:02:40]:
Um, well, they are considering moving slower. Today they follow a tight release schedule, releasing a new version every 6 months. They're looking to slow that down a little bit, uh, you know, slow down that release cycle. And I didn't— I, I didn't see they didn't really say specifically what they're looking for on a timeframe. But, you know, if they push things out too much further, that is just further away we are from eventual, our eventual Mint Wayland future. But it might not be all that bad. So Linux Mint's project lead Clement is basically saying the 6-month treadmill is eating the team alive. When you're constantly fixing, testing, polishing, and shipping, you're not building, you're releasing.
Rob Campbell [00:03:40]:
And, you know, and as I frequently pointed out, they aren't moving forward as fast as they should be. So, you know, maybe that's part of it. So, you know, the, the idea on the table is a longer development cycle, more, um, as they say, kind of ready when it's ready, uh, less deadline-driven. And that could mean, uh, you know, fewer releases, maybe a little stretched out further, maybe bigger as they come, but— and potentially better ones. You know, they could be better, better releases that they're actually spending more time working on it. You know, more time for ambitious features that don't fit neatly into a 6-month window. Also, you know, one funny little detail, maybe helping them move a little bit, they've run out of alphabetical codenames after the Z release. So it's a perfect excuse to reset the cadence and get creative, try something new.
Rob Campbell [00:04:45]:
Now, about the Wayland future. Mint still calls Wayland support experimental. And they're clearly not promising a flip of the switch moment like some others seem to have been able to do. Okay. But they are knocking down blockers. The biggest headline is a brand new Cinnamon screensaver. The current one is an old X11-only, I mean, GTK app, old like so many things on Mint. And that's a hard stop for real Wayland readiness.
Rob Campbell [00:05:21]:
I mean, or is it? I mean, I haven't really seen a screensaver in years besides when I think Jonathan demoed one.
Jonathan Bennett [00:05:30]:
Yeah, KDE has support. So it surprised me because they're very quiet about it. You have to really go searching for it, but KDE has support for screensavers.
Rob Campbell [00:05:38]:
Yeah, but these days I don't really see many people using it. So, you know, is that really a big blocker? I don't know. At least they're moving it forward. Maybe that would have been my focus, the screensaver. But anyway, the replacement will work on both X11 and Wayland, and it'll be rendered by Cinnamon's compositor, meaning smooth lock animations, tighter integrations, and better security posture. They're also improving keyboard layouts for people who use logical mappings, building a new user admin tool, MiniSysAdmin. With stuff like home directory encryption during user creation, webcam profile photos, and better high DPI handling. Even their forums got a serious upgrade after hammering or being hammered by bots and crawlers.
Rob Campbell [00:06:26]:
So now they have way more CPU, more bandwidth, better filtering. So maybe the slow slowdown will actually speed up their development a little bit. But right now, I'm still going to remain hesitant until I see, you know, see a trend of keeping up.
Ken McDonald [00:06:49]:
With.
Rob Campbell [00:06:49]:
Uh, uh, modern technology for a while. You know, maybe once that trend's there, I won't be so hesitant. Um, and I still think, uh, Mint should dump Cinnamon for a modern desktop environment. I think that would go a long way to speeding up their, their, uh, development. Um, I always say KDE, uh, as someone I think pointed out to me, all their apps are GTK because it's, um, based off of old Cinnamon being an old version of GNOME. So I don't know, maybe, I don't know, go with modern GNOME. Really, I think just pull off the band-aid and go with KDE. But whatever, I guess maybe they'll get Cinnamon modernized and have more time to keep it modernized in the future.
Rob Campbell [00:07:33]:
But, uh, we'll see if the slowdown speeds things up or what it does, or what they'll even do. I mean, it's still just a thought at this time.
Jonathan Bennett [00:07:43]:
Yeah, well, yeah, when you thought— it seems like they, they sort of have, have come to the conclusion that something has to change, right? And that's— what is it they say? The, the first step on the road to recovery is admitting you have a problem. Mint has admitted it has a problem.
Ken McDonald [00:07:59]:
Mint has admitted. Recognizing that you have a problem.
Rob Campbell [00:08:03]:
Yes. Mint's a sleek desktop. I understand why a lot of people like it. I mean, if you don't look under the hood too much, it's nice, clean, works. But yeah, I mean, there are, there are, there are.
Ken McDonald [00:08:18]:
It's a good thing you don't look under the hood of Ubuntu then.
Jeff Massie [00:08:22]:
Well, it's kind of one of those when You know, when you thought Debian Stable was just going too fast and reckless, then you had Mint, you know.
Jonathan Bennett [00:08:30]:
Sad but true. My goodness. Yeah, I did a Mint install for someone years ago and turned on automatic updates and came back a couple of years later. It's like, it hasn't changed any. Linux is way beyond where this is at and this Mint install is not changing. Then I went like, how do I upgrade a Mint install? You do reinstall. That's that kind of soured Mint for me. Those two things.
Jeff Massie [00:08:53]:
I don't know if that's right.
Rob Campbell [00:08:54]:
I don't think that's still the case.
Jonathan Bennett [00:08:55]:
That may not still be the case. But at the time, how do you upgrade Mint? You do reinstall.
Jeff Massie [00:09:00]:
And Cinnamon is based off, I think it's GTK2. It's like an older version, or maybe 3. It's an older version of GNOME.
Ken McDonald [00:09:09]:
GTK2 version of GNOME?
Rob Campbell [00:09:11]:
Yeah, it's an old version of GNOME. It's forked GNOME.
Jeff Massie [00:09:14]:
Yeah, but it's an old version of it.
Jonathan Bennett [00:09:17]:
Which of course puts them at a disadvantage for trying to do things like Wayland support. And they don't get to use all of the fun upstream stuff that GNOME has already worked out.
Jeff Massie [00:09:25]:
Yeah, somebody in the audience could probably tell us, but I thought it was 2 because it was right before GNOME went through one of their big API restructuring things and people said, no, we like it how it is. So they forked it and made Cinnamon. And then GNOME went through a huge, uh, yeah.
Rob Campbell [00:09:43]:
Yeah, it's based off GNOME 2. MATE is the fork of GNOME 3.
Ken McDonald [00:09:49]:
Okay. And what's the fork of GNOME 4?
Rob Campbell [00:09:52]:
I don't think anyone's done that yet. GNOME. Has it been big enough.
Ken McDonald [00:09:58]:
To— Nobody wanted to touch GNOME 4.
Jeff Massie [00:10:02]:
Well, you know, and there is, you know, as much as we laugh, there is a little bit of truth to that because they really soured a lot of their base audience or fans because how much they kept changing everything. And it was, it was kind of almost a little too much too fast. And there was a lot of questions of, well, why are you doing this? You know, some of it was stylistic and some of it wasn't. I think there was some technical reasons, but there was also a lot of stylistic.
Rob Campbell [00:10:29]:
And it just— a lot of it was them being opinionated.
Ken McDonald [00:10:32]:
And of course, none of it had to do with going from GDK 2 to GDK 3 to GDK 4.
Jonathan Bennett [00:10:41]:
Well, I mean, they locked themselves into using an old version as a base. And like I said, they didn't— there's the challenge that if they wanted to continue with their code base and not switch upstream GNOME, they would have to basically redo all of that work to switch to a newer GTK. So they kind of— they stylized themselves into a corner.
Ken McDonald [00:11:04]:
Yeah. And then they finally did switch.
Rob Campbell [00:11:09]:
Which they caused a few other problems. Yeah, but they also future-proofed themselves from that kind of, uh, lock-in again.
Jeff Massie [00:11:18]:
Well, and, and that's— I think they learned the lesson from probably from KDE, because when they went from 3 to 4, that was a terrible rough transition. And that's why KDE can switch faster now, is because they said, okay, we're, we're building this for future changes. And instead of, you know, having some things so locked in, it's, it's much more flexible code base. And it wouldn't surprise me they did the same thing, but that first lift is.
Jonathan Bennett [00:11:45]:
Painful. Yeah. And, uh, there is no GNOME 4, by the way. That sounded weird to me, so I went and I looked, and GNOME 3 was the last release of the old versioning system, and the one that followed that was GNOME 40. And now they're doing new versions every 6 months, I think. So we're up to what, GNOME 49 is now the most recent.
Rob Campbell [00:12:07]:
They're talking about 50 already. So I just think of the 40 series being GNOME 4 and the upcoming 50 series is GNOME 5.
Jonathan Bennett [00:12:18]:
They just removed the dots.
Jeff Massie [00:12:21]:
Yeah. Sorry, I'm on the metric system. It's 4 to.
Jonathan Bennett [00:12:27]:
Me. That's great. All right. So let's see what's coming up next. Speaking of Ubuntu, somebody tried to make a segue in there. We'll let Ken take it and tell us what's up.
Ken McDonald [00:12:38]:
It wasn't an intentional one.
Jonathan Bennett [00:12:40]:
It may not have been, but still it works. Now, Ken, surely you have a typo here because this is all about Ubuntu 24 LTS. There's no way we're still talking about the Ubuntu 24.04 LTS, right?
Ken McDonald [00:12:55]:
Yes, we are because we've got it.
Rob Campbell [00:12:57]:
That's also a little.
Ken McDonald [00:13:01]:
Behind. Actually, no, we're not, because this week Bobby Borissow, Michael Lariville, and Marcus Nester all wrote about Ubuntu team releasing Ubuntu 24.04.4 LTS. Now, this is the fourth maintenance update in the novel Numbat series, and it brings Linux kernel 6.17 and focuses on cumulative fixes since the original 24.04 release. Yes, that was back in April of 2024. Now, according to Marcus and Michael, this newer kernel should support more hardware while the newer Mesa 25.2 graphics stack should improve gaming performance. According to Bobby, the main updated packages include LibreOffice 25.8.4, Mozilla Firefox 147, and everybody's favorite, snapd.2.73. Now, they all remind you that Ubuntu 24.04 LTS is supported for at least 5 years and will continue to receive software updates and security patches until June 2029. Meanwhile, Ubuntu 26.04 LTS is due out this April.
Ken McDonald [00:14:36]:
Who's looking forward to that.
Jonathan Bennett [00:14:41]:
Release? Eh, I'll stick with Fedora. 43? Well, I mean, by then I'll probably be running parts of Fedora 44 on my machines. Yeah, I forgot there for a moment that 25.04 is not an LTS. Ubuntu does every other year LTSs.
Ken McDonald [00:14:59]:
Yep. And everybody gets support for 5 years. Yeah. If you have subscribed to Ubuntu Pro, that extends it to 10. And if you really want to pay, you can get it longer than.
Jonathan Bennett [00:15:15]:
That. Yeah, I'm sure.
Ken McDonald [00:15:16]:
If you need to keep running it that long.
Rob Campbell [00:15:19]:
Yeah, 15, I think it is.
Jonathan Bennett [00:15:21]:
15 years is a long time. Ubuntu 24.04 is going to be real crusty in 2039.
Rob Campbell [00:15:29]:
Oh my goodness.
Jonathan Bennett [00:15:30]:
Me too. Yeah, admit the truth. I mean, you're already kind of crusty, Rob. Let's be honest.
Ken McDonald [00:15:39]:
Not as crusty as I am.
Rob Campbell [00:15:41]:
I thought we were all friends here.
Jonathan Bennett [00:15:42]:
We are all friends here.
Ken McDonald [00:15:44]:
That's why we can say that about each other.
Jeff Massie [00:15:46]:
We're more like brothers. You know, there's, there's a lot of jabs in there.
Ken McDonald [00:15:50]:
Indeed. Oh yeah, don't, don't bring up the fights me and my brother used to.
Jonathan Bennett [00:15:56]:
Have. Yeah, yeah, fun to see this stuff. All right, uh, Jeff, we're gonna talk about kernels.
Jeff Massie [00:16:07]:
Yes. So this week we— what? This week we have— I'm going to just move on. This week we have the release of the 6.19 kernel. Now, it did take a little longer to get out, but that was planned because of the Christmas and New Year's holidays. Linus had said that he was going to give it more time because of the holidays. And then even though in the later RC cycles, he said things were looking pretty good, he kept his word though. And, you know, he thought he's having an extra cycle just because of the holiday. So nobody felt rushed.
Jeff Massie [00:16:38]:
And as always, there are a ton of goodies and fixes, but as always, you know, we're just going to hit the high points here. Now, pretty normal for the releases, you know, there's, there's a ton of AMD and Intel CPU items that have been added and massaged, such as the AMD GCN 1.0 and 1.1 GPUs. They now default to the AMD GPU driver rather than using the legacy Radeon DRM driver. It lets, uh, RADV Vulkan support, you know, work without messing— you know, it works out of the box. You don't have to mess with it and tweak it a bunch. So it just— it'll just work now. And that will give ideally significantly better performance for those that have older Radeon GPUs. Uh, there's lots of supporting software for the upcoming, you know, the I should say a lot of supporting code for the upcoming Wildcat Lake and Nova Lake CPUs, which are going to be out in the future.
Jeff Massie [00:17:42]:
Wildcat should be out soon. You know, it was, it was announced at the Consumer Electronics Show. Sounded like it won't be too long, we should see it. And based on rumors, Nova Lake should be towards the end of this year if, if things hold. AMD was also putting in some foundational support for for their Zen 6 hardware, which last rumors I heard will be about the end of this year or early start of next year. So, you know, maybe January of '27, some, somewhere in there. We'll see, we'll see how, see how it goes. Uh, support for LASS, L-A-S-S, that's Linear Address Space Separation, is now working.
Jeff Massie [00:18:24]:
Uh, it's a security feature found on newer Core Ultra CPUs and Xeon 6 CPUs. The really short version is it restricts virtual address space access, so it keeps people only in the blocks of addresses and memory where they should have access. And other updates, and something we've mentioned in the past, ext4 now supports block sizes larger than the kernel page size. Now this is following the move by other file systems doing the same thing. Bottom line, it aligns with other file systems and, you know, it can, it can see some performance improvements from this change. Uh, you, you know, you're going to be moving quite a bit of data to see a lot of improvement, but if you do, ext4 now should go a little faster for you. Uh, they also have— ext4 also had a performance increase using, uh, cache in certain situations. And there's also some error code improvements and other general small cleanups in the ext4 file system.
Jeff Massie [00:19:25]:
It was actually kind of a bigger code update for the ext4 file system on, on this release, on the 6.19 kernel, which normally file systems have a much slower progression of their code. But there was a nice little tweaking in there. Nothing— there wasn't any major errors or anything they were fixing. It was just making things go faster and improving things. Networking also got love in the form of driver improvements, enablements, and speed increases for some Realtek, Synopsys, MotorCom, and NVIDIA hardware, just to name a few. So it was basically tweaking, you know, either existing drivers or adding new hardware support, you know, adding more features to existing hardware, things like that. One of the biggest overall improvements happened by replacing a busy lock at the transfer queuing layer with a lockless list to yield a 4x improvement in heavy transfer workloads. So this, this was an overall networking improvement, not just on a specific piece of hardware.
Jeff Massie [00:20:31]:
The merge request said sending twice the number of packets per second for half the CPU cycles. So that, that's pretty exciting to see in there. I, you know, I'm not going to go into any more detail on this release, but it's just a tiny fraction of the code that made it in the kernel. A lot of the hardware is better supported, a lot of overall security updates have been made in there. Even hardware which was supported before now is even better supported, more features, all sorts of stuff. Take a look at the link in the show notes for full details. And that article, there's a ton of other links which go into much greater detail. And each thing we touched on and the 100 other things that we didn't, you know, such as Rust— there are several Rust improvements in there.
Jeff Massie [00:21:17]:
So definitely have a look and happy delving.
Jonathan Bennett [00:21:22]:
Yeah, lots of, lots of interesting stuff. What, what are you most excited about in 6.19, Jeff? Which do you think is the coolest?
Jeff Massie [00:21:29]:
I think it's the network, the, the some of the 4x improvement in speed. I thought that was pretty exciting when they're, uh, able, able to speed that up even more, uh, you know, because a lot of stuff I'm to be honest, on an x86, you know, I'm not seeing huge major code changes there. A lot of the hardware stuff is— my hardware is already well supported. I'm not running the latest greatest, you know, some of the other stuff. So a lot of it for me was just the networking.
Jonathan Bennett [00:22:03]:
Yeah, makes sense. There's some, there's some really cool GPU stuff in there too between the the, the new driver support with the RADV compatibility, and then the, the DRM color pipeline API. The, the Valve stuff finally landed. Pretty cool to see that.
Ken McDonald [00:22:19]:
Hopefully it'll, it'll help us all out. And, uh, there's one of the— in, uh, Michael's article, there's a link that takes you to another article that has a link to another article by Michael where he's talking about, uh, bringing many driver core changes for Rust and housekeeping CPUs being exposed. So that's an interesting read there.
Jonathan Bennett [00:22:45]:
Yeah.
Jeff Massie [00:22:46]:
And that's precisely why I said this is a whole link tree. There's a ton of, you know, link to this, to this, to— so whatever your heart desires, it's in there.
Jonathan Bennett [00:22:55]:
Dive as deep as you want to.
Rob Campbell [00:22:56]:
There is a rabbit hole you can go down.
Jonathan Bennett [00:23:00]:
You can— the coolest thing about this is you can actually just go hit the, uh, go hit the Linux kernel mailing list and, uh, read the whole thing, every bit of it, if you want to. Yep, it's all out there in the open. All right, um, we are going to talk about Vim and something very interesting in a changelog there that caught my eye. We'll do it right after this. So Vim 9.2 is out, it's released, and I'm not a heavy, heavy Vim I'm not a heavy Vim user, but I saw the changelog and there was something here that really, really intrigued me. And that is that Vim is advertising Wayland compatibility. And I was very much taken aback by that. Wait a second, Vim is something we use in the command line.
Jonathan Bennett [00:23:46]:
Why does it need to be able to talk to Wayland? So I did a little bit of research into this and looked at some of the other things in the 9.2 release of Vim. There's some other cool stuff in there like fuzzy matching for for insert mode completion, but other UI enhancements. But the one that really got me is the way the modern platform support is what they call it. And this is Wayland UI and clipboard support. I went, oh, okay, okay. Clipboard I get. That's going to be a little bit different than the old X11 clipboard. I didn't know that Vim could talk to the clipboard.
Jonathan Bennett [00:24:22]:
But then I discovered something called the Vim GUI or GVim. I didn't know that this existed at all. Apparently you can run Vim as a GUI instead of just as a command line program. I've never actually run this. It's something I need to check out. But that is essentially what they're talking about here, mostly with Wayland compatibility. They've added, they've made that now in Vim 9.2, GVim is a Wayland application. And then there was a couple of other things that were in here that was really interesting.
Jonathan Bennett [00:24:54]:
The which was that they were talking about the Vim 9 scripting and the fact that you can now do a lot with it. And apparently someone vibe-coded a couple of games in VimScript. And these are not in, from what I understand, these are not in Vim proper, but it's outside GitHub repositories where you can go grab VimScript to run either Battleship or one of the logic games, a number puzzle. Right inside Vim. I thought that was super interesting that, that one, the scripting was to the point where you can do that. It was enough, close enough to Turing complete or Turing complete that you could do that. Um, so yeah, super duper interesting. And, uh, yes, Vim 9.2 just released, and believe it or not, Vim now.
Jeff Massie [00:25:43]:
Wayland compatible. You didn't know you needed it. Yeah, playing the game seems more like.
Rob Campbell [00:25:47]:
An Emacs.
Ken McDonald [00:25:49]:
Trick though. Now, I've got to ask, how do you exit Vim when you're done, especially the GUI version?
Jonathan Bennett [00:25:56]:
The GUI version, you can probably just click on the X.
Rob Campbell [00:25:59]:
That'd be great. Or Alt+Tab. It'd be great if you couldn't just click on the X though. That would make it the best.
Jonathan Bennett [00:26:06]:
I mean, it's fairly easy to do that. I do that in my GUI applications accidentally. You just usually consider that a bug, not a feature.
Rob Campbell [00:26:13]:
I think that'd be a great feature just for traditional humor.
Jeff Massie [00:26:18]:
To save your document, you got to win a game of tic-tac-toe or battleship or something.
Jonathan Bennett [00:26:25]:
I can't let you do that, Hal.
Ken McDonald [00:26:26]:
You've got to win the number puzzle.
Jeff Massie [00:26:28]:
Yeah. Oh, you lost. Sorry, you've lost everything.
Jonathan Bennett [00:26:33]:
Please don't do that. That.
Ken McDonald [00:26:36]:
Sounds terrible. And with that Wayland support, hopefully it allows you to Alt+Tab out of the GUI to another GUI.
Jonathan Bennett [00:26:42]:
Yeah, it should. It's not like Emacs. It's not a full-blown operating system. All right, Rob, Rust is over. We've talked about this before. This is not the first time we've talked about.
Rob Campbell [00:26:57]:
This story. Did we? Did I do a repeat again?
Jonathan Bennett [00:27:01]:
Well, yeah, yes, but no. We, we've, we've covered that Rust is no longer an experiment, but now Linux 7 is out, and so like it's more official. You take it away, you tell the story, and we'll make fun of you afterwards for yourself.
Rob Campbell [00:27:14]:
So Linux 7 is going to be landing soon. The merge window is open, and that comes with a headline that the Rust experiment is over. Linux 7 is officially concluding the Rust experiment, and we all knew it couldn't last. So I guess we just go back to our C roots for Linux and end the whole progress movement we were making. Because, uh, you know, the experiment is over in tech usually means one of two things. Uh, it didn't, um, work and someone is quietly escorting it out the door, or it worked and it has graduated. Well, the experiment is over, but Rust is here to stay. Uh, what Linux 7.0 is doing is, uh, symbolic because of some massive Rust feature bomb or not because of some massive Rust feature bomb dropping.
Rob Campbell [00:28:15]:
But Linux 7 is mostly a version number because Linus likes round numbers, as I think we've said before. At least sometimes he does. But in the merge window, there's a very intentional change. The kernel docs are being updated to declare that the Rust experiment is complete. Miguel Ojeda, the guy who's been one of the driving forces behind Rust in the kernel, basically spells out the intent, calling the experiment done. And it's meant to signal commitment, not just to hobbyists or curious contributors, but the companies, maintainers, organizations that want to invest time and money without worrying that it's going to get scrapped somewhere down the road. Rust isn't in limbo anymore. It's already being used in production environments.
Rob Campbell [00:29:08]:
Some distros are shipping kernels with Rust code already. Millions of Android devices are already running it. And at this point, it's not a science project sitting on a folding table. It's in the building now. The patch itself isn't flashy. It's not like suddenly half the kernel got rewritten overnight. It's just documentation plus some building tools improvements like a new Rust helper annotation to play nicer with the kernel LTO and some incremental crate updates. If you're paying attention, you could feel what they're trying to do.
Rob Campbell [00:29:50]:
They're trying to, you know, stop treating Rust like a guest with a temporary badge to be here and start treating it like a family member. But this is only the beginning of the great things to come with Linux 7. And I know that Jeff is going to want to expand on those.
Jonathan Bennett [00:30:11]:
Later on. Yes, yes. So we have indeed talked about this before a couple of times. I think, I think this is the third time we've covered this story that Rust is no longer an experiment in the kernel.
Jeff Massie [00:30:21]:
But not referencing kernel 7.
Jonathan Bennett [00:30:24]:
It is different this time. Yeah, yeah, yep, for sure. Um, well, yeah, it is, it is interesting to see that, uh, uh, Ethiolegarchs wants to know Linux kernel in Rust by 2036. I mean, I guess that's one way to solve the 2036 bug, but they.
Rob Campbell [00:30:41]:
Have— they haven't actually documented in the kernel code itself that the Rust experiment is over before 7.
Ken McDonald [00:30:48]:
Is that the, uh, Rust helper that.
Jonathan Bennett [00:30:52]:
Michael talks about in one of his articles. Yeah, the actual documentation change may be new, but this was said in the.
Rob Campbell [00:30:59]:
Mailing list quite several weeks ago. Yeah, I think people have said it and now it's in the kernel. It's actually a comment within the kernel itself.
Jonathan Bennett [00:31:09]:
Yep, there you go. And you know, there's actual live code running on, well, depending upon what drivers you have, there's real code running with the Rust. Code. So yeah, it's, it's in there. It's actually.
Ken McDonald [00:31:23]:
A thing. Um, it must be a thing. I've been going through a little, uh.
Jonathan Bennett [00:31:27]:
Tutorial on it myself. Oh, writing, writing what, your first Rust kernel driver?
Ken McDonald [00:31:34]:
Uh, writing my first.
Jeff Massie [00:31:38]:
Rust program.
Jonathan Bennett [00:31:38]:
Ah, your Hello World. Yeah, the Hello World.
Ken McDonald [00:31:41]:
Yeah, that was my first one. My second one's a, uh, Hello, ULS. Program to convert from Fahrenheit to Celsius.
Jonathan Bennett [00:31:51]:
I started the Advent of Code project a couple of years ago in Rust and got through, I think, 3 or 4 of the challenges before I just absolutely ran out of time and couldn't finish it. But, you know, the very early stuff with Rust, it seemed pretty reasonable and pretty familiar. And then I started seeing syntax things that it's like, I don't know what this does at all. And that's about the time that I bailed on it. It's got some funny syntax things in there, which I'm sure it makes sense once you understand what it's doing. It's syntactical sugar, but it's just a.
Ken McDonald [00:32:26]:
Little bit— Syntax and indentation make a difference.
Jonathan Bennett [00:32:33]:
Is Rust whitespace sensitive?
Ken McDonald [00:32:35]:
I don't think it is. With one of my mistakes when I was programming, it was. It said this, close brace is in the.
Rob Campbell [00:32:50]:
Wrong.
Jonathan Bennett [00:32:53]:
Location. Wrong. Wrong. Yeah. Wrong location, but not necessarily because of whitespace.
Jeff Massie [00:32:57]:
Yeah.
Rob Campbell [00:32:57]:
It's not like Python.
Jonathan Bennett [00:33:00]:
Yeah. Python is the canonical example of this language cares about whitespace.
Jeff Massie [00:33:06]:
Or.
Jonathan Bennett [00:33:10]:
Fortran 77. Wow.
Rob Campbell [00:33:11]:
That's from the past. Anyway, to wrap this story up, to wrap this story up, if you want to get the source code yourself, you can do that and you can look through it and you can find in there where it says Rust experiment is done. It's complete. It's no longer an experiment.
Jonathan Bennett [00:33:29]:
And then really don't like Rust.
Ken McDonald [00:33:30]:
And then really don't like Rust. And then really don't like Rust.
Jonathan Bennett [00:33:30]:
And then really don't like Rust.
Jeff Massie [00:33:30]:
And then really don't like Rust. And then really don't like Rust.
Ken McDonald [00:33:30]:
And then really don't like Rust. And then really don't like Rust. And then really don't like Rust. And then really don't like Rust.
Jonathan Bennett [00:33:30]:
And then really don't like Rust.
Rob Campbell [00:33:30]:
And then really don't like Rust.
Jeff Massie [00:33:30]:
And then really don't like Rust.
Rob Campbell [00:33:30]:
And then really don't like Rust.
Jeff Massie [00:33:30]:
And then really don't like Rust.
Ken McDonald [00:33:30]:
And then really don't like Rust.
Jonathan Bennett [00:33:31]:
And then really don't like Rust. And then really don't like Rust. And then really don't like Rust. And then really don't like Rust.
Rob Campbell [00:33:31]:
And then really don't like Rust. And then really don't like Rust.
Ken McDonald [00:33:31]:
And then really don't like Rust. And then really don't like Rust.
Rob Campbell [00:33:31]:
And then really don't like Rust. And then really don't like Rust.
Ken McDonald [00:33:31]:
And then really don't like Rust.
Rob Campbell [00:33:31]:
And then really don't like Rust.
Jeff Massie [00:33:31]:
And then really don't like Rust.
Rob Campbell [00:33:31]:
And then really don't like Rust.
Jonathan Bennett [00:33:31]:
And then really don't like Rust.
Rob Campbell [00:33:31]:
And then really don't like Rust.
Ken McDonald [00:33:31]:
And then really don't like Rust.
Jonathan Bennett [00:33:32]:
And then really don't like Rust. And then really don't like Rust.
Rob Campbell [00:33:32]:
And then really don't like Rust. And then really don't like Rust.
Jonathan Bennett [00:33:32]:
And then really don't like Rust.
Rob Campbell [00:33:32]:
And then really don't like Rust. And then really don't like Rust.
Ken McDonald [00:33:32]:
And then really don't like Rust.
Jonathan Bennett [00:33:32]:
And then really don't like Rust. And then really don't like Rust.
Ken McDonald [00:33:32]:
And then really don't like Rust.
Jonathan Bennett [00:33:32]:
And then really don't like Rust.
Ken McDonald [00:33:32]:
And then really don't like Rust.
Rob Campbell [00:33:32]:
And then really don't like Rust.
Jeff Massie [00:33:32]:
And then really don't like Rust.
Jonathan Bennett [00:33:32]:
And then really don't like Rust.
Rob Campbell [00:33:33]:
And then really don't like Rust.
Ken McDonald [00:33:33]:
And then really don't like Rust.
Jeff Massie [00:33:33]:
And then really don't like Rust.
Rob Campbell [00:33:33]:
And then really don't like Rust.
Jonathan Bennett [00:33:33]:
And then really don't like Rust.
Rob Campbell [00:33:33]:
And then really don't like Rust.
Ken McDonald [00:33:33]:
And then really don't since C helpers.
Rob Campbell [00:33:35]:
Cannot be inlined into Rust.
Jonathan Bennett [00:33:36]:
If you don't like Rust, use Windows. Well, I was going to say, actually, you could download the kernel code and patch that particular line out and then recompile it.
Rob Campbell [00:33:44]:
Then it's just experimental.
Jonathan Bennett [00:33:47]:
Experimental in your build. Forever. Yes. Yes. All right, Ken. What's IPFire? I recognize some of these words.
Ken McDonald [00:33:59]:
Well, then you're going to want to thank Bobby Borisov and Marcus Nester. Since they both wrote about IPFire launching IPFire DBL. Now, DBL is Domain Block List. Now, it's a community-driven system that lets administrators fine-tune network filtering, filtering policies. According to Marcus, it is designed to organize millions of domains into specific threat categories based on your security and content policies. According to Bobby, the project was developed in response to longstanding concerns about existing blocklists. According to IPFire, many available lists aggregate third-party data without clear redistribution rights. IPFire DBL supports formats like DNS Response Policy Zones, or RPZ with the AXFR/IXFR transfers, SquidGuard for proxy filtering, direct HTTPS downloads in several plaintext formats, and AdBlock Plus syntax.
Ken McDonald [00:35:19]:
Now I'm going to recommend reading Bobby and Marcus's articles since I have only touched on the highlights. And I don't think you want to hear me tripping over all those.
Jonathan Bennett [00:35:32]:
Long words. So it's a domain block list. It's essentially a, this is a known malicious domain. We're going to keep you from visiting it.
Ken McDonald [00:35:40]:
But you can open WART, open WART. Basically use tags to identify what type of block list you want to use for certain sites.
Jonathan Bennett [00:35:51]:
Yeah, you know, the, the problem with these, particularly trying to do them open source like this, is that it's so much work to stay on top of the problem because it's so easy to, to generate a new domain or a new subdomain or take over somebody's domain and make it malicious. I wonder, is there anybody that's actually paid to keep up.
Ken McDonald [00:36:14]:
With this? That's a good question since it's a.
Jonathan Bennett [00:36:18]:
Community-Driven domain blocking list.
Rob Campbell [00:36:21]:
Yeah, I'm— yeah, they probably just— do they just use public ones, publicly available ones?
Jonathan Bennett [00:36:26]:
I mean, that's possible. There are some publicly available lists like.
Ken McDonald [00:36:30]:
Um, that just block everything that's on it.
Jonathan Bennett [00:36:33]:
Yeah, so like they could, they could pull from the Spamhaus list, for example, and, uh, there's a, there's a known malicious list there.
Rob Campbell [00:36:40]:
And I know like, uh, in Pi-hole you can select from a bunch of different sources for blocking.
Ken McDonald [00:36:47]:
But then when you implement it as the system administrator, you can only choose only the categories that you want to block. Like say you just want to block malware or phishing or advertising categories, wallet and some of the others such as gambling and games and DNS over HTTPS through.
Jeff Massie [00:37:10]:
Yeah.
Rob Campbell [00:37:10]:
And that kind of stuff is pretty common in like in enterprise firewalls.
Jonathan Bennett [00:37:14]:
So, right, right, right. Yeah. So this, this is, this is tuning it for something appropriate for blocking at a business versus blocking maybe at your house. Yeah. It's cool that it's out there. I hope, I hope that they have funding in place to where it can actually continue to survive because, you know, without, without somebody being paid to take care of it, this sort of thing oftentimes just flounders because it's a lot.
Ken McDonald [00:37:36]:
Of work to set up. IPFire may be actually funding it from the money they get for selling.
Jeff Massie [00:37:48]:
Their prescription IP service.
Jonathan Bennett [00:37:50]:
Indeed, a prescription is.
Ken McDonald [00:37:54]:
Something different. Subscription.
Jonathan Bennett [00:37:57]:
Mine like a sieve. Exactly.
Rob Campbell [00:38:00]:
You hear, doctor, I'm going to write you out a prescription to this firewall. You.
Jonathan Bennett [00:38:08]:
Got viruses. I mean, if you're running IPFire, you might need a prescription for that. You might need it for something.
Jeff Massie [00:38:20]:
All right.
Jonathan Bennett [00:38:20]:
We're going to let Jeff talk about the 7.0 merge window and the stuff that has landed there right.
Jeff Massie [00:38:28]:
After this. With the kernel 6.19 out, one of the questions gripping people was what will the next kernel be named? Well, Linus decided that this next kernel is now going to open the merge window for the 7.0 kernel. To paraphrase Linus, he doesn't like to max out his ability to count on his fingers and toes the number of kernel versions. So next whole number. Now, before we talk about things getting pulled into the kernel, you know, and just, just as an overview, so this is kind of— I'm giving you a little overview of the process here. If you want to add something to the kernel, a person would code it up, they would have it posted in the kernel mailing list for others to look at, you know, you review the code, you know, you compile it into the Linux next kernel, and you do that to make sure it compiles, it functions, you do some testing, and then you can go in and have it added to the merge window, and then it'll, it'll get pulled in. Now I should say, give me a little leeway, experts, I know this isn't exact, but for the very simplified non-coder view, I think it works. Well, something broke for the Linux Multimedia Card, MMC, when they asked for a pull request.
Jeff Massie [00:39:45]:
The code was to see some new hardware support and to revise some of the existing, existing eMMC hardware. Well, Linus had this to say, and I've edited it for brevity and to remove some of the code details, which would be really boring for me to just read out. Linus said, no, those changes are complete garbage and don't even compile. It has apparently never been in linux-next or has been build tested in any way. In other words, that commit— and I'm just editing out all the commit details, it's just long numbers and stuff— is pure unadulterated untested garbage. I do not want to see a quote fixed pull request from you. This was entirely unacceptable and I will not be pulling anything more from you this merge window. Stop sending me untested crap that hasn't been in Linux Next and doesn't even pass the most cursory smell test.
Jeff Massie [00:40:45]:
You can try again for 7.1, but only if it's actually been in Linux Next and properly tested. So now that means the estimated time for the 7.1 merge window, it's probably going to be mid-April depending on how things go. To the person that maintained and, you know, initially had the pull request, I do want to give them some credit and say that it looks like, you know, they're— they basically said the code should have reached Linux Next but didn't, and the author didn't double-check to make sure it actually landed. So there was how they thought and what was supposed to happen. They didn't fully follow up on everything and, and whether they had to or some scripts broke, I don't know. But basically they took ownership for the issue, apologized to, to Linus and said they're going to get it going smoothly for 7.1 by fixing the issues. And they, they did talk about they're going to wait for the first RC. They're going to go get it pulled into Linux next.
Jeff Massie [00:41:52]:
They're going to patch it. They're going to do all the stuff they need to do. So I do want to give them credit for just taking ownership of the, you know, oh, this went wrong. And, uh, you know, I thought that was, that was pretty cool. But, you know, enough though on how things get into the kernel and what didn't make it in. In part 2, meaning my next segment, I'll go over some of the things which are probably going to make it in. Now, no guarantees. Things can get pulled out even if they were first accepted for various technical reasons and whatnot.
Jeff Massie [00:42:24]:
We, we have seen it before. We've commented on it. But, uh, stay tuned for, uh, kernel 7.0 part 2.
Jonathan Bennett [00:42:34]:
Yeah, I, I like the additional context that these guys came back and said, yeah, we messed up, we'll do it right. Like, that's the, that's the point. That's what Torvalds wants. They've got a system there at the kernel. And obviously they have to have a pretty structured system just because there's so much that happens each cycle with the kernel. So many patches fly around. And now that you've got AI hallucinations and LLMs and all of that stuff, making it so easy to write code that at least looks reasonably decent, they've got to have sort of this black magic process to be able to get stuff looked at because otherwise they would just be swimming in LLM contributions the whole time. One of the projects I'm involved in, we had a pull request get open the other day and we called it out and we're like, man, this probably is not, this is not a spec.
Jonathan Bennett [00:43:26]:
We're not going to pull this. And the guy's response was, I kid you not, I didn't realize that Claude Code opened this pull request for me. Like this, this explains so much of what's wrong with open source right now.
Rob Campbell [00:43:41]:
Odd can even open the pull requests for you now.
Jonathan Bennett [00:43:44]:
It's agentic AI, man. That's what it's all about.
Ken McDonald [00:43:46]:
And close it according to Leo.
Jonathan Bennett [00:43:49]:
Well, I mean, we closed that one for him.
Jeff Massie [00:43:52]:
Well, and yeah, I really wanted to just show that it wasn't just somebody just putting out total garbage. It was, They, they weren't watching the process probably like they should have. But I, you know, I really respect people that make mistakes but own it and go, okay, you know what, we did it, my bad, I'm sorry, here's how we're going to fix it, we learned, it's, you know, our lesson going forward. And yeah, I think that just speaks a lot for the character of a person.
Rob Campbell [00:44:22]:
Yeah, it's a lot better than some of the, um, ones we saw last summer. At least one, I can think of one that did not respond so gracefully.
Jonathan Bennett [00:44:37]:
Yeah.
Jeff Massie [00:44:37]:
Yeah.
Jonathan Bennett [00:44:37]:
And you know, all of these situations are different. So, I mean, that one from last year may have been, well, no, I know it was, it was not a.
Ken McDonald [00:44:45]:
Single thing last year.
Jonathan Bennett [00:44:46]:
There was a couple of them last year, but both of them, it wasn't a single thing. It was frustrations mounting over multiple iterations through this, right? But yeah, I agree with you, generally speaking, that yeah, it's good for someone to man up and yes, this, this could have gone better, we'll do better next time.
Ken McDonald [00:45:07]:
And have the finger pointing back at themselves.
Jonathan Bennett [00:45:10]:
Yeah. And, and, you know, I think it's probably worth pointing out here that you can do that without groveling or, you know, making yourself look too ridiculous, but.
Rob Campbell [00:45:18]:
Just a, I blame you, Ken.
Jonathan Bennett [00:45:20]:
This could have gone better, we'll do better next time.
Jeff Massie [00:45:23]:
Yeah, and, and I read the kernel mailing list, the thread, and it— he didn't grovel, he did— he just apologized, said here's what happened, you know, all that, all that kind of stuff, and said, oh yep, we— if, if it would have made it into, I think, Linux-next, they would have found the problem, would have— it would have went smoother. It was just kind of, oh, I was waiting for something to say, oh, this broke, or but it never made it that far.
Jonathan Bennett [00:45:49]:
Yeah. And sometimes stuff like this is just the timing. The timing is just non-ideal. You found something that needs fixed. You want to smash it in, and it's just that time in the kernel. So you throw it in there and hope for the best.
Ken McDonald [00:46:00]:
You got a week left.
Jonathan Bennett [00:46:02]:
Ah, sky's falling.
Jeff Massie [00:46:03]:
And well, and Linus doesn't like those pull requests right at the end. He, he likes a lot of, he likes a lot of runway, a lot of lead time. And the last minute, oh, quick, get this under the wire.
Jonathan Bennett [00:46:16]:
Well, I mean, that makes sense. I mean, how many times— I've multiple times in an open source project thrown something in under the wire and regretted it. Oh my goodness, I've regretted it. We had a— This is for the next release. Yeah, yeah. We had one deal where there was a simple little fix. We were trying to sort a data object in memory. And so we just used a standard sort and it went in right before a big release.
Jonathan Bennett [00:46:39]:
It's like, it's a standard sort. You can't get this wrong. You can't break it. And we had random results from it for months afterwards. That was terrible. It took me like a year to finally realize what was going on with that too. It wasn't what I thought it was. I have finally concluded that our code was multi-threaded and we didn't realize it, which that's a long story.
Jonathan Bennett [00:47:05]:
Anyway, that's a very long story to get into all of that, the fun parts of running with NMB. Embedded code project. Um, Rob, what is up with Pear OS and why is it staying alive? Was this in question?
Rob Campbell [00:47:21]:
Well, Pear OS, it's one of those names, uh, we didn't hear much about for years. And, you know, now we are here talking about them again, only months after it was resurrected from the dead. Essentially, I had a story, I don't know, must have been a couple months ago. But it doesn't just look alive now. ParoOS 26.2 just dropped and this update has big, we're not messing around energy to it. First off, the aesthetics. ParoOS is leaning hard into what it calls the quote Liquid Gel design language, and it's aiming for that glossy, cohesive, everything-feels-connected vibe. They're redesigning the Dock and the Launchpad, added an arc effect to the Downloads folder, which I believe is similar to what macOS Downloads folder does for those who are familiar, and focused on the little stuff that makes a desktop feel premium.
Rob Campbell [00:48:34]:
Smoother animations, cleaner feedback, and fewer moments where the UI feels like a bunch of separate parts stitched together. To back up, for those who don't remember or don't recall, Parrot OS is— it's a KDE Linux desktop, but it is designed, configured to mimic macOS as much as possible. So to continue on, there's the big technical shift. Wayland by default. Yeah, even a little niche distro like ParoOS can do Wayland. Come on, guys. So ParoOS is flipping that switch. Tells us they're not just doing a theme pack on top of Linux.
Rob Campbell [00:49:20]:
They're willing to take on core platform changes. They're also tackling the experience side in a serious way with a brand new installer. Alongside that, there's a new control center. App-wise, they added Puffari, a Safari-inspired browser. For those not familiar, Safari is the one that comes on macOS. And Piri, P-I-R-I, as in Siri, but it's Pear OS, a voice assistant. That acts like an AI helper without actually running any real AI code. So more like a smart command runner, you know, launch apps, run actions, interact with your system naturally, kind of like what Siri used to be when it was dumber.
Rob Campbell [00:50:08]:
Um, you know, all that kind of without predicting it's— or pretending it's your new digital best friend like they're all doing now, which drives me nuts. And sometimes they've had some really surprising results lately. It's, it's, it's, it's interesting. Um, anyway, I was gonna go on a tangent.
Jonathan Bennett [00:50:27]:
I won't.
Rob Campbell [00:50:28]:
Uh, so some of the standout features are surprisingly practical too. Uh, using your display as a ring light during video calls, so you don't need to buy a ring light anymore. You just flip on that little thing and it'll brighten your face for you. Uh, privacy bubbles that, that alert you when the mic is hot. Or the webcam is on, or the screen is being recorded. So, you know, that's the kind of built-in trust signals you usually see more in a polished consumer-focused operating system. And finally, under the hood, ImperOS 26.2 ships with Linux kernel 6.17 and KDE Plasma 6.5.5. And the live ISO even calls out support for both UEFI and BIOS plus NVIDIA support.
Rob Campbell [00:51:14]:
Out of the box. So it seems like they're doing a lot of things to really polish it for those who like that macOS feel and look. You know, we already have a lot of them out there that mimic Windows perfectly fine. So it's nice to see. I mean, they're an old one. They kind of died off for a while. They came back. It's nice to see them going again.
Rob Campbell [00:51:40]:
You know, hopefully they can, keep things going. In fact, over the years, it wasn't even maintained by the same person. So, you know, people get burned out. Hopefully, hopefully they can find something sustainable and keep this smooth-looking.
Jonathan Bennett [00:52:03]:
Project going. Yeah, I, I, I have to question whether it's worth it. Does the world really need a Linux distro that's tuned to look like iOS or Apple?
Rob Campbell [00:52:12]:
I mean, why do we have all these that look like Windows? Do we.
Ken McDonald [00:52:18]:
Need them?
Jonathan Bennett [00:52:18]:
That depends on who you ask.
Rob Campbell [00:52:20]:
It does. And that's the exact same answer. Nope. I like the macOS.
Jonathan Bennett [00:52:27]:
Interface. Eh, there's some things I don't like, but. I not terribly long ago got an iPad. Um, one of the companies I work with sort of, sort of told me I needed to, to be able to test stuff.
Rob Campbell [00:52:38]:
And I'm not talking about the iOS interface, I'm talking about the macOS interface.
Jonathan Bennett [00:52:42]:
Yeah, so, well, that's part of— that's part of where this is going, where I'm going with that. So for one thing, the iOS interface drives me nuts, but I'm used to.
Ken McDonald [00:52:49]:
Android, so it's not terribly— and I bet I know why they're calling it Liquid Gel. Yeah, um, because they can't call it Liquid Glass, right?
Jonathan Bennett [00:53:00]:
So, um, it's interesting that the, the guys that actually use macOS and also have the iPads— and we've seen the changes to iOS, we're trying to do, um, you know, multi-application stuff and the, the, the really not very good way that they're doing it in on the iPad. And the guys have told me multiple times, like, really what we would like to be able to do is just run full macOS on the iPad, and the hardware can do it. It's like Apple just is unwilling to for whatever reason. Um, I guess because it would cut into their sales of their, their Mac laptops, right? Uh, yeah, it's a, it's an interesting thing. I also watched a, uh, watch part of a, a talk given at one of the most recent Ubuntu conferences, um, by a professional GUI designer. Uh, he almost took over the, the, the, the UX, the head. He was almost head of UX at Microsoft right before Windows 11 was released. And he's like, you know, the, the vision I had just didn't match what they wanted, so they told me thank you.
Jonathan Bennett [00:54:01]:
And he goes, like, he's like, I really dodged.
Jeff Massie [00:54:06]:
That bullet. I saw that talk.
Jonathan Bennett [00:54:08]:
That was, that was very good. Yeah, yeah. I can't remember what his name is, but yeah, it was super interesting. Um, yeah, his, his whole point was, by the way, We've got a really interesting opportunity working on the Linux desktop. We've got a really interesting opportunity here to go do something new and notable and like actually take the next step in user interface design. And instead we're spending all of our time in some distros at least, uh, copying what Windows and Mac are doing. And maybe that's not the thing that.
Rob Campbell [00:54:39]:
We ought to be doing.
Ken McDonald [00:54:41]:
Well, sometimes I hope they're not copying the code.
Jonathan Bennett [00:54:44]:
No, no, not copying the code. Um, copying the experience.
Rob Campbell [00:54:48]:
I mean, sometimes they copy us too because Windows 11 looked a lot like KDE before.
Jonathan Bennett [00:54:53]:
And you have things like the virtual desktops that was absolutely born on Linux. And I think that is to some.
Rob Campbell [00:54:58]:
Extent come to the other OSs. Maybe, maybe it was too late, but maybe if he got there, he could.
Jonathan Bennett [00:55:02]:
Have made it better.
Rob Campbell [00:55:04]:
Maybe. You know, speaking of Windows 11, I hear they have a great new feature coming that Windows 11 has not seen before.
Ken McDonald [00:55:14]:
Is it where you just run everything in WSL?
Rob Campbell [00:55:18]:
Well, that's great too, but no, no, they're going to let them move the taskbar to other locations.
Jonathan Bennett [00:55:24]:
Oh yeah, because no Windows has ever let you do that in the past. Yeah. All right. Ken also has thoughts about window managers and Wayland. We're going to let him get to.
Ken McDonald [00:55:35]:
Them, but right after this. Well, Jonathan, this week we're hearing from Liam Proven. About how River Projects is working to break a hard problem into smaller ones. Back at the FOSDEM 2026, Isaac Froude's River Compositor brings a little old-fashioned modularity and customizability to the brave new— Rob, you're going to like this. Wayland world. Now, according to Liam, one of the great joys of the FOSDEM conference is catching program items that introduce radical ideas you had never considered might be possible. Now, almost by accident, he found that one of these talks, a talk titled Separating the Wayland Compositor and Window Manager, an interesting talk. In it, Frued introduced his River project, which he describes as a non-monolithic Wayland compositor.
Ken McDonald [00:56:43]:
Now, River brings to Wayland the idea of a window manager as a separate program. What does this do? Well, it lets you, uh, set up so that you have it already supporting a list of— having the River project supporting a list of 10 different window managers that can work with it. Now currently a Wayland compositor combines 3 primary functions into one. It acts as a display server, manages windows, and then it composites these windows together to be displayed on the screen. Now the River project, it actually splits this up. As I said, it, it's going to be the display server and compositor but it doesn't do any window management. Instead, it provides a documented window management protocol so that another separate program can do the window management. Now I'm going to recommend that everybody read Liam's article for more details into how he's doing all this, and you may find where he talks about the Q&A section at the end of the talk, to be interesting.
Jonathan Bennett [00:58:00]:
Yeah, this is really interesting. It's, it's something I've, I've noted about the way these Wayland desktop systems work. They are, they're, they're, for lack of a better word, they're monolithic. They're all-in-one. Whereas, you know, with the old X style, you had the X11, which was the server and did some of the compositing, and then you had the other parts built on top of it. And so the, the way Wayland has generally worked is it's encouraged KDE and GNOME and those to make sort of a vertically integrated, to use the business term, where, you know, the whole stack is one piece of software. This is an interesting different way to go about this. I kind.
Ken McDonald [00:58:39]:
Of like it. It gives you the option of changing.
Jonathan Bennett [00:58:41]:
Out your window manager on the fly. Yeah. Now, what'll be really fun is if you can run one of these window managers inside of KDE. Or inside of GNOME. I think that would be, that would be particularly interesting.
Ken McDonald [00:58:55]:
Oh yes, I don't think, especially if you found one that worked better with running Steam and another one for doing like, say, uh, video processing and just switch on the fly from one to the other depending on what you're doing at the moment.
Jonathan Bennett [00:59:12]:
Yeah, it's definitely a cool idea. Definitely. All right, Jeff, do you want to talk about 7.0 again, the cool stuff that's actually landed?
Rob Campbell [00:59:22]:
Yeah.
Jeff Massie [00:59:23]:
All right, now see part 2. So now we're going to talk about what we have to look forward to. So AMD is putting more hardware blocks in for RDNA 3.5 and RDNA 4, and they're also adding blocks for GC/GFX 12.1. Now I say blocks because AMD has a pile of code blocks instead of a single monolithic driver. When a new piece of hardware comes, comes out, it's based on the, on the hardware. The driver pulls different code blocks based on what the features— on, based on the features it supports. So this helps keep things a secret because you don't fully know what blocks will be pulled into the hardware, you know, based on what hardware. So there's no reverse engineering a large monolithic driver ahead of time.
Jeff Massie [01:00:13]:
To figure out exactly what it'll support. And it makes it more flexible, so you just pull in what it needs and it doesn't pull in what it doesn't. So it, it's, it's been something that AMD's been working on for a while, and it's just in, in more of that. But we know the future hardware is at least going to support some of these new code blocks coming in. The Novu kernel driver is getting support for larger pages and compression, which should help performance along with other code improvements. There's Several things they're doing there to make, make the driver even better. Intel's in the game as well by supporting GPU firmware updates on their discrete GPUs. Oh, so what, you say? This is on non-x86 platforms.
Jeff Massie [01:00:57]:
So this will let people using an Intel GPU, like a discrete GPU, on an ARM64 or RISC-V to get the updates rolling. So now you, uh, aren't, aren't going to have to go, I can't, I can't update my, my firmware. Now you, now you'll be able to no matter what hardware platform you're on. Continuing with speed improvements, there's more caching code is being replaced with Sheaves. Very simply, Sheaves is an array-based caching layer which should give an increased speed for the kernel. Now the code started The sheaves code started in the 6.18 kernel, and but it's— and even now it's currently opt-in. But hopefully with this latest, uh, pull, there's going to be enough of the code to run some performance numbers to get a true verifiable performance increase. It should increase, but we don't have any hard numbers yet, so it's still kind of theoretical.
Jeff Massie [01:01:59]:
But it's opt-in for now, so nobody needs to worry about anything, you know, breaking something. Google is putting in a revocable resource management. Not quite like it sounds, but if you have a hot pluggable device like a USB and someone removes it, the resources are removed and the memory freed. It can cause a use-after-free issue. So this code that Google's putting in will make sure that when someone tries to access the device, it's still valid, so it hasn't been unplugged or anything hasn't happened to it. And if it's not, it fails safely. So it, it basically is, uh, taking care of a few issues there. A lot of hardware support is going in for things like better sensor monitoring for ASUS desktop motherboards, fan target and temperature thresholds for Framework 13 laptops.
Jeff Massie [01:02:53]:
The LG Gram Style 14 laptops get speaker support. And there's some things going away, like for example the HIPPI. It's a networking standard for supercomputers which almost went to 1 gig in the '90s, so it's, it's an old deprecated standard. Also the Intel 440BX chipset EDAC driver, E-D-A-C, it's, it's going away. It's been broken for 19 years anyway. And you know, the thought is people running Pentium 2 and Pentium 3 systems are not loading on the latest greatest version of Linux. The, the chips are from the '90s as well. Uh, the old mount API code is going away and it's being replaced by the new mount code.
Jeff Massie [01:03:42]:
Uh, in the entry file systems, so the file systems in the kernel have all been converted to support the new system, and it should, but it should be noted this is for inside the kernel. The user space old API mount calls are still supported, so it doesn't break anything for users who are using software that has not been converted to the new API system. Take a look at the article linked in the show notes for all the full details and a ton of other links in there, another Linktree covering many more changes, upgrades, improvements, things like and I, which I didn't cover. And there's a lot of, uh, more technical stuff like there's IOU ring, IO pole improvements, and RISC-V user space control flow integrity support and things like that that I didn't even touch on and get, uh, rather, rather down deep in there. So, you know, dig in, find what tickles your fancy, and happy reading.
Jonathan Bennett [01:04:41]:
Yeah, an example of that is they also— he also mentions the Focusrite Forte USB audio interface support. I always find those things fascinating because I've spent some time throughout my career fighting with pro-level audio interfaces and trying to make them work under Linux, and sometimes that's not the best experience. So it's always nice to see another.
Ken McDonald [01:05:01]:
One officially getting added. What I find surprising and fascinating is that with the human interface devices subsystem merges, it's going to include support for the Rock Band 4 guitars that were designed for PlayStation 4 and PlayStation 5.
Jonathan Bennett [01:05:22]:
Yeah, that is interesting to see too. So I wonder, is that usable like as a controller for emulation? Can you do other things with it? Can you play real music with them now?
Rob Campbell [01:05:34]:
I'm sure. Can you map it to keyboard keys.
Ken McDonald [01:05:37]:
And just type it? Does that mean you can use it as a MIDI controller to a synthesizer.
Rob Campbell [01:05:48]:
To jam out on?
Jonathan Bennett [01:05:48]:
If you got the software. Yeah. I mean, if you can capture the inputs, then sure, you can convert those into MIDI. There are, yeah, there are plenty of plugins that'll do that. I'm, I want to look into that. That fascinates me too. Not that I ever played those games.
Ken McDonald [01:06:12]:
But again, the music thing is— Then you might want to follow the link.
Jonathan Bennett [01:06:17]:
I just posted into the Discord. I was already climbing the Linktree. Yeah.
Jeff Massie [01:06:25]:
Already looking at it. Yeah, see, I played, uh, I think it's Rocksmith where you plug in actual instruments and you, you, you play real instruments and play real music.
Jonathan Bennett [01:06:38]:
Yeah, that one's pretty cool too. It was almost a, uh, a true.
Rob Campbell [01:06:44]:
Learn-To-Play sort of, uh, game. A game where you plug in real instruments too?
Jeff Massie [01:06:49]:
Yeah, and they, they start you out real simple and My, my son learned to play bass by doing that, and.
Ken McDonald [01:06:55]:
Now he can just play it. Yeah.
Jonathan Bennett [01:07:03]:
And who sings tenor? Oh, probably nobody in the household. No. All right, uh, I've got one final story that I want to chat about, and that is that Mesa 26 is out just in the last couple of days, um, February 11th. I believe was when it released. And Mesa 26 has some really fun stuff in it. So Mesa is your video card drivers, open source video card drivers. This is the one that AMD does a lot of open source work in, Valve sponsors a lot of open source work in. This is where things like the Vulkan ray tracing support lives.
Jonathan Bennett [01:07:40]:
This is where the, the, the RADV Radeon driver is at. And a lot of fun stuff got added. One of the big changes in 26.0 is that AMD ray tracing is going to be a lot better, a lot more performant, and a lot closer to what you would expect comparing the same hardware on Windows or using the AMD closed-source drivers. There's, there's a bunch of highlights in the official release list. But one of the, again, one of the big ones that I kind of keyed in on was the performance improvements, particularly around ray tracing, which is something that I fiddle with from time to time. And so I went, I got curious, and I went looking on the Fedora desktop behind me because obviously it's Fedora, pretty bleeding edge. And so I did a, I did an update, and I said, okay, what version of Mesa am I going to get on this? And I ended up pulling down Mesa 25.
Ken McDonald [01:08:39]:
3.
Jonathan Bennett [01:08:39]:
Uh, we went from 25.2 to 25.3, and Mesa does a, I believe, a 3 times a year, uh, release cadence. So I, I'm pretty sure there's not a Mesa 25.4. Um, it does 3 times a year, and so the one directly after 25.3 is 26.0, and 25.3 was released, I think, November November of '25, sometime around then. So Fedora is on Fedora 43 at least about 3 months behind. They give it that much time for it to stabilize. This may be another case of me going and installing packages from Fedora 44 a bit early. Surely that won't break anything. It never does.
Jonathan Bennett [01:09:23]:
I never regret doing that. Yeah, Mesa 26 is officially out, but it's going to take a little bit before it lands in our actual desktops, unless you take a take a, take a drive.
Jeff Massie [01:09:39]:
On the bleeding edge.
Jonathan Bennett [01:09:41]:
Yeah, it's— 26 is big. It's got a lot of cool stuff in it. Yeah, it's, it's huge. Huge. Uh, yeah. Um, okay, so those are the, the stories. Let's get into some command line tips and we're going to let Rob go.
Rob Campbell [01:10:02]:
First, but we're going to do it right after this. So my command line tip, it's a simple one. It's called Cull, C-U-L-L. It's a Go run program. It's an interactive TUI, it's TUI disk space analyzer. It scans directories. You can find what's eating up your disk and delete it all from within this 2e terminal. So for those watching, I'm going to show you.
Rob Campbell [01:10:29]:
If I just type cull in here, C-U-L-L, enter, and it brings me to the 2e interface. I can go through. There's not a whole lot on the system because it's a fresh demo system, but it shows you has a nice little graph of stuff. Yeah, that's all there is to it. I'm not going to show you at the bottom. I could select multiples with the S and I'm not going to delete anything. Suppose I could, it's just a demo system. Hit D, it's going to delete them all.
Rob Campbell [01:11:10]:
Hit yes. And oh, I just deleted those 3 files, which is fine because it's just a demo system. I can do that kind of thing. But yeah, at the bottom it shows you what the commands are. S to select, D to delete. F to filter, T to.
Ken McDonald [01:11:27]:
Sort, E to preview. Um, and how easy.
Rob Campbell [01:11:33]:
Is it to undo? Uh, you're not going to use this program to undo. It's just like, uh, if you do an RM, uh, file, it's not going to be easy. You're going to need some third-party thing to that. Depends how good.
Jonathan Bennett [01:11:48]:
Your backups are, Ken. So we've covered another program that does something similar, a command line.
Rob Campbell [01:11:57]:
NCDU?
Jonathan Bennett [01:11:57]:
Yes, NCDU. That's the one. I've used that a lot. It'll, it'll go through and it'll tell you, you know, your disk usage and it'll let you explore down the tree and delete things. It looks like this is pretty similar. I'm gonna have to run them side by side and see which one I like best.
Jeff Massie [01:12:15]:
Because they find that one monster log file that's hogging up your disk space.
Jonathan Bennett [01:12:20]:
Or that folder that's got, you know, 5 ISOs in it that you forgot about. Yeah, absolutely. Um, I'll check this out because it's pretty cool. Looks like it does the same thing. Uh, I like it. All right, up next is Ken, and we're going to talk about, uh, looks.
Ken McDonald [01:12:36]:
Like some systemd stuff again. Yes, we are. And it's a simple command. Let me go ahead and bring up my command line. Let me go ahead and get these up so y'all can see it a bit better. Is that one big enough.
Jonathan Bennett [01:12:56]:
For y'all to see?
Ken McDonald [01:12:57]:
Make them a little bigger.
Jonathan Bennett [01:13:05]:
Okay.
Ken McDonald [01:13:06]:
There, how's that?
Jonathan Bennett [01:13:06]:
And now for the other side.
Ken McDonald [01:13:12]:
Yeah, there you go. But the command itself is systemd-detect-virt. Now, have you ever wanted to determine if your script is running in a virtual environment? If you have, then this command is what you need. Now, on the left side, I'm just going to run it. And that's its response. Now over on the right side, for those of y'all listening, I've switched to my openSUSE Tumbleweed that I've got running in a virtual machine. And when I hit enter for systemd-detect-firt, it comes back with the response KVM. With the other one, it said none because that was on the hard middle.
Ken McDonald [01:14:00]:
Now it's got a man page, so you can go in and look at some of the other options that I'm not going to bother to explain because they're just minor, and it's easy enough to pull up the man page. But.
Jonathan Bennett [01:14:21]:
That'S it for systemd-detect-virt. Also very useful to determine if you're inside.
Rob Campbell [01:14:26]:
The matrix or not. So yeah, I imagine that's not something you would have a lot of use just running because you probably know where you're at, unless you have a whole bunch of tabs open, like which am I inverting? But the useful thing I could see there is if you have a script.
Ken McDonald [01:14:42]:
Running and you want to see if it's running on bare metal or invert.
Jonathan Bennett [01:14:45]:
Yeah, you want to have different behavior. One or the other. The other place it seems like it could be interesting is if you rent if you rent a server and you want to know, am I actually running on bare metal or not?
Rob Campbell [01:14:58]:
A quick way to find out. You're supposed to be paying for a dedicated bare metal server and you go.
Ken McDonald [01:15:04]:
Around and say, that's not dedicated.
Jonathan Bennett [01:15:06]:
It's not dedicated bare metal. Yeah, I've got an OVH server I've just started renting. I need to go run that on.
Jeff Massie [01:15:10]:
That one and see what it is.
Jonathan Bennett [01:15:13]:
Hey, wait a minute. Yeah.
Ken McDonald [01:15:17]:
Exactly.
Jonathan Bennett [01:15:17]:
Want a refund. All right, Jeff, what is, What is preload? What are we talking about here?
Jeff Massie [01:15:23]:
Well, we've kind of got two intertwined commands. They're, they're different, but they're very close. Uh, preloading, for those who don't know, is the action of putting and keeping target files— you put them into RAM. The benefit is that preloaded applications start more quickly because you're reading them from RAM, and it's always quicker than from a hard drive. Even, even NVMe, your RAM is always faster. However, part of your RAM will be dedicated to the task, but no more than if you kept the application open. Therefore, you know, preloading is best with, you know, large and often used applications. Now it's, it's interesting because the wiki gives a program example like Firefox, but for me it wouldn't be because for my web browser when I boot up, I open it up, I never shut it down, so it doesn't load and unload out of memory all the time.
Jeff Massie [01:16:20]:
But if I had something else that I load and shut down all the time, that would give me a benefit. Now, GoPreload and Preload are similar programs but works a little differently. Now, GoPreload is a program where you specifically call out the program or programs you want to reside in RAM so you can load them faster. Now, preload is a program that tries to look ahead and figure out what should be preloaded by trying to predict what you're going to want by looking at your history. Uh, for the programmer or predictive math savvy, it uses Markov chains for prediction. If you really want to get into the math and all your system predictability, that's, that's a nice little read. If this sounds like something you would want, they could speed things up. Take a look at the link in the show notes for a link to the Arch Wiki with more details.
Jeff Massie [01:17:22]:
Now I do know that preload is widely available. You can— it's on Debian and Fedora and it's everywhere. Go preload, I didn't check on that. Uh, it, it is on all Arch— you can get it on Arch-based, and there's a few others as well, but I don't know, like the Red Hat and Debian, if, if there is widely available. But, and they're, they're really pretty simple. Preload is just basically preload, and go-preload is pretty much go-preload and then your program.
Jonathan Bennett [01:17:59]:
So, but Happy, happy, uh, loading. Very cool. All right, so I've got, I've got one that really tickled me when I first heard about it. Um, I'm going to talk very briefly about a— it's a script, uh, and it's the New Script script. And so this is actually written by, uh, Bill Schatz, and I had him on Floss Weekly this, this past week. And Bill wrote the Linux command line guide, which I've ordered and I will have a physical copy of probably by next time we do this show. But he described this as, so you want to write a new script and you run this script for it to, it'll do the setup for you. You tell it what the script name is, you tell it what the options need to be, the help text, the thing that you want it to do, all of that, and it generates the script for you.
Jonathan Bennett [01:18:53]:
And when he said that, I started laughing. I said, so it's vibe coding without the LLM involved. And he's like, yeah, I guess it is kind of what it is. I thought it was hilarious. But then he says, really, I would love to see more people taking a look at this, help us find bugs and ways to improve it. So I thought, I'll throw it out here. It's pretty cool. Essentially what it is, is it's a template.
Jonathan Bennett [01:19:13]:
It's a new script template. Where you, you run this and it generates a new script in the right place for you, and then you can go in, you can actually flesh the meat out and put the things in that you actually want it to do. But it does— it looks pretty interesting. It looks, uh, it looks like it would save a lot of time putting the boilerplate stuff on, uh, on the file system for you and then go in and actually do the work. So if that's something— if you are a script wizard and you write lots of new scripts, give this a try, and then Bill would love some feedback on it. And this was the one thing that he said he wished people would take a look at. He thought it was pretty cool and not enough people, not enough other people did. So there you go.
Jonathan Bennett [01:19:54]:
And I just, I just love the description that came up with vibe coding without the LLM.
Ken McDonald [01:19:58]:
I thought that was clever. Yeah, it looks interesting.
Jonathan Bennett [01:20:02]:
I'll definitely have to take it for a spin. Ken was definitely the one that I thought of that would like this the most. All right, that is the show. I'm going to let each of the guys get in the last word if they want to on something or plug whatever they've got to.
Ken McDonald [01:20:18]:
We'll.
Rob Campbell [01:20:21]:
Let Rob go first. All right, for those who like what I do, want to see more of this marvelous mustache that I brought to you today, I brought this to you just for the show. You can come. And, uh, find me on the social webs. My website is robertpcampbell.com, right there at the top. That's robertp, uh, campbell.com. And on that page, you'll find links to my LinkedIn, Twitter, uh, Bluesky, Mastodon, and a place to donate a coffee or, you know, some, some little mustache wax so I can get these, uh, handlebars to curl a little better for you. Uh, I'll use that, uh, one of.
Jonathan Bennett [01:21:02]:
Those coffees for that.
Jeff Massie [01:21:07]:
Huh? Yeah, do that. He's twisting his mustache right now if.
Jonathan Bennett [01:21:12]:
You'Re on audio only. It's like a.
Jeff Massie [01:21:19]:
Like a cartoon villain.
Jonathan Bennett [01:21:21]:
Dastardly or something was his name. That used to be Dan. Yeah, yeah, that's who he looks like. Uh, one of my social media, uh, personas somewhere. Uh, I forget if it's Facebook or Twitter or GitHub, you know, one of them. I had that handlebar mustache thing going on. I went through that phase. It didn't last super long, but, uh, I, I did it for a little while.
Jonathan Bennett [01:21:42]:
All right, Jeff, save us from the handlebar mustache.
Jeff Massie [01:21:46]:
Well, this week it's going to be a kind of poetry corner. So this, this should be, and I'm not going to do it. It should be sung to the, uh, Aerosmith song Come Together. Here come old laptop, he come booting up slowly. He got broken touchpad. One more update, he got no, no space on this PC. One thing I can tell you, it's got no warranty.
Jonathan Bennett [01:22:21]:
Have a great week, everybody. Aerosmith is not my band. I'm going to have to go and look up this particular song to get.
Rob Campbell [01:22:28]:
What they're talking about. I want to say one more thing. Donate $5 and tell me if I should keep this or shave it because I may just shave it off. But if you think I want to, if you want me to keep it.
Jonathan Bennett [01:22:40]:
Donate and put a comment there. All right. There you go.
Ken McDonald [01:22:45]:
All right, Ken. Well, I've got a link in the show notes that I wanted to share with everybody about what could be the death knell for sysvnet support in a lot of the independent distros. Basically, Linux from Scratch is dropping it.
Jonathan Bennett [01:23:05]:
With their next release. There you go. All right. Thank you guys for being here. Sure appreciate it. Had a lot of fun tonight. If you want to find more of me, there is Hackaday. You can find FLOSS Weekly there.
Jonathan Bennett [01:23:19]:
We tape on Tuesdays and go live on Wednesdays and have a lot of fun with that. You're all welcome over there as well. And other than that, just want to say thank you to everybody that's here, whether you watch or listen, whether you get us live or on the download, we sure appreciate it. And we will be back next week.
