LINUX Unplugged - 589: 6 Reasons to Love Linux 6.12

Episode Date: November 18, 2024

The Linux 6.12 kernel isn't just another update — it's a game-changer that deserves our full attention, from performance improvements to fascinating new features.Sponsored By:Jupiter Party Annual Membership: Put your support on automatic with our annual plan, and get one month of membership for free!Tailscale: Tailscale is a programmable networking software that is private and secure by default - get it free on up to 100 devices! 1Password Extended Access Management: 1Password Extended Access Management is a device trust solution for companies with Okta, and they ensure that if a device isn't trusted and secure, it can't log into your cloud apps. Support LINUX UnpluggedLinks:💥 Gets Sats Quick and Easy with Strike📻 LINUX Unplugged on Fountain.FMLinux 6.12 Features Are Super Exciting With Real-Time, Sched_ext, Intel Xe2 & Raspberry Pi 5 - PhoronixRelease v6.12 · torvalds/linuxLKML: Linus Torvalds: Linux 6.12The 6.12 kernel has been releasedLinux_6.12 - Linux Kernel NewbiesLinux Kernel Release ModelHow do kernel version numbers work? : r/linux4noobs2. How the development process works — The Linux Kernel documentationoracle/bpftune: bpftune uses BPF to auto-tune Linux systemsWhat's scheduled for sched_extSched_ext at LPC 2024sched_ext: scheduler architecture and interfaces (Part 2)sched-ext/scx: sched_ext schedulers and toolsRaspberry Pi 5 Initial PatchPrintk Changes Submitted For Linux 6.12 Finish NBCON Console Preparations - PhoronixReal-Time "PREEMPT_RT" Support Merged For Linux 6.12 - PhoronixReal-time programming with Linux, part 1: What is real-time? - Shuhao's Blogrealtime:documentation:howto:applications:preemptrt_setup [Wiki]Annual Membership — Put your support on automatic with our annual plan, and get one month of membership for free!binary boost soundLINUX Unplugged 563: Nix's People ProblemBtrfs Assistant — Btrfs Assistant is a GUI management tool to make managing a Btrfs filesystem easier.kernelconfig.io — search config information for linux kernel modules

Spin up a few more of these over here. You know, just torture the system. Okay. Hello, friends, and welcome back to your weekly Linux talk show. My name is Chris. My name is Chris. My name is Lance. And my name is Brent. Well, hey, gentlemen, coming up on the show today,

the Linux 6.12 release isn't just another kernel update. It is an actual game changer, and it deserves our full attention this week. From the performance gains to the massive upgrades, we're going to dig into the top six features of Linux 6.12. It's going to blow you away. And then we'll round it out with some great boosts, a double mint pick segment, and a lot more.

Wow. Double mint picks, Wes. I don't even know if I know about this. So, wow. You're working overtime today. I snuck in a surprise. So before we go any further, let's say time-appropriate greetings to our virtual lug.

Hello, Mumble Room. Hello, Chris. Hello, Wes. And hello, Brent. Hello there, On Air. And look at that. Powerful, quiet listening up there. It's powerful.

Jeez. It's awesome. Thanks for joining us, everybody. And a big good morning to our friends over at Tailscale. Tailscale.com slash unplugged. It is the easiest way to connect your devices and services, wherever they are, whatever they are,

go get tailscale on 100 devices for free at tailscale.com slash unplugged. It is the best way to build a simple network across complex infrastructure, creating a flat mesh network powered by. Wow. That's right. It's really slick. If you got like five systems, it'll take you three minutes to. That's right. It's really slick. If you've got like five systems, it'll take you three minutes to get them all going. It's so great.

I love it. Tailscale.com slash unplugged. Gentlemen, before us is Linux 6.12. It may be released right now as we're recording. We're doing this a little early, so we don't know. It's on schedule to be released today. Yeah, you never know, right? Like Linus makes the call,

he could decide that an RC8 makes sense, but either way, it's coming real soon. If he did that, it looks like the next release would be then next Sunday. Right. I'm thinking we're going to see it, though. I bet you by the time we hit stop, I bet you it's out.

Yeah, and you know, there's already a little pressure. Plenty of folks are getting stuff queued up for the 6.13 merge window, which will open once 6.12 is out. So it presses on. There's a lot of exciting new features in Linux 6.12, and I don't think we've ever, ever dedicated an entire episode to a kernel release. Yeah, we've talked about big features from releases,

but usually there's something else going on in the show. Not so today. But this is probably a kernel that you are going to end up running because of, well, we'll get into it. And that is our first feature. Gene Bean, take us away with top six. Number six. Number six. There you go.

Thank you, Gene Bean. Well done, Gene Bean. It is. There you go. Thank you, Gene Fien. All right. Well done, Gene Fien. It is the next LTS. One kernel version is selected to be an LTS each year. The selected LTS kernel is then maintained for two years, sometimes longer. Yeah. Often, you know, these are the kernels that end up in LTS distros, and then they're maintained and patched, and you see them on servers for years.

Android phones? Oh yeah, absolutely. So the idea is you want them to be stable. You want to be able to support them for a while. You want them available for enterprise and embedded systems. And so the LTS kernel receives regular updates and bug fixes and security patches that get backported from newer kernel versions. I know Greg KH is intimately connected to it.

ported from newer kernel versions. I know Greg KH is intimately connected to it. On average, the LTS kernels receive 9 to 10 patches per day compared to 10 to 15 patches for the regular stable kernel. Huh. You know, my first question, if I had Greg here right now, if he was in the mumble room, wouldn't that be something? Yeah. Why do you decide which patches go to the LTS kernel

versus, that's a bit of a delta. I imagine there's, you know, part of it probably is, uh, how much, you know, who's demanding it, who wants it, what's the,

you know, is it a security thing that needs to you? Is it a feature improvement that is like widely adopted? And then of course there's all bugs, just bugs, Wes. That's true. True.

Uh, but then there's the question of how much darn work is it? Yeah, that's probably, I mean, you know, some things are relatively simple to backboard and others are a huge pain. Yeah. And then of course I would imagine towards the end of that LTS kernel life, it becomes more and more of a challenge to backport. So it becomes trickier and trickier

because the mainline kernel is moving further and further forward. Right. You're going to either have to like drag a bunch more code back for what you need or re-implement it or. So right off the top, the reason why this is going to be an important kernel is because it is the next LTS. Yeah. You know, on the show, we live fast and crazy and run RC kernels sometimes, but the reality is a lot of places, especially like in a work production environment, you don't necessarily have the luxury of picking what kernel. You got what you got on the supported distro that you happen to be running unless maybe your workload really demands something specific so it makes sense to be familiar with the features you're going to have for a long time coming plus this is kind of the new baseline yeah it'll be once it is deployed for like what linux can do yeah yeah and we'll get into why that's going to be such a big

deal number five a lot of file system fun in this kernel a lot uh obviously um bcash fs gets some love we'll talk about that but just some good stuff in here for xf xfs really nice to see uh iou ring now supports async discard so you're going to get some significant performance improvements there fuse gets some nice improvements here butter fs gets some performance enhancements. Is that right? Everything gets a little love. Bcache is expected to have the experimental flag removed next year. And 6.12 sees some key features, like the base. And 6.13 is going to have even more for self-healing features of BcacheFS.

Hey, and that's what you want, right? You have a fancy copy-on-write file system that supports having replication and multiple copies of data, and you don't want to have to do a scrub or run a, you know, file system check. It should be able to just figure it out and keep on going. Also in 6.12 for BcacheFS, improve scalability to support petabyte-sized file systems. Not a problem I have, but... I'd love to have that problem. Yes, and if you eventually want enterprise adoption, probably something you need. Also a better user experience around degraded mounts. But I think maybe more so than the features that we're seeing in BcacheFS for 6.12, it's the fact that these patches are in 6.12. Right.

You know, actually, I did a little firsthand testing because I got 6.12 RC7 running on my system with an encrypted root Bcachefs file system. And I think at some point there must have been, you know, some stuff wrong with the file system, because the first time the new kernel ran, it reported fixing some stuff. Oh. Yeah, and then later just to see, I did like an offline

full file system check and it fixed a bunch of stuff. And now I can boot between either kernel and both are totally happy. Okay. It's working. So there's clearly been work and improvement in both the runtime user space tooling and the kernel tooling.

And I know coming in 6.13 too, there's work to make those two paths basically the same. So it's always going to be consistent. But do you agree the bigger story maybe here is the fact that BcacheFS hasn't been forced out of mainline and that patches are landing in 6.12. They are landing in 6.13 already. Yeah, definitely. Seems like BcashFS is still on board. Yep. I mean, we haven't

seen further, you know, incidents or explosions or heated discussions. I think at the end of the last one it seemed like Linus and Kent had gotten on the same page. You know, we're understanding each other's goals and motivations and we're willing to work together. And let's hope that continues.

ButterFS got some fixes and some code refactoring and cleanups. That's where the performance comes from. So there's also, along with that, some improvements now to error reporting, preparations for compression in subpage mode and reducing the extant lock scope for read operations, which will allow for smoother integration with IO map later on. So they're like also laying down the future for butterFS there. Doing some plumbing. ButterFS is in a really, really tight spot now. Uh, if you haven't checked it out, it's not 2010s butterFS anymore.

It is whatever. I don't, I don't have a good... It is much, much better. It is much improved. Are we all running to some extent... I know you're on Bcash on this. Do you have ButterFS anywhere? Yeah, I do have ButterFS on some other systems. Brent's got ButterFS now. On his laptop. So do I on my

home NAS and on all of my desktops. And none of them have had any problems. Just getting better and better. Yeah. And you know, they've got more... you still shouldn't use all of my desktops and none of them have had any problems just getting better and better yeah and you know they've got more you know you still shouldn't use five six but they've got raid one c3 and four if you uh need a lot of data production and what's neat is you can also use that for your file system metadata even if you don't use it for the data to just help make sure you have a really robust file system i was doing some tests uh as bre Brent was exploring this NAS path where I made a little

file system with four disks and arrayed 10 and then I killed one disk and then I killed a second disk just to see could I still mount it and how much of the data could I recover? And yeah, it killed about 33% of the files as it was expected

to because I nuked more than I was able to survive. But I was able to recover the rest. The rest. Yeah, at least you could get that, the remaining. Mm-hmm. Huh. That's a fun experiment.

Yeah. You madman. Of course, I didn't do it with any real data, but. I hope not. Just Brent's data. Right? Yeah. Okay.

Reasonable. Number four. This one, I think, is hard to wrap your head around initially. Number four. framework for scheduling. I don't actually know how you pronounce it. The AI version of Chris pronounces it wrong, and now I'm afraid to pronounce it. Is it Skedge EXT? Skedge External? Skedge Ext? Skedge Extension, I would think. The Schedule Extension functionality, which was initially planned in Linux 6.11,

actually now is landing in 6.12. It enables the schedule policies to be implemented as this is the big part, guys. BPF programs. So you can have your own custom scheduler for your specific work case. So you could see how Valve could use this on the Steam Deck to create a Valve-designed engineered scheduler specifically for the Steam Deck.

Or you could see how AWS or Azure could use this to have a BPF scheduler that optimizes resource allocation for different customer workloads. One thing that's kind of neat about this is by leveraging eBPF, you know, it's already designed to be implemented and loaded from user space. It's already got safety and like runtime checks and, you know, a lot of checking to make sure that you're not going to crash the kernel. So you get to reuse all of that for your scheduler. You know, so Meta did a case study. There's one of their largest workloads in their fleet. For that workload, how much work the machine could perform is determined by the 99th percentile request response latency.

could perform is determined by the 99th percentile request response latency. Usually, the CPU utilization hovers around 40%. So with that much CPU time unused, they were thinking like, there's got to be room for some scheduling improvements here. They wanted to know like, are idle CPUs hanging out when there could be doing work? Is a task waking up and looking for a CPU to run, you know, how do we choose the right one the most efficiently? So they played with that. They implemented several different schedulers using this technology. And here's their conclusion. This experience has clearly confirmed the benefits of Sched EXT to us. There's no way we could have experimented with an internal scheduler change this significant

in production. It would have taken too long and too much coordination and pain. Even if we could experiment and obtain the same results, the path of deployment would be too uncertain and arduous. With SchedXed, I spent a couple weeks writing SCX layered, which is one of the things they built, and another week testing it on the live production workload while continuously tinkering to obtain the result. That's what's so neat, right? You don't have to reboot. You don't have to do it. You just tinker with the little program you wrote. Also, because it's so safe,

we can easily expand the testing to wider scale and deploy as is. I didn't spend too much effort optimizing the configuration and the resulting configuration isn't very specific to the workload. The goal behind SCX Layered is to allow application teams to easily apply their domain-specific knowledge and experience to improve scheduling behaviors. So now like your infra team can set

this up so that you can, you know, like your workload, you can tweak the Linux scheduler. You don't need a custom kernel. You just load it in. And you go from like, there's like three or four schedulers you can pick from, and then you've got to have a kernel that's built for that. and then you've got to have some boot flags, and you've got to reboot every time you swap out schedulers. To now you can hot load a purpose-built scheduler that runs inside a kernel VM handling that stuff for you at super fast speeds. And you can use Rust to do it if you want. Oh, jeez. How about that? Rust to do it if you want. Oh, geez. How about that? So as we get along here, 6.12 starts to emerge as the kernel that really lets you build around your specific workload. And schedule

extension is a really big part of that because we covered when eBPF just came to the Linux kernel years ago, and it was really for more diagnostic type stuff. Yeah, right. I mean, we'd always been envious of things like DTrace, which turns out now exists basically as a wrapper on top of eBPF. So yeah, right. It was ways to hook in and add trace points and look at what the kernel was doing and have counters and stuff in a way where you didn't have to recompile the kernel to get extra information. Yeah.

This is a big feature, and it's not even in our top three. And this is one of those that's going to have long-lasting impacts on Linux in cloud workloads, in high production workloads, in gaming workloads. I mean, would it be possible for people to tinker with this kind of stuff to see if they can optimize battery life? I think it is possible. I just want to know if I can make a scheduler that prevents you from running Nano. 1password.com slash unplug. That's the number 1password.com slash unplug. You know, you really

have this problem in modern day IT. Users have their own devices. They have all these services they want to access. How do you make sure that your end users, and I mean always without exception, work on only company-owned devices and use IT-approved apps? Is it even possible this day and age? I mean, let's be real. In today's world, it's like herding cats, which they say is impossible. At least I've never tried. So you have to make sure that if your employees are using their own phones and their own tablets and their own laptops,

maybe their own smart fridge to access some sort of crazy like proxied calendar service. You got to make sure that your company data is secure. So how do you do that in this actual reality that we face? One password has the answer to this. It's extended access management. One password extended access management helps you secure every sign on for every app on every device. It solves problems that your traditional IAMs and MDMs just weren't designed to even touch. It's security for the way

we actually work today. It's generally available for companies with Okta and Microsoft Entry and it's in beta for Google Workspace customers as well. So support the show. Check it out. They've got a lot of great information and more at their website. So you just get started there. You go to 1password.com slash unplugged. That is the number 1password.com slash unplugged. Number three. Raspberry Pi 5 support. The Raspberry Pi 5 was announced on September 28th, 2023. It began shipping on October 23rd of 2023, so over a year ago as we record now.

The patches to support at least the basic SoC landed in the 612 development cycle on September 16th. So just before the year mark of the announcement. So to their credit, they did get the kernel drivers in under the year mark, but we're not actually going to get them shipping to us until now. And it's taken some time to just get the basic BCM2712 Raspberry Pi 5 support working. And more code still needs to be done. I think there's stuff that just has to be ironed out still to get all the functionality. But, like, you know, this is a big part of it. And what this means is you don't have to use the Raspberry Pi OS anymore.

You can use something that just uses upstream. That makes it so much more flexible. And it starts with 6.12. I feel like this is really the advantage that Raspberry Pi has going for it over some of the other SoCs. Like to have this baked in to Linuxes everywhere. I feel like that's such a beautiful thing. all the stuff in the 4 series. It made it such a more practical device for a Linux user because I'm not particularly interested

in Raspberry Pi OS just because I am running this headless and I kind of just want it to be similar to my other systems. So it's very nice to see that. There's a lot to like about Raspberry Pi OS, but what? They've got their own window manager now. It just gets kind of more and more of its own thing, which for a lot of folks

is maybe what you want, but maybe not for us. Number two. Crash QR support. We've been talking about this because we've been watching this. It landed August 29th of this year, 2024. This feature may actually be the thing that gets most people using Rust because the crash QR code support is written in Rust. And you've got an example right here up on your screen. Yeah, that's right. As part of building our custom kernel this week, we wanted to enable this stuff. So when you run into a

problem that normally you would get like a kernel panic stack trace written out on the text console. If you're lucky, it doesn't get cut off. Instead, you get a screen with a customizable colored background and foreground color. It's got a cute little penguin in the upper left-hand corner and an exclamation mark in ASCII art style. Of course, the main segment is dominated by a giant 40x40 QR code. And then it has a nice little message that says, Colonel Panic, please reboot your computer. And then, I don't little message that says, Carnal Panic! Please reboot your computer. And then, I don't know what the

full scope of what all it can show, but it's got a little line that tells you. So, like, I used the magic sys request trigger stuff, slash proc, slash sysRQ to trigger this. So it tells me sysRQ triggered crash. Well, that's kind of nice. Why Rust, Wes? Why do they have to use Rust?

You know, that's a good question. I think it just came down to... The developer wanted to learn Rust? Well, I don't know about learn, use. Yeah. It's a way to leverage this new functionality. It also maybe was just a little bit simpler to get the QR code stuff going in Rust. It does kind of push the Rust issue forward in the Linux kernel.

Yeah, because it kind of means this is one of the first things, you know, it's not, a lot of stuff has been pushed forward to develop drivers, but drivers are necessarily kind of often at least coupled to the hardware. So you're not going to use it unless you have that hard work or whatever. Which means it's not going to be needed to commonly enable a kernel

that has Rust so people don't figure out all the issues and, you know, push it forward. I think Enterprise wants this feature. Yeah. I mean, I think there's out all the issues and, you know, push it forward. I think enterprise wants this feature. Yeah. I mean, I think there's a lot of folks because you can, the QR code

can just be a data dump of like a lot of the data you would get in a kernel stack trace. Very dense QR code. But it can also have an inbuilt URL, right? So your corporate IT

or your Linux distribution could put in a URL that like auto starts creating a bug report from the kernel crash data. And your user can just like scan with their phone, you know, right there. There's also a connection here to something that people are developing in the background, which is disabling the virtual terminal in Linux. I'm sorry. Yeah. So it's called config VT. People are doing config VT equals no. And VT is at is say no to config vt equals no is what i say

i want a shirt so it's the virtual terminals that you're used to the text console enabled in the kernel but it's a lot of old crufty code and these days you could totally have a user space process that can be your console you don't necessarily need this you know the kernel to do that it can't do a lot of modern features in terms of like, you know, GNC font support, even just like any kind of modern stuff that like a regular terminal can do. It does not have support to do that. I think it's made some things painful

with like getting the real-time stuff merged, you know, various other improvements that kind of have to route around this stuff. So there's desires to remove it or be able to turn it off just to simplify the code base and improve multi-seat capabilities, enhance security. Yeah, I suppose it would.

So that's where the DRM panic stuff comes in. You could have the terminal turned off but still get that output. Right. So you still wanted some kind of output to display the stack trace if you don't have a text console to display it on. Yeah. So this ties in with that functionality and extends it with additional capabilities. Yeah. play it on. So this ties in with that functionality and extends it with additional capabilities.

Yeah. The developer who created this does work for Red Hat. And they say the mission to enable config VT, the virtual terminal is what you were just discussing here is that broader effort to modernize and streamline the kernel's display and input handling system. So you can make it more secure, but still get more information. And then there you can tune the kind of information that QR code supplies. So you could see maybe if you're in a very sensitive environment, it's pretty minimal. But if you're in a trusted environment, maybe you get, you scan that QR code, you basically get all the information right there on your device. Yeah. Maybe you're like lab, right?

You want a whole bunch of information. So you get the quick to debug as you're trying to get your new kernel version or whatever. Lab machine. Sure. Yeah. Yeah. It's fun to play with. I will say it's early support, right?

It was just merged, so not a ton of DRM subsystems have support, like the i915 Intel driver did not. So I had to set up a virtual machine using the simple DRM driver, which is built into the kernel. So I think in a lot of cases, depending on your hardware, that's why I did a VM just for ease, but if you blacklist your existing driver, or you don't let your actual GPU driver load, then you can probably get the simple DRM one to do it. It's a little fussy. So when you were testing all this, so you're like, first I got to get a kernel, and I got to get all this, and I got to get it working and building, and that takes forever.

And then you're like, once you get it all working, did you then realize, oh, I have to now crash this thing? Or did it occur to you the entire time that you were going to have to crash this? Like, at what point did you realize, oh, now I have to figure out how to make the Linux kernel panic? Well, I had that realization did happen kind of in the middle. Because you were, yeah, you were setting up. Right. Okay.

Also, when I was like, oh, yeah, I guess I'm doing this. Well, I'm also using the RC with this file system. It's, you know, experimental. Yeah. Yeah. I'm sure it'll be fine. But really, you actually don't have to crash it. You can. I wanted to because that's experimental. Yeah, yeah. I'm sure it'll be fine. But really, you actually don't have to crash it. You can. I wanted to

because that's one of the main things that I'll use. But if you enable it, which is another kernel option this patch added, under sys debugfs, there's something you can, like, echo 1, 2, and it'll display the, basically do this thing it would do if it crashed without actually

crashing your system. Number one. And now we get to the big one. Preempt RT landed September 20th after 20 years? Seems like, yeah. Of development? I mean, seriously, guys, this is a massive one. Yeah, I guess it was, what, 2004-ish or so when it kind of started. Wow. So real-time kernel support has finally been merged as of Linux 6. Yeah, I guess it was, what, 2004-ish or so when it kind of started. Wow. So real-time kernel support has finally been merged as of Linux 6.12

and it is available on ARM64, RISC-V, x86, and of course x86 64 devices right now. And I guess the pre-advantages the pre-advantages, haha, the pre-empt advantages are it ensures consistent response times by making

the kernel fully pre-emptable, critical for real-time applications. It converts interrupt handlers to preemptible kernel threads, which reduces system latency spikes. So I kind of understand all that. I kind of get it. And yet I didn't expect it to make the difference it did. And I don't know, Wes, where do we start with preempt RT? it did. And I don't know, Wes, where do we start with PreemptRT? Because I go back to our conversation we had about the Mars rover and how they were custom patching their Linux kernels to enable PreemptRT because they needed it in the space application. Yeah. You know, there's a lot

of things for Linux where Linux is just sort of default optimized for throughput. And, you know, when you're deploying servers, that's often what you want. The response time is, you know, you need it to meet some threshold, but it's not like a super tight boundary. Not life or death or, you know, noticeable even sometimes. But if you're running a, you know, billion dollar rover or a helicopter or a CNC machine or medical equipment, you got to make decisions and you have very tight and consistent timelines with which you got to do that decisions and you have very tight and consistent timelines with which you've got to do that,

otherwise you're outside of safety margins. And by default, the Linux kernel was just not meant to do that. There are dedicated real-time operating systems, some of which you might even still need. This would not get you all the way there if you need sort of like, you know, really tight minimal code base that's optimized for exactly that situation. I think one of the reasons why Linux kind of lagged initially

in the automotive industry and things like QNX succeeded for a while. Right. But there's a lot of workloads kind of in between there, right? Like it needs more than the stock kernel can do, but it doesn't need this like dedicated, perfectly real-time operating system that's had proofs worked out about it.

And that's where the preempt RT patch set kind of came in. Right, okay. And, you know, these days, over time, we finally saw, like, the bulk of it actually merged and available and the preempt RT option, the config option is now in 6.12.

But, you know, it was a gradual process. So the kernel's gotten more and more preemptible over time. A lot of things had to be tweaked or changed. Yes, a lot of things had to be tweaked and figured out, things that you might not expect. And it means like, even if you don't go full RT, like you can still go, I think a lot of stuff, the kernel comes compiled preempt dynamic,

which lets you sort of tweak how preemptible it is at runtime. But you can also configure it to be more preemptible. And then RT kind of like takes it to the max. Yeah, yeah, takes it to the max. Yeah. Yeah. Takes it to the max. It does. You did some testing because of course we're talking about space and mechanical operations, but in media production, real-time kernels matter as well, especially when you're applying effects and processing. Yeah. Or, you know, in live music production. Yeah.

Things where, you know, you got to be- Like we're doing right now, actually. Yeah. Look, we're doing it right now. You got to be consistent. You have deadlines because you need to fill your buffers if you're going to go send out the next waveform

or the next packet onto the network. So you did some audio sampling so we can hear the difference between a real-time Linux system and a non-real-time Linux system. So here is the version that has non-real-time. So this is just stock 6.11.7. Stock Linux, and you get what's called sometimes X runs, and you'll hear them as almost like packet drops in this audio. Take a listen. BcacheFS has now joined the party of various kernel components,

sending in preemptive pull requests ahead of the Linux 6.13 merge window that's expected to open following the Linux 6.12 release, which might be as soon as this coming Sunday. So you may have heard those pops in that audio. In fact, if you've been listening to the show for a while, every now and then you may hear those pops in the actual released audio. Yeah, right. So like the kernel has to shuffle data around, right?

It's got to go read data from my interface where it's coming from the microphone and the digital conversion. And then it's going to go shovel it to Pipewire and Pipewire has got to shuffle it around your graph and feed it to Reaper or Audacity or whatever is happening. And that's a case where, you know, if the kernel got distracted because it was trying to write some bytes to the disk and it didn't meet that timed deadline, you don't have any audio to write. So you get that. You get what is essentially a packet loss. And for context there, I was compiling the kernel. So that was maxing out all of my CPUs. I was also doing a ping flood on the network stack. Yeah, keep that subsystem busy. And as you heard at the very start of the show, that was me spinning up a whole bunch of DD processes

writing to the disk. Yeah, one of the things I've definitely found spinning up a whole bunch of DD processes writing to the disk. Yeah. One of the things I've definitely found in media production is it's generally when multiple subsystems in Linux are stressed is where you start to notice this. And so if you've got a lot of throughput, that can happen. Now, here it is again, the same audio, but with Linux 6.12 and preempt RT turned on. Yeah, I didn't change anything else. I just rebooted and chose a different kernel when I booted. In addition to BcacheFS working its way through its remaining bugs, well, for Linux 6.13, there's some more feature work in particular around self-healing support, enhancing RefLink repair capabilities, and improving option handling on the rebalance data path. Oh, not a single drop,

right? Oh, it's so great i cannot wait to get this on our production systems there might have been one or two over like the whole testing that i was doing on the rts not like it was perfect because i was really abusing my system uh but yeah way more consistent way easier to to use and interact with wow you know the funny thing is i was gonna say hey there was one there near the end, but I think that's actually your live kernel sending one to me here. We got to switch this thing. I know. Right. It also made me aware of a PW top, which is pipe wire top and cool. It lets you see those X runs. Oh, so it made it really clear when I was on the other kernel, even when I wasn't recording or using Reaper, you know, doing anything with audio, the X runs were just steadily increasing in there. And when I was on the RT kernel,

no, they were pretty much static. So I just, um, a nod to the decades of work that have taken place to make this possible. Um, Wes did a fantastic job of breaking down the timeline in our show notes and, you know, it starts, the journey starts 2003 2004 really till today and um we had a chance to load it on our systems we wanted to see if we could notice a difference using a preempt rt kernel on just a desktop load and i wasn't particularly interested in even trying it i have to be honest with you because i've been told it really doesn't make a difference. And I've been told that it's kind of all just a bunch of hype. And so I wasn't particularly, I didn't even feel like it was necessary for me to try it to report here on the show because I didn't think I'd be able to observe anything.

Wes, though, I'm very, very happy that you had a different attitude and you went ahead and tried it and you then dropped a little hit in the chat that, hey, I think my desktop feels faster. That's all I needed, buddy. Yeah, right. That got you. So Wes very kindly built a Linux 6.12 kernel for me and created a little flag for me to get it set up on my NixOS system. What were your first impressions of a desktop with preempt RT?

It felt snappier. Yeah. You know? Everything. Yep. Firefox in particular felt snappier. Yeah. You know? Everything. Yep. Firefox in particular felt snappier. Changing tabs just. Even the freaking tab changing is snappier.

Yeah. And a lot of the little KDE things also snappier. Console's faster. Yep. I mean, I'd never complain about it being slow. It's faster. It's faster in the way like I was talking about a couple of weeks ago. My MacBook feels a little bit faster.

It's just everything's a little snappier. I think you can tell that. And again, what we haven't tried is taking a stock kernel and applying all the techniques you can do without having to rebuild it. This was just like stock, stock. But I think you can tell that the kernel is doing a better job of always being available to do what needs to happen immediately and not finishing what it was still doing. I mean, I don't know if I can convey how surprised I was that I noticed a difference in my desktop workload. Like it's still kind of subtle, right? It's not like a radical change. But if you're using your desktop or your workstation day to day, I think you'll notice.

It's like the tabs are just a little snappier. I couldn't really give you a full review because somebody over there decided to uncomment all my desktop applications. So I was like, I've got nothing to test. I don't have any of my apps. I don't know. So I was like, I'm using Firefox. I was using Console and Dolphin. I'm like, okay, this feels pretty good.

I was hoping you would uncomment them. That was just an effort because the first approach we were doing, I was sending you a full system closure. So I figured I didn't need to also send you a whole bunch of electron apps. Yeah. Well, I thought I'd open it up and I thought the only thing that would be commented would be nano and that Vim would be in there, but no. Dang it. I missed an opportunity. You did. It's pretty great. And I'm not going back. And I don't know how much of a headache I'm creating myself by now saying that I'm a preempt-only kind of guy. But once 6.12 is officially released, I'm going to figure out how to run a real-time version on NixOS.

And my desktops are going real-time only. And our recording systems. You could hear the difference in the audio when the system's under load. It's getting me really excited to spend some time redoing the studio. 6.12 is literally a game changer for us, but there's so many things. This top six is the literal definition of

scratching the surface. Michael Larble has done an unbelievable job of trying to document some of the features. I would imagine Cardinal Newbies will eventually once it's out too. It's a massive task to cover everything in this release and it's i think brent you said it best it kind of puts all other free software project releases

into perspective yeah it took minutes and minutes and minutes to just go through like the summary of the feature changes and i know this is an lts so i imagine they're squeezing a lot of things in but just to i mean this episode this whole episode kind of suggests that the number of like wildly amazing features they're bringing to a single kernel release is is unbelievable and i know there's a lot of people working on it so it's maybe not fair to compare that to smaller projects. But geez, they are doing amazing work. And we are lucky as Linux users to benefit from that work really down the road. Linuxunplugged.com slash membership. Support the show directly and we got a couple of perks for you. Number one, you get the ad free version of the show if you'd like. It's the same show, nice and clean and tightened by Drew, but none of these types of ads. Or the second option is the full bootleg. Now that's probably

double the show because we put a lot of stuff in there that doesn't make it in the main show, the stuff we trim so we can focus the main show. Like in the pre-show this week, we talked about the new Fedora Linux installer that's landing in the next Fedora. We talked a little bit about Bluefin 41, a RIS risk motherboard for framework, and then some back and forth that Tuxedo Computers has had with the Linux kernel regarding the licensing of their modules and the resolution they came to. That amongst, you know, of course, just gassing and getting together with friends, that's all part of the pre and post show that you get as a member. And it's just a way to say thank you. You know, I realize that we are entering a phase now where

you're probably getting tired of hearing me ask for support. I probably would be if I was listening, but we are entering an era where I just simply have no other option. We're going the way of Linux magazines. But thankfully, podcasts, we don't have this huge, massive cost of print infrastructure and mailing out magazines. We don't have that huge overhead. There's a solution here, and it's direct funding from our audience.

We run a lean, mean operation. I think you'd be shocked. Well, maybe not. If you're a listener for a while, you probably have an idea. I mean, if you're like a – if you just – we're another business owner, and you saw what we got done and the budget we get done. The magazine publishers could never have touched that. And so we have a real special opportunity here. That's why your direct support matters so much. We have an annual membership that we'll have linked in the show

notes that gets you support to all the shows. Of course, linuxunplugged.com membership just to support this show. Or you can boost in and then you can support at your own cadence amount on your terms as you like, and you get a message right on the show. So there's a couple of options there, and we're really appreciative of whichever one you choose. And of course, we always are grateful if you just spread the word and help grow the audience too. LinuxUnplugged.com slash membership or send a boost with a new podcast app.

You can get those at PodcastApps.com. Well, while you guys are away playing with kernels, I got this giant bag of boosts here. And now it is time for Le Boost. Doma, when the walls fell, boosted in with 48,197 sats, and he is

our baller this week. Hey, rich lobster! Got that baller on sale. Good for you, boy. It's a day late and a dollar short. Finally got my booting issues resolved. Anyways, my first experience with Linux was SUSE 7 back in high school. Yeah, buddy.

Nice. Shout out to Mr. Halle. He made a huge impression on my life. Oh, a teacher? I had a really great high school teacher as well that was a Linux geek, so it was really nice because it was him and I against the Microsoft Brigade. Wow. Oh, look out, Wes Payne.

I got news for you there, buddy. It is a zip code boost. Yes, zip code is a better deal. All right, you want to try to find it? I do. All right, get them in. Oh, this is a fun one, I can tell. Oh, there it is. All right, we got to try to find it? I do. All right, get them in. Oh, this is a fun one, I can tell. Oh, there it is.

All right, we got 48197. Yep. And I believe that is a zip code in Ypsilanti, Michigan. Hello, Michigan. With what a name, Y-P-S-I-L. Whipsy. A-N-T-I. Whipsil?

Whipsitilly? Where are you getting the W from? I don't know. I like to whip it, I guess. The internet suggests ipsilanti, but dear booster, you'll have to boost back and tell us how we did. That's got to be one of the trickier ones. Thanks for the zip code boost. We love these.

Yeah. Thank you, Mr. Walsfeld. Appreciate it. Timbo boosts in with 27,000 cents. I hoard that which all kind covet. Keep up the good work, team. I'm sending an Australia postcode boost. Just subtract 21,000 cents.

Yes, zip code is a better deal. Wow. Well, this is an Australian zip code. I can tell you that much. Yeah, gents, it's called a postcode. True, true. I think a postcode is a fancier way to say zip code i do like it yeah but our postcode's a better deal it's more it's more that's true okay looks like that maybe is in perth oh fun is that our first perth perth western

australia that's awesome you know i feel like we have a kinship with western australia because they're wa and here in washington state we're wa yeah i was know, I feel like we have a kinship with Western Australia because they're WA and here in Washington State we're WA. Yeah. I was thinking the same thing. We have a boost here from Grounded Grid for 2,600 sets. Everything's under control. Hey, last episode, Chris seemed surprised that nobody might use a Pi as a desktop. So I thought I would share that for over two years, my primary system is a Pi 400 booting Arch off a 5 terabyte Western digital passport.

And I love it. Eventually, an X11 update broke the system, but I ran it with a rolled back version of X for another six months. I'm not a power user. And the system was actually a big upgrade for me. I mean, you rolled back your X for six months. That's kind of a power user. And the system was actually a big upgrade for me. I mean, you rolled back your X for six months. That's kind of a power user move. You know, if anybody was going to do it, they were going to do it with a Pi 400. Because I used the Pi 400 for quite a while as like a standby system, always ready to go with the terminal.

Because every now and then I'd show up and like my desktop wouldn't be working. Or I forgot my laptop when I was only using a laptop. And it's like, oh, crap, I need a system to SSH or to do a calendar thing or whatever. And the Pi 400 was so great for that. So I could believe it grounded, but I'm really impressed. You managed to stick with that for two years at five terabyte external hard drive probably helped quite a bit. I bet. That's amazing. I wonder, you know, I wonder if you could do it. I wonder if you could do it today with a modern distro. They're a little heavier. I have only one question. Okay.

How many tabs can you keep open? That's a good question. That's a good question. You supposed! True Grits came in with a Jar Jar Boost, 5,000 sats. You know, now that you guys don't livestream on YouTube and Twitch anymore, you could bring back the pump-up music idea again. Maybe create a specific pump-up song using AI like you've been doing with these other songs. We do play a little more music.

You just gotta show up on the live stream. You gotta get up here a little earlier. Or the members feed. Yeah, yeah, that's true. That's true. Yeah, we love the music. You know, because it gets the mood going. We were looking for a specific pump-up song for a while. Which we never... No. Ironically, we ended up on a song to playup song for a while no ironically we we ended up on

a song to play at the end of the show when we're voting after when we're all done we ended up with a voting yeah we got the opposite of what we were looking for i think the voting's going well so maybe that's yeah yeah maybe maybe i will say i think that whole pump-up song journey was really fun to hear from the audience like all these weird, esoteric songs from all over the world. So thanks for sending those in. And if you have others, we'll hear them. Please. Yeah. Somewhat Justin boosts in with Spaceballs Boost.

So the combination is 1, 2, 3, 4, 5. Oh, okay. This is Rename. Booster formerly known as User39 something something something. Yeah. Checking in. All right. Hopefully my username is actually fixed this time.

Hey, well done. Yeah, we can confirm. This is the way. Last week, I asked about desktop hardware and loved your suggestions. I was very close to getting a B-Link until I spotted an excellent deal on Facebook Marketplace for an SFF PC with a modern Ryzen 9. Ooh. Nice.

My seller was excellent, and I'm very happy with the purchase. It can be a small gamble, but in my case, the seller did this as a hobby and was very thoughtful. Nice. For the Tuxes, have you considered adding a category

that's about the show itself? You know, like best episode, best interview, best challenge, best host, et cetera, et cetera. I've never considered that because we don't really like

thinking and talking about ourselves much, but it would be good information for us to have. True, yeah. Yeah. Hmm. Hmm. Good score on the machine, by the way. Well done.

Good move on that somewhat. I think that's a pro move of the week right there. I will think about that for the tuxes. We do need to get cooking on them tuxes, so you're going to hear more about that real soon. We have an excellent boost here from LegalMTMode2601Satoshis. Welcome. Glad to hear from a fellow tab hoarder. I exceed 2,600 tabs the other week.

Whoa! On one browser, on one computer, I did not bother to calculate a grand total across all my devices. Yeah, because you're going to have to use floating point. Yeah, no kidding. Who has the time? Tell you what. Oh, boy.

Here it is, everybody. Gene Bean's back. 17,238 sats. All systems are functional. He's got a whole row of boosts catching up on the shows. He enjoyed the message. How many is that across there, sir? 14.

I'm a pro summarizer, though. Watch me go. He says he enjoyed the meshtastic episode, and he's tempted to get started in there. Brent, by the way, you could try a bookmark manager to solve all those tabs. Which one? Well, you decide and come back to us. I'll do a little research in a set of tabs. Wes, Gene enjoyed hearing your mom's journey to her new laptop.

Aw, thanks, Gene. I liked that, too. Also, we heard us mention the call-out for the scale call for papers. He's going to think about putting a talk in and looks forward to seeing us there. Yes! Love that. I'm really excited to use more determinate nicks. It sounds awesome.

Regarding Linux for a very young child, I'm considering it more now because my Chromebook at school as part of your kid has a Chromebook. So, Gene, my kids all got Chromebooks, and I still decided to give them laptops to go alongside their Chromebooks. And I think that was a good decision because when they get home, they basically don't even touch the Chromebooks and they even prefer, if they can, to do their homework on their actual laptops and they can just sign into their Google account for school and they use their actual laptops. So I say do it because now they're getting introduced to the concepts and features. Solved and figured out how they need to do the stuff they got to do. He's got ideas for Planet Nix for you and Alex. Alex should talk about his single flake for everything and Wes should do an introduction for packaging an app for you and Alex. Alex should talk about his single flake for everything, and Wes should do an introduction for

packaging an app in a flake. Ooh, yeah, okay, I like that. And the longest distro was Mint as his go-to for years. The longest continuous server was CentOS. Nobody sent in their server OS. Thank you, Gene. We should have asked that. That's a great point.

My longest continuous desktop has to be a tie between Gentoo, of course, and Pop! OS. Oh, yeah. Pop! OS is really making a name for themselves. And tips us on the taking over Brent's laptop. And everybody loved the takeover of the laptop. Not everybody loved us. A lot of people just wanted us to fix the package manager.

Yeah. Fair enough, I guess. I don't know if that would have been as much fun, but... I've been fixing the package manager for like six plus months. I thought that was a little old hat. Listener Jeff sends in a binary boost soundbite audition. Should we hear it? Yeah. This should be interesting.

All right. Oh, and he sent in 10,110. Odd. Okay, here we go. Ones and zeros everywhere. And I thought I saw two. That's good. That is pretty good for a binary boost.

I think that's a winner. Thank you, PJ. Odyssey Westra boosts in with 5,000 sets. You're supposed. I have to say, Fountain is working quite nicely. Oh, good. I had an audio bug, but they got back to me with a fix in the 1.1.8 release. With that

said, I would love an ability to add a song to my library from the player screen or a quick add of the track list of a music podcast to a playlist. It would also be nice if posts would format better on Noster overall. It looks nice on Fountain with linking to clips, podcasts, or songs, but on Primal, for example, it just shows the boost bot, which isn't helpful.

Also, cross-posting to Activity Pub would be awesome as well. Noster.pub by Soapbox could be a way to do it. Either way, I'm excited about what Fountain is doing and what it's going to do down the road. Me too, Odyssey. And I think it's gotten incredible. But I am still asking for people to boost in if they're having any problems with Fountain because the team's going to be reviewing the boost and we'll create a new action list from that. But the Nostrand stuff does seem to be coming along quite nicely.

I could see some tweaks to how it gets shared. Yeah, that could be really nice. Thank you, sir. We have a Spaceballs boost from Biggles. Yes, that's amazing. I've got the same combination on my luggage. Keep up the good work, and hello from Dublin, Ireland. Hello.

Well, hello. Thank you for coming in. I've gone way down the rabbit hole on NixOS thanks to you guys. However, NVIDIA and CUDA support for my AI workstation was a real challenge until eventually I found a Mamba, an efficient implementation of Conda. Well, there you go, Mamba. Interesting. If you have any of your stuff up on Git somewhere, maybe boost us a link. Thank you, Biggles. Todd from Virginia comes in with

11,101 sats. No message, just the value. Pure value. Seems like a binary boost to me, if you look at it that way. It could be. It could be a binary, but it's not today. Ow. What did PJ do to you?

Zach comes in with 10,000 sats. It's another binary boost. Okay, all right. Okay, fine. I don't even remember where I had it. I don't even think I saved it. No, I'm getting it.

I'm getting it right now. It's a binary boost. Ones and zeros everywhere. And I thought I saw a two. All right, that's pretty good. Congrats on getting Brent's laptop fixed. Though I was kind of curious on what it would have actually taken to fix the Tumbleweed zipper issues. Yeah, I know.

I get the impression that Tumbleweed really does not want you to add more repositories and just keep what is already there. That's part of why I eventually jumped off of it. Maybe leave the repositories alone and just use Flatpak and App Image to fill out the rest. That's my playbook right there. And maybe Nix. Though at that point, might as well use Fedora or NixOS. And speaking of Fedora, their KDE spin is the closest I've seen to tumbleweeds, where it just feels like a first class citizen. Good to know. Yeah, I want to apologize to the Soos fans out there that we're actually hoping for an episode where we just fixed his current install and what i don't think we properly conveyed was that we went into it knowing that brent was ready to migrate and so that wasn't really on the

table and it was more about like what is the fun kind of hardest way we could migrate that laptop and make content we wouldn't we wouldn't go looking for hard fun things, would we? But people tuned in expecting us to fix the existing install, and that's sometimes a great option. And so I apologize to those who got frustrated with that. But yeah, as Brent was just saying, right, like it was, you have been through the, we didn't share it on air so much, but you'd been through the cycle of like, should I keep it?

Should I keep it? Should I keep it? Should I keep it? Should I keep it? Should I keep it? And that's why you were ready to migrate. Well, actually, we did share it on air like a year ago. We did an episode where we had a little tiny segment where we fixed my tumbleweed.

And I think that was maybe even in the title. And so it's not a new journey. I think we could have done a slightly better job explaining that path that I've been through and where I was just at this point, just kind of ready to move on. Yeah, but it was fun. So, and now it's been working solid. I hope.

Oh yeah. We haven't heard anything. So I mean, I have to install a bunch of stuff, but it's been ideal. Now you're making me feel like I need a new kernel though. So we'll see. Yeah.

Yeah. We got to get that thing RT. But you might wait for 6.12 to actually, you know, to this afternoon when it actually comes out. Just so the audience knows, I did offer both of you the kernel. Yeah, that's true. And I, first Brent was the taker. But then once you started talking desktop performance, I went right to the front of the line. Yeah, and once you said desktop performance, I was like, ah, Chris has this.

Yes, please. Like, ah, Chris has this. Yes, please. Well, we got SatStacker7 sending in 4,444 boosts. That's across two boosts, some rows of duck. I fight things that are looking up for all the duck. I got one feedback and one question here. You mentioned Mealy in the last episode. I use it daily and I find it amazing.

in the last episode. I use it daily and I find it amazing. Specifically, the scraping feature that collects all of the steps and ingredients from a blog or recipe website. And that works fantastically. By the way, there is some standard that most websites follow that makes this much easier.

It's defined in schema.org slash recipe. We heavily use the meal planner feature to plan the entire week's dishes and prepare for the weekly grocery tour. Nice. That's the good feedback. I've wondered, because I knew about this schema, I've wondered if I came across ones, if Wes and I could

create a script that would reformat it to match the schema and then import it. Oh yeah, why not? Because I've got a bunch of markdown files and I'm wondering if they could be schemed up to be imported. I bet there's some chatbots that could help, too. Oh, yeah. I bet you're right. Now here's the second boost for the question.

I'm using Fountain for boosting. To my knowledge, it only allows using its internal custodial wallet. However, I've heard from other boosters that were able to connect their own node via Albie Hub. I'm running an Albie Hub myself. So how can I use it for boosting? Do I need to use another podcasting app? Any hints would be appreciated. Well, first of all, well done getting Albie Hub going.

That's a freaking fantastic. You are already way ahead of the game. And you got it. You figured it out. So the nice thing about Fountain is that it's all just kind of managed for you. And they just host all of that. And they do all of the plumbing. But you can't plug in your own self-hosted.

A lot of other podcasting apps can. Podcast Guru, if you're on iOS, Cast-O-Matic definitely does. Podverse will work with your Albie Hub. There's a lot of them, actually. I think even TrueFans, if you want to use the TrueFans app, that'll work with your Albie Hub. There's a lot of them, actually. I think even TrueFans, if you want to use the TrueFans app, that'll work with your Albie Hub. And on the desktop, if you get the Albie extension in your browser, then Podverse or Podcast Index are places that you can do it. Yeah, you can boost from there, from the web.

Fountain has a boost via invoice functionally. I think right now it's only enabled for live streams, but in the past it worked more widely, so it may be another option. So the way I do it is my hub is my main Lightning node, my Albie hub, and then I just, you know, every couple of weeks as I kind of go through my budget, I just send some sats over to my Fountain wallet.

And I just kind of keep a stash in there, and that kind of keeps it separated, and I don't mind that at all. You could do the same with Breeze, if you liked, which is a similar setup, but it's technically non-custodial because they run a node in the app, but there are fees involved to pay for the liquidity that they provide.

Yeah. It makes it very easy to boost. Yeah, and then you don't even have to use a different, you could keep your podcast app and just use Breeze to boost, because it's just pulling the feed and boost from there. Good question. And SatStacker, it's nice to hear from you.

Thank you very much. Slash home slash JLA comes in with 2,000 sats. I don't know. JLA? Do you think maybe it's a JLA? I like JLA. JLA. And they write, just got NextCloud 9 installed using the web installer.

It seems solid so far. How's it been working for you guys? Up and down. Well, we have a couple of different Nextcloud instances, so it kind of depends on the one you're talking about. Right. Mine that I just reset up at home using the NixOS module is working like a freaking champ.

But our Nextcloud that we use as our daily production back end for the network, which has been in there for like four plus years now, just recently has been giving us issues specifically with retrieving files. We'll upload files. It'll look like they're there, but then we can't pull them down. And we suspect it has something to do with the fact that we use S3 object storage as our back end storage for NextCloud. Yeah, they ended up being just zero bytes in S3. Yeah. It's strange.

The UI shows the file there, but you check the actual bucket and the file's not there. I did some database cleanup and just general maintenance on it and updates. And since then, it seems to be doing better. But we'll see if it's actually fixed or not. We're still tracking it down. But that's kind of a particular setup, right? Because we're running on a VPS. It's using S3 object storage as it's back in.

On my home setup where I've just got a Postgres database and I'm just using my local file system, it's been super good. I think Brent just has a new one too. Weren't you playing with the Nextcloud Nix stuff, Brent? Well, I'm playing with a few in a few different places. I now have a new home server as as of what was that last week? And my plan is to host an extremely private, uh, next cloud here, literally in my cabin in the middle of nowhere. Um, so I'm playing with it there and I have had a test instance on a VPS, uh, for many months now that I've just been kind of like, don't do this, but I've been handing out credentials to friends

to just like bang on that thing and see what they think. Um, and that thing's been super solid and extremely snappy. So I am looking forward to implementing that, but I will say the next cloud I've had installed for five to six years now, based on the snap system, just got, uh, next cloud hub nine, I don't know, a couple weeks ago. And it seems a little faster in the web interface, which I'm happy to see. The Snap's always been a little slower for me in this particular setup.

So nice to see a little bit of speed there. I do agree. I think even on our shared instance that we were complaining about, the web... NextCloud Hub 9 is the happiest i've been with the web interface yeah yeah i'll co-sign that chatty mike chats in with a row of ducks hello from scotland hello listener from the lunduk era don't borrow chris's shaver yeah good advice

good advice yeah boy that's a good callback right there. What's your view on the Knicks community goings on? Oh, good question. You know, it's interesting to see how this directly impacts some folks and seems to completely just wash over and not impact other folks. And I've heard it cited as, you know, concern that the Knicks community cannot go on. And I've heard it cited as absolutely no big deal. I think I fall somewhere kind of in the middle. You know, I am concerned about a project that could be overrun by moderators, but I also want it to be a welcoming place for as many people as possible that can work together. And that's a tough line to strike, having run a community for, you know, almost 19 years. I can

definitely say we've been on both sides of that and made mistakes before. So those things take time. And as a Linux distribution grows, it often goes through these kind of dramatic community events that seem like they're just the absolute biggest deal in the world. And then you fast forward a few years later and people have completely stopped talking about them. And you see it for every major distribution that's come along and really survived is they've gone through this metamorphosis stage where they have kind of pushed through a moment in time and they come out a little bit different creature on the other side. And I think because Nix has been around for so long, what?

2004? Early 2000s? 2004. It has the staying power to survive this. And now we also see commercial implementations, enterprise implementations, and the user base has grown so exponentially in the last three years.

There's a lot of inertia and a lot of folks riding on it now. And people recognizing the value and understanding it better now because the market need has met the product match. There was a product out there for a long time that didn't have a market need, but now there is a market need and the two have met.

And I think that gives it staying power. There's also, you know, it can be hard to judge from afar. You know, you see a lot of discussions that get a lot of attention. But I think it can be hard to figure out how that actually does impact you, right? If you are relying and think about Nix as like a giant package of up-to-date libraries and you're worried about the maintenance of that dropping off or something, I can kind of get the concern. But if you're using it as a build tool or, you know, you're comfortable doing a little bit of stuff yourself, because Nix is so flexible, right? Like even if someone hasn't maintained, is a little behind on the

maintenance of a package, if you can just update the pin and rebuild it yourself, Nix is still perfectly usable for you. So even if that happens, I think Nix will still be super useful. And it can be really hard to tell, even as a project may lose prominent members, you know, how does the growth balance that out? Does it? Does it not? This is the stuff that takes years to settle out. So we're still watching and having a lot of fun using it.

Yeah, I mean, I'll keep an eye on some of the forks too. And it could be something to test in the future. You never know. But yeah, that's the other part. It seems like even a lot of the folks who've had particular, you know, who've been involved, some have left. But I think for many of them, the ideas rooted in Nix, like the approach, it's hard to shake. So like the drive to make other tooling in the space or, you know,

cousins or other things in the family. Yeah. So, you know, humans do. Yeah, exactly. I will say,

if you do need a little bit of a primer on what chatty Mike is asking about here in the next community, we did do a, an episode on it. Five, six, three Linux unplugged five, six,

three. If you want to check that out, Nick's people problems. Now hybrid sarcasm sent in, I think I'll call this a co-boost for 5,000 sets. Oh, my God. This drawer is filled with Froot Loops. Yeah, he wants to pull forward BitTiger. Yeah, Hybrid Sarcasm says, I want to get you reading BitTiger's boost.

Yeah, BitTiger came in at 100 sets, which was below our, obviously, below our 2,000 sat cutoff. But we'll read it because hybrid sarcasm boosted in as like a co-support. I like that. We do what hybrid says. Hello, Chris Lass and the Jupyter Broadcasting crew. It has been a long journey getting here. I have been a longtime listener since the early days of Linux Action Show. I was finally able to set up Fountain FM and get my Lightning Wallet.

Hey-o, congratulations. Now I'm earning income purely on Nostr, creating original, spicy, thought-provoking content. I wanted to thank you personally so much over the last 14 years for introducing me to Nostr, Linux, Bitcoin, and countless other open-source software through all the Jupyter Broadcasting shows.

You've been a positive influence in my life. I wouldn't be who I am today without Jupiter Broadcasting. This will definitely be the first of many booths to come. Keep up the truly meaningful work. Wow. Well, thank you very much. Double applause there. Really.

People are listening. That is really nice to hear. And, you know, BitTiger, I think you're kind of on it. Sats are scarce, and you're getting even more scarce. But there are ways today to earn sats, like Noster and people zapping your content. I think also, BitTiger, if you haven't, you should check out stacker.news, because same mechanism there. People can zap. So instead of an upvote, they have a sat zap.

So you zap somebody a sat, and that's considered an upvote, and the best content rises to the surface. That's stacker.news. And then if you're really crazy, you could try the fold card, which will give you sats back on your bill pay and stuff like that. But that's sort of the full trifecta there. And BitTiger, thank you so much for listening and taking the time to get all that set up. I know it really can be a bit of a journey,

and it's really special to always hear from you long-timers. And thank you, Hybrid, for co-sponsoring that boost. Really appreciate it. Brad comes in with 11,000 sats. Just pump the brakes right there. Brent, how have you found NixOS configuration editing? Has it been problematic, or does it feel like a new home? That's a great question.'s a good question i think this is in reference to this computer i'm sitting in front of

just getting new nix os on it and therefore i need to edit that config i think wes you probably have a few tricks here i think i'm going to name the ones that i think are really important um i think linking your configuration outside of etsy so that you can modify it with just your normal user instead of having to be a super user every single time you want to do an edit is key to making it smooth you know whenever you run into oh yeah i think i want this app and you just want to go change your config quickly sure sure is helpful to not need to enter your password every time you want to change the config. Yeah, you could change the, you know, chmod folder, etc folder if you wanted to, but NixOS rebuild is set up.

If you symlink your flake.nix, it'll just follow that. So you can have a git repo in your homeder or, you know, homeder slash source or whatever you do, and the system will just pick it up. Yeah, you can then use your favorite editor. I lean towards Kate. I've been using Kate for many, many years. Chris, you were running into an issue here where Kate wasn't too happy doing super user edits in your config. I remember that.

I think you solved that now, but Kate does some highlighting for Nix specifically, which is really super helpful. Just makes it a little easier to browse through there. And the other trick is steal from your friends. Like I'm stealing little code bits all the time and throwing them in my configuration file from Wes, from Chris, from all sorts of people. So, you know, don't simply try to figure it out yourself. Go ask the community. We got a Nix room. You can ask a bunch of stuff.

If you're trying to get an application running, often it's just a copy-paste away. And a lot of people post their configs to GitHub. You can search around in GitHub and find a lot of example configurations. Yeah. Alex from Self Hosted, our dear Alex, he's got a great config up on...

He's got his perfect media server config posted up there. Yeah, and his personal stuff. And you can try out grep.app as well, which is like a GitHub search that you don't have to sign into GitHub to use, it's also pretty good at finding repos that have Nix in them. grep.app is a good, that could almost be a pick, but that's just a tip right there. What you said too, I want to underscore, I love it when we get together and we're working on systems and we're posting each other's configs and I'll just start, I'll snag stuff from, you know, we just, it's great.

Oh, that's how you solve that. Oh, I'll grab that real quick. Yeah, my newest favorite one is Pretty Rebuild. Pretty Rebuild is nice. Yeah, that's the name I came up with for a little script that just uses Nick's output monitor. I need to clean that up and put it on game software. But it makes it look nice.

It makes it look nice. Thank you, everybody, who participated this week in the booth. It's always a lot of fun for us, as you can tell. We had 37 of you just listening and streaming sats as you listen. I want to call out Sikander, too, who sent us not a ton of sats or anything, but a bunch of them and was clearly trying out CurioCaster. Oh, neat. So we see you. Hey-o.

Shout out. So 37 streamers stacked 113,802 sets just on their own, just streaming those sets. Nicely streamed. Yep. When you combine that with Elboos, we stacked a grand total of 302,525 sets. Thank you, everybody. This is a Value for Value show. We don't want to go the way of Linux magazines. You can become a member or you can set your support with a membership on autopilot, or

you can do via boost at your own cadence, at your own amount, at your own time. Get your message read on the show. And it's all open source, free software, no middleman, peer to peer. Really appreciate you. And big shout out to our members as well. As we wrap up, I have, well, I kind of went overboard this week, boys. I apologize. There's no containing you.

There isn't. Brace yourselves. It's a big pick segment. So first up is an application that Brent has been getting a lot of use out of. You're going to love it. It's probably going to tell you what it does right in the name. It's called ButterFS Assistant. And Brentley, you've been playing around with this.

It's a GUI on top of ButterFS. Yeah, I have. As I teased earlier, I've been kind of diving into ButterFS on a few different systems here, mainly because I now have a ButterFS native NAS here on my local network. And so I'm going to be migrating from using Borg, which has worked really well for me for backups, to a more file system native backup solution.

I don't know how long it'll take me, but tools like ButterFS Assistant is very helpful in this regard. It just is a gooey to do a bunch of the stuff you can do on the command line, of course, but it just surfaces a lot of your ButterFS volumes, for instance, sub-volumes, and makes it just easier to see what's there without needing to go poking around in the command line and to also just edit a few snapper entries so you can have easier visibility on your backups and sending things in different places.

So if you are running ButterFS and you're interested in just doing that in your GUI, highly suggested. I think what was kind of appealing to me is, like you said, it gives you one place to just view all your sub volumes. But getting kind of a GUI on top of these features, maybe you didn't even know ButterFS had, one spot to manage all your snapshots, to't even know ButterFS had, one spot to manage all your snapshots, to do all your ButterFS maintenance, to know what the difference,

but to even know, to even know what the difference, what the different ButterFS maintenance options are. That's nice. Just have a GUI and you learn it.

And, you know, it's just doing command line stuff on the back end, but that's, I kind of prefer that actually. Well, it's just probably handy because, you know, some of the configs

that we were sharing in Nix was turning off man pages and docs, so there's no man page for Brent to read. He needs the GUI. That is one of the that, actually. Well, it's probably handy because, you know, some of the configs that we were sharing in Nix was turning off man pages and docs. So there's no man page for Brent to read. He needs the GUI. That is one of the things we did. No man pages. Save time. And then, Wes, while you were messing around with kernels this week, you found kernelconfig.io.

Oh, yeah. This was very handy. Not only did they actually have the data for the RC kernel that I was playing with in there. Wow. But, like, I don't compile kernels a ton. And it was a little tricky at first because you have to turn on expert, config expert before you can turn on config preempt RT. That's fair. But if you don't do that, then you kind of get weird errors that are like, got config RT equals yes, but wanted no.

What? What is that? You're not an expert, so you don't want it. Right. Yeah, exactly. So reading through the docs and being able to quickly, plus as I was trying to figure out the options, I really wanted my DRM panic QR code because you can change the background color.

By default, I think it's just like black and white, but I wanted that blue screen look. So I found- Heck yeah. I asked AI, like, what's the Windows blue color and hex notation and then plug that in. And so, yeah, being able to quickly search through kernel config options

and see the docs on them and what they take and stuff was super handy. I wonder if that's the same blue they've been using since Windows 7. It is a very pleasing blue. I'll give them that. But the blue screen blue

might be a different blue. That's a lot. Kernelconfig.io and Butterfest Assistant. And there's so many more. In fact, I'm going to do it, boys. I don't even have it listed. I was going to save it for next week.

But we're going to go for three picks. Surprise, third pick. Before you do, just for reference, you know, for the color nerds up there, 0000C0 is what I read was the Windows. Thank goodness. Is that correct? Let me know. It feels right in my heart, Wes, and that's what matters. Okay, here we go.

Third rando bonus pick, not even linked in the doc. The boys have no idea I was even going to make this pick. It's called Television, and it is a fuzzy search for the Linux terminal, and it is a full-on TUI that it just looks incredible. Who found television this week? Because it wasn't me. Must be Brent, not me. Brentley, did you find that the revolution

will not be televised? Was that you? I don't have television here, so I don't think it was me. It might have been a listener. Oh, no, this was me. I just forgot the name. Yeah, it's really, really neat.

And so that's why I'm making it a bonus pick. And yes, of course, you guessed it, written in Rust. It's a blazing fast, general purpose, fuzzy finder, 2E. So it's a terminal user interface written in Rust. It's inspired by the NeoVim telescope plugin, and it's designed to be fast, efficient, and simple to use and easily extensible. And it's beautiful. be fast, efficient, and simple to use and easily extensible. And it's beautiful. It's beautiful, too. So really, the results are fantastic.

It has a fuzzy matching algorithm to filter the list of entries, the algorithm that then depends on the input pattern that you provide. So I could see this being really useful. For example, one of the things I thought we might do tonight after the show is go through fake NAS and figure out where we could trim some of the fat and a tool like this makes it, makes it really easy. So there you go. Three fantastic picks,

butterfest assistant. Really? If you have a butterfest file system, you should look at some of the maintenance options you have there. Kernel config IO, just play around and learn about some of the options with the kernel and how that works. And then television, which we'll have linked in here. It's up on GitHub and it is a very cool terminal application. It's funny. I feel like 2024 has been the year of some really

kick-ass terminal applications that are better than some of we ever had, which I don't know. Go figure that one. Hey, did you know we do this show live? We do it on Tuesday, as in Sunday. You can join us Sunday at noon Pacific, 3 p.m. Eastern. See you next week. Same bad time, same bad station.

We'd love to have you join us. We've got a mumble room where you can get a live Opus feed right off of our mixer using a full free software stack or, of course, in a podcasting 2.0 app of your choice or plug jblive.fm into your radio app of choice. See what happens. And just listen. We love to have you there. And then the whole thing's made available to our members

as a special thank you. And, of course, we'll put links to everything we talked about today at linuxunplugged.com slash 589. We're getting close. 590 next week. Starting to get a little nervous. Going to be working on them tuxes too. All right, everybody.

Thank you so much for joining us on this week's episode of the Unplugged program. And I hope to see you right back here next Sunday, as in Tuesday. Thank you.