LINUX Unplugged - 278: Shell in a Handbasket

Episode Date: December 5, 2018

We chat with a developer who's gotten Linux running on iOS devices, do a deep dive into Clear Linux, and discuss Xubuntu ending 32bit support. Plus why Android in the cloud, and a bunch of community news. Special Guests: Alan Pope, Martin Wimpress, and Theodore Dubois.

You remember that old joke about put it in a Beowulf cluster that used to be the go-to line for like every Linux user ever? That's kind of died off, hasn't it? You don't happen that much. Well, I think it's because Amazon provides them for you in the cloud. You just have to pay them. The Beowulf cluster was this go-to way to cluster a bunch of machines together. But, you know, also, Wes, like you could just put a bunch of CPUs in a system

and then just design the workloads to spread across a farm, I suppose. Maybe that's why the Raspberry Pi cluster might just be the ideal candidate for some. It's a new system that lets users pack up to five Raspberry Pi Compute Module 3s into a single, compact, relatively low-cost, power-efficient computing cluster. Yeah, the MiniNode's Raspberry Pi 3 carrier board is designed to house multiple CM3 boards for carrying out computation at the so-called edge of the network. For instance, building automation, running a monitoring station, things like smart traffic lights in a city. The network connectivity is provided by an integrated gigabit switch to connect all of

them together. So that's pretty nice. And it feeds five boards, which are also like five awesome Raspberry Pis that have gigabit connections. That's really nice. Yeah, they've also got four gigs of flash storage. Looks like no Wi-Fi or Bluetooth at the moment, though. Decent processor, 1.2 gigahertz

quad-core, as you probably would expect. Okay, so prices, it's $259 for the board, and then each Raspberry Pi compute module is $30. Yeah, if you look at it, I mean, it is like a bigger board, and you just sort of slot

these other little pie boards right into it. Now we could have a Beowulf cluster of Raspberry Pis. You know, this is for all that edge work that you need your Raspberry Pis to do. I was thinking maybe if you're using Raspberry Pis, maybe you want a little build

farm at home. Well, there you go. I feel like we need to have a committee and the Jupyter Broadcasting maybe should form it and then we can vote if we're going to allow terms in. Like, we should have had a vote on hybrid cloud and we should have a vote on edge devices. Because like you read this article and it makes my skin crawl. Here's just one paragraph. Edge computing is becoming increasingly common in situations where IoT sensor data needs to be processed with low latency or in near real time, or it needs to be handled locally for regulatory reasons. Edge computing is just another way of saying electronics devices, right? It's just another

way of saying devices that are in the consumer's home or in the end destination. An edge device is just something that is network connected out on the edge. I don't like this term at all. It kind of separates between the compute device that's batching and processing this data and the sensor, which could be an Arduino or some other low-end microcontroller device. So for example, there was a company that were doing a thing where they have cameras in elevators and they were taking pictures of people in the elevator

and then using some intelligence to scan the picture and figure out if it looks like the person was a bad guy from the, I don't know, the look of their face or facial recognition, or it looked like they may have a gun under their clothes or if they get the gun out or something so it could see that. But the thing that's connected to the camera doesn't have the compute power to do that. And if you had multiple elevators in a building, you'd have multiple cameras feeding into an edge device that had the compute capability to do that. So there's certainly, you've got to have some term for the thing that's doing the compute that isn't

the device that's at the very, very end of the line. Here's where I think the line should be, I think the line should be is if it's in your network on their premises, you could consider that the very edge of your LAN or your WAN. You could consider that to be the edge of your network. But if the device that you are communicating with is on a separate network, that's not an edge device. That's just a node on the internet. That is just an internet node. It's not an edge device at that point. I grok that it could be a sensor collecting data, transmitting it back to a larger processing system.

But if they're on separate networks, I feel like edge device is, it's a dumbification of the term. Although all of that said, I did get a tweet this last week from an elevator where that elevator was doing a file system check. So I don't know what to call that. Chris, I think you should just think about it like this.

You got to have devices to funnel data to your blockchain-based machine learning platform to analyze it. Nailed it. This is Linux Unplugged, episode 278 for December 4th, 2018. Welcome to Linux Unplugged, your weekly Linux talk show that's all over the world this week. My name is Chris. My name is Wes. Hello, Mr. Payne. Good to be connected with you.

I am in Dallas for, well, for some shenanigans that you're actually about to join me for in just a couple of days. We're going to have our company party now that we're part of Linux Academy. Very excited about that. But we've put together a show this week that really looks back and pulls out the best bits. We've got some great community news that we're going to go through here towards the top of the show. A fascinating look at Android in the cloud. It is actually a thing. And then I had a chance to chat with the developer of Ish. Ish, you might be familiar, is a Linux shell for iOS. And the developer is hard at work

at implementing an entire Linux environment that iOS users can use. And so far, Apple hasn't pulled the plug. I'll talk to him about that, why he's doing it. I wanted to find out too, if he was like a Linux diehard that was trying to bring Linux to iOS, or if he was an iOS diehard trying to learn about Linux. So we talk about that and more. A follow-up on some crazy issues I've had with my laptop after connecting that GPU-powered dock, and you might run into it as well. In fact, it impacts more than just GPUs. And then we'll take a little deep dive into Clear Linux, which has been making more and more news recently and just refreshed their desktop offering. You could say double down on the desktop offering.

And Wes is going to kick the tires, and I did a deep dive on the technical side of things, and we'll be taking a look at Clear Linux towards the end of the show. But before we go any further, we have one bit of business we must attend. We must bring in that virtual lug. Time-appropriate greetings, Mumble Room. Morning. Howdy, howdy. Hello. Hello, Bruce

and Popey and Sean and the Silent Drifter. All of our friends. It's good to see you. Yeah. It is nice. We have a few up in quiet listening and in staging as well. Mumble Room's cool because you could just listen in the Mumble Room and then you get like the raw

untapped or I don't even know what the unfettered studio feed because it's connected to our soundboard. I was going to make that sound really cool and then I just realized it's not that cool. It's just an audio interface that plugs into our soundboard. But at the end of the

day, I should make it sound like the raw direct feed from the studio mixer. That's pretty good, right? That was great. That's way more hype. Thank you. Thank you. All right. Well, let's start with some community news.

A story that I'm a little grumpy about. And I didn't want to be. I didn't want to be. I wanted to be the guy that came on the air and said, I think it's time. You know, it's been 10 years. It's time. But I can't.

And I understand, though. It's just what it is. The world's moving on, and Zubuntu is one of them. And it's official that Zubuntu will stop producing 32-bit ISOs beginning with 1904. The development team has now decided to go ahead and eliminate the 32-bit ISOs beginning with 1904. The development team has now decided to go ahead and eliminate the 32-bit builds moving forward. The change will affect Ubuntu 1904 and beyond, but not the current 1804 LTS.

Yeah, the developers announced this this morning following a team vote. Interestingly, it was the sixth out of ten votes, so clearly not, you know, it wasn't unanimous. There's still people out there representing 32-bit, I would imagine, especially in the Zubuntu following. This leaves LXDE or LXQ as the only Ubuntu desktop distribution still offering 32-bit images moving forward.

And there's just no way that's sustainable. That puts all of the work for testing, for all of that, for building all on that project. And it's not a focus of the project. The project has, Lubuntu has stated, they want to focus on more modern hardware. Now, the common meme here is that 32-bit processors haven't shipped since the Pentium 4. But the reality is they have shipped as recently as 2011. And I grant you that's not yesterday, but it's also not 10 years ago. And I don't know. I hate to see something that seems to have a well-defined purpose

to facilitate really good performance on hardware that is maybe, wouldn't be worth a YouTube video. You know, maybe it's not a Threadripper. Maybe it doesn't have the latest and greatest GPU and the latest and greatest CPU. Maybe it's, God forbid, even running spinning Rust disks. Maybe.

Isn't that where, like, ARM is now, though? Like, if you just have an old 32-bit processor, shouldn't you just get a nice ARM chip and probably have a better time? And we should have distributions focusing on that. Because like, Zubuntu has its own things

outside of that, right? I mean, you can have a lean and mean Zubuntu desktop on just about any system, not 32-bit going forward, I guess. Yeah, yeah, you're right. I guess, I guess if you have, you're saying if you have a system that old, you could almost just go get

a cheapo ARM box. That said, I mean, I do appreciate it because one you have a system that old, you could almost just go get a cheapo ARM box. That said, I mean, I do appreciate it because one of those things Linux has been. Like it feels like there has been this part of our culture where it's like, oh, it doesn't matter. It's super old Mac. Okay, we'll put Linux on it. Whatever system you have, you could put Linux on it. And some of that's fallen away. True, yeah.

And I think part of it for me is this is something that these distributions were serving. Like they were super serving a specific type of user base. Like that was the role they played. And it made room for other distributions to go hog wild in other areas. But, you know, that's just my thoughts on it. Wimpy, I would obviously love to hear your thoughts, not only as an Ubuntu Mate maintainer, but also just having years of experience with ranges of different hardware and distributions. Yeah, so we obviously made this decision, or rather I had the final decision for Ubuntu Mate to end the 32-bit Intel support. And basically that came down to the fact that there weren't people within the team that had the hardware to test and qa this stuff you know so yes whilst there may well

be devices that have been produced as little ago as seven years that have that instruction set and that processor um you've got to have people out there that are using it that are maintaining the distribution and what wes is also quite correct that for intel 32-bit systems of that era the arm platforms these days offer competitive performance and are inexpensive by comparison and just because anabuntu flavor is not supporting 32-bit intel as a release architecture anymore doesn't mean that the sky is falling. There are dozens and dozens of distributions out there that cater to supporting these, you know,

what are becoming niche hardware platforms and specialize in that. And I don't think any of us would be saddened to see owners of that hardware go to those niche distributions where they can be well supported. Sure. I guess I can't disagree with that. You're right. It's not like there's not somewhere else for these users to go. But what is the role then of Zubuntu and Lubuntu? I guess if their main advantage is that they have XFCE or LXQt,

I don't know if that seems like enough of a differentiator. You can get that anywhere. Like the focus is sort of what made it a go-to. Like it put them in a channel for a particular type of user where the randos on the internet would filter people into that channel when they're looking for something. Like in a way, it was part of their brand. And now the main thing about their brand is going away.

I don't mean to dwell. I want to give Popi a chance to jump in, too. I don't think it's the main thing about their brand. It's a factor that would help you decide which ISO you were going to download if you had that old, crusty laptop and you wanted to run software on it, sure. But what I find frustrating is whenever we announce the death or depreciation of something, people come out of the woodwork and tell us we should continue working on this thing. And I'm sorry, but you don't get to tell us what we work on.

If you want to work on it, you could volunteer some of your time to work on it as well. Like there are plenty of people out there who will wring their hands when something is no longer supported. But they won't step up and help when people put the call out for it. And I find that incredibly frustrating. If you're not going to do anything about it, then shut up. Yeah, I can see. And that is sort of a definite, it's not the friendliest answer, but it is definitely a legitimate answer in the open source world. It's like, well, the source is there. If you want to do the work, you can do that work. And like I say, I think for the most part, it has been a really long time and the market has moved on and the people that are running 32-bit probably don't have to.

I've said that before on this show. They don't – they likely don't have to. But it still seems to me like – I don't know. It's like in the world of cars, when you have a carmaker that has something in their cars that makes them special, when they stop doing that thing, it just sort of takes something from it. Mr. Drifter, I want to give you a chance to jump in, and then perhaps we will move on. What do you think? Well, I agree with Popey. I think that there's a lot of responsibility, even in, you know, large corporate offices to put all of these old crusty machines and it's, it's kind of hard to maintain it, but you had mentioned, or at least ask the question, you know, what would we use? You know, what is

the use case then for these like XFCE and, you know, these desktops for 64 bit. And I think that, you know, when you go on there and you just need essentially kind of a jump box or something that doesn't take a lot of resources, I still think that these different distributions are going to really excel in those areas. Right. And I think at the end of all of this discussion, you have to keep in mind the project. And if the project doesn't have the resources to do the testing and they don't have the passion to go build 32-bit boxes to make it work, then they're probably not the right project to be doing that work. And it doesn't really matter how you feel about it.

It's just the reality of human nature and how these kinds of things work. This next story, I only included this in the show because just the holy shit factor of it. It's really surprising. Microsoft is throwing in the towel with Edge HTML. And today it's come out that they're instead building a web browser based around Chromium, which uses Blink as its rendering engine, essentially, you know, their WebKit fork.

This is massive. Woo-wee! They were also recently spotted, them being Microsoft engineers, committing code to the Chromium project to help get Google Chrome running on ARM. Perhaps some of that work will translate over to getting Anaheim running on Windows 10 on ARM 2. I don't know about that, but it is interesting. I mean, think back to when IE6 was the big thing.

I know. Is Microsoft done with the browser game? They're just going to ship one, you know, kind of phone it in? I got to say, not particularly jazzed about another Chromium-based browser, but mind-blown about the end of Internet Explorer,

essentially. Like, this is really it. Like, I realize Edge wasn't IE, but to me, this is really quite surprising because it is a massive strategic beachhead to have one of the popular web rendering engines. But it really seems like Microsoft in all areas right now is looking at it in a very humble, practical way. Where have we lost? Where are we just not going to win no matter how long we pay a team to work on something that we think has great battery life and great performance or whatever they want to advertise about Edge? And they had to go and have that conversation and say, we got to just throw in the towel here. We just got to ship a Chromium browser.

Now, we don't really know if this is going to replace Edge fully. Maybe they're just going to swap out the ass end of Edge and never even say anything to the end user. Maybe it's only going to be available on ARM versions of Windows. We don't know, do we? No, I mean, we don't. There's not a lot of clarity here. But I just want to welcome Windows users to the rest of the internet. Clarity here. But I just want to welcome Windows users to the rest of the Internet. Nice to have you.

And you can find the free stickers over in that corner. This next story, cue the Valve SoundLink. Cue this, what is it called? What's that Valve gong or whatever? Cue that. Cue the Valve gong because the Steam Link lives on on the Raspberry Pi. Last week we talked about the Steam Link hardware going away, and I got a couple of notes from several people that said I went and bought one, and then they were totally sold out.

I almost did. I mean, it was very tempting. They were like $10 for a while. That's pretty convenient. Wes, they got down to like $2.99. Oh, really? $2.99. So anyways, there is now a beta to run the Steam Link app on your Raspberry Pi 3 and 3B Plus running Raspbian Stretch. It's a little bit funky. You have to run it and then it downloads the full app and creates a desktop shortcut.

And then we'll automatically update as needed from there. So it's not your traditional package, but sounds like it's working yeah well you know we do have a reporter on the scene let's go to him now live on location mr popey what are you learning uh so i tried it out on a new pi b plus three wherever the new one is the fast yeah one and uh i used a debian stretch raspbian image that i downloaded from the raspbian the raspberry pi website chucked it on a stick and then grabbed the deb installed it ran it and then connected to another machine on my network that had steam installed uh plugged in an xbox 360 controller into the raspberry pi and streamed a game directly to my raspberry pi

from my steam machine and it worked pretty good it okay. There was a few network stutters. But other than that, I mean, it's beta quality, but it worked. It worked just fine. How would you rate the quality of your Wi-Fi in your home? Because it seems like that's a pretty important component. Yeah. Wi-Fi in my house is middling to shitty. And so that wouldn't have helped. But I did do the test that it does internally, and it said it was fine. I didn't have a super high frame rate game. I was playing Ultimate Chicken Horse, which is a puzzle game.

But still, it's one of these things where you're standing on a platform, and if you're one pixel off, you off so you know it still could affect gameplay um and it wasn't perfect but like i say it's beta quality but it's worth trying out and could be a good opportunity to dedicate uh raspberry pi to sit underneath the tv and play some games streamed from a hulking great loud Steam machine somewhere else in the house. Yeah. Well, if this is part of the strategy, maybe they're going to go Netflix on this.

Like when they said that they're going to rely on the quote-unquote app, maybe what they really meant was full-spectrum domination of all devices like Netflix did. Like when Netflix started streaming, it was called Instant Watch, and it was supplemental and free if you had a DVD subscription, right? And then they had the Instant Watch service, which you could watch in your web browser if you were on Windows or Mac or something. I mean, we all remember that.

The dark days. Yeah, it took years to get it on Netflix, on Linux. And then there was all these different hacks and ways you could do it for years. And in this intervening time between the DVD service and now where it's at, they launched, well, they didn't, but in partnership with Roku,

they launched the Netflix box, which later became the Roku, which is an entire platform now and now everything you buy from toasters to televisions has a Netflix app built into it at least in the States and if Valve could go somewhere in this direction

and maybe they could get pre-installed on things like the Nvidia Shield TV or a Roku box or maybe even a smart TV, if it's a decent one, then the death of the Steam Link starts to make a little more sense, doesn't it? Because they're following the Netflix model at that point, which seems to have worked. You don't have to buy a new device. You just have to install a new app on the device you already have and, I guess, figure out a way to have a controller connect to it. So maybe not every device. Yeah, there is that.

But really with Bluetooth, your options are wider and wider. And I don't know about you, but I don't really feel like buying another box. I may have bought the next Steam Link, but probably not. I don't have a 4K television. I mean, and that's because you're a niche person

who knows it exists and might have considered it. It seems like the discoverability of something in an app store is a little bit better. And I think people are sick and tired of buying these set-top boxes and hooking up another box to their television. Yeah, and then they have to get a new receiver because you have so many different things that you have to plug in. They might be onto something here. And if they can let you roll it on a few of your own devices, they integrate it into SteamOS. So if you build a Steam machine, it's just built in. You have Raspberry Pi options.

All of a sudden, I'm starting to like their new strategy. So we'll see where this goes. And I think I kind of want to try it out. I kind of want to play around, maybe compare it to the Steam Link one day, the actual physical hardware. Why don't we take a moment and say congratulations to a friend of the show, the handsome Barton George, who is celebrating with his team six frickin' years of Sputnik. It all started in the beginning of 2012 with Barton's pitch to an internal innovation committee

that they had at Dell. Like their own internal seed fund that they could use on a project internally if somebody had something good. And then a month after the pitch, the team or whoever he was pitching to seemed to like it. The committee gave him the go for this quote-unquote exploratory project

to test out the idea of a developer laptop. Just test the waters. Thankfully, I mean, there was an interested and supportive community. So with enough support, people was an interested and supportive community. So with enough support, people actually said that they wanted this. It became a real product. On November 29th, way back in 2012, the first Dell XPS 13 developer edition was born. Yeah. I remember the week I got to cover that in the Linux Action News or in the Linux Action Show. And man, huge. What

a big day. Dell shipping a Linux, like for legit Linux laptop. And Linux Action Show. And, man, huge. What a big day. Dell, shipping a Linux, like, for legit Linux laptop. And it was announced, and then it was, before the orders were available, people figured out, like, which Windows version you could buy that was going to be the same thing with Broadcom Wireless. Yeah, but now fast forward,

they're on, they're the seventh generation of laptop, so it's been six years, they've done seven generations of that laptop, uh, shipping it with 1804 by default right now. And then they've extended that to the precision workstation line, which I just recently reviewed, which I think is a pretty killer system. And, uh, I don't think they're stopping. In fact, I think they have demonstrated to the market that this is a viable consumer segment to sell to.

And I personally believe it's played a part in why you see the subsystem for Linux on Windows 10 and why you see Linux capabilities on Chrome OS. And why I think ISH or ish, the Linux environment for iOS is smashingly popular right now. Like people are going crazy about this. So I had a chance to chat with the developer of ISH and it was just a quick chat. He was fortunately very accommodating and willing to join me. And just I wanted to talk about why he's doing this. And I think it all plays a role here. I think it's all part of a bigger picture

about a common Linux runtime and environment that developers want and admins and system administrators. So I had a chance to talk to this guy. His name is Theodore. And Mr. West, go ahead and play our clip. So Theodore, welcome to Linux Unplugged. And congratulations on all the attention ISH is getting. Yeah, it's been honestly surreal to see it just go from, like,

20 stick-up stars to a few weeks later, 2,000. I feel like there's, like, three podcasts I listen to right now that are talking about it. So it's a Linux shell for iOS, and I thought maybe I kind of wanted to just pick your brain on this crazy project. And, and also maybe we could start even lower. Like don't, don't you have to emulate an entire x86 basic system to pull this off? Yeah, it was this, this took like a year and a half to get to, to like the stage where I could actually run it on an iPod on a device.

Wow. So yeah, it was, it was a big project that I mean, the real reason I've been doing this is to like learn everything I can about, um, Linux and x86. Cause it's just an interesting topic to me. Are you by day a Mac user? You're not a, you're not like already like a Linux user who's just trying to get Linux on an iOS device. No, I mean, I'm, I'm mean, I'm a Mac user. I also sometimes use Linux in a VM. Oh, okay. Like for at least six months, I was developing Ish under Linux and then ported it to Mac. Oh, that is really interesting. And is that phase the x86 emulation? That was when I was getting the x86 emulation to work. This was before it could at

all run on an iPad. Now, is that like QMU under the hood? Like, how are you doing that? It's a, I wrote my own x86 emulator for this. Oh, you mad man. Yeah, exactly. Which is why it took so long. But it must have been an interesting experiment. Oh yeah. Yeah. Kim, you can't run on, uh, on iOS because of the restrictions of the sandbox. Cause it's doing dynamic binary translation.

Um, the, the sandbox doesn't let you, uh, have any executable code that didn't come with the app. Um, so it has to be done with an actual emulator of the kind that runs each instruction with C code. To try and make it a little faster, I decided that I'm going to now write an assembly language implementation of every instruction.

So that at the end of the assembly language gadget, I can just make a direct jump to that construction and hopefully save a little time with decoding. That's pretty clever. and hopefully save a little time with decoding. That's pretty clever. And so once you got the x86 system working, you had to pick a Linux base, and you went with Alpine. That's an interesting choice.

Yeah, that's because it is the smallest Linux distribution ever created. The result is that the app is like a 3-megabyte download for Linux, and it includes an entire Linux distro that has a package manager and can install anything. That's a great point. It is. It's impressive. And the only way to really get my hands on it is to go through TestFlight right now. And I feel like that maybe we should talk about that. It seems like that could potentially be the big elephant in the room is you could end

up spending, you know, theoretically, what, two years on this project? And then Apple just never accepts it to the App Store. I mean, that could happen. But I've already learned so much about how Linux works and about how x86 works. It's, even if that happens, it's still worth it. And it seems like you've struck upon a real nerve here that there is a bit of a need for a command line like environment on an iOS device, especially the iPad Pro. Yeah, the amount of interest that I've seen on this has really kind of surprised me because, like, who wants to program on an iPad? But, yeah, apparently people do

because Apple has been telling them that their iPad is essentially a computer, so they want to now be able to program. There's all these little edge cases where you've got a great portable system, but you just need access to a command line utility. Maybe it's SSH, maybe it's something like YouTube DL. Who knows? It's just one of these things. Everybody's got their edge case. Yeah, that is true.

There are plenty of nice command line utilities. There are plenty of things I can do easily on the command line, like on an iPad, I would have to go searching through the app store, maybe find some app that costs $5.99. I mean, you know, who am I, Theodore? But if I was sitting at Apple right now and I was looking at the market, I would be looking at the success of the Windows subsystem for Linux on Windows 10. And I would be looking at Google adding Linux app and command line capabilities to Chrome OS, which was always supposed to be the most simple OS out there.

It's just nothing but a web browser, and now you can get a Debian command line. And iOS is sort of the oddball out on this. Yeah, it's like the only thing that, the only OS that doesn't have like a built-in Linux linux or unix like uh shell um so yeah i am hoping i'm really hoping that apple will uh like see the demand on this and realize that it's something that they should like think about like i mean the fact that they

haven't that they haven't uh yanked it from test flight already um is i guess a bit of a positive signal yeah yeah that's a good way to look at it. That's what I was thinking when I installed it. The fact that they're doing this is a potential signal because I hope they recognize it might not even be that people want the command line because I suppose it would never happen, right? But in theory, Apple could ship a Bash app that would give you access to the Unix subsystem

that may be rolling around underneath iOS. But they're not likely going to do that, are they? And people want Linux. No, I mean, if they shipped a bash shell with iOS, that would essentially be a jailbreak. Right. Yeah. It would be, I tell you, it would be more shocking

than a lot of the other things Apple has announced. It's more shocking than USB-C on the iPad Pro. It would be unheard of. And so I really, but at the same time, Theodore, I have to be honest, the technologist in me cries a little bit that you have to construct an entire virtual machine on a mobile device to get a command line. Yeah. It's kind of disappointing, honestly.

And it also makes it incredibly slow. Well, you've got an amazing performance, and like you mentioned earlier, size out of this, too. I'm very impressed, but I wonder if this thing isn't going to just start growing like crazy, because you must be having feature requests coming in like nuts for everybody's little edge case. Oh, yeah. My GitHub vicious page is filled with those.

Um, someone, there's an, there's an issue that's titled improve the performance. So yes, I saw that. And that's like, Oh, thanks. Okay. Um, I mean, yeah, if anyone, if anyone has any brilliant ideas on how to make this faster without jailbreaking their device, please tell me, I mean, really, it is justbreaking their device. Please tell me. I mean, really, it is just you right now, yeah? Yeah. Do you want to keep it that way? Would you want to actually make this a... I would love to have contributions.

I've gotten quite a few pull requests, actually, from people who want to fix this or that system call. And that's been really great. Though I would, of course, love to have more of that. And you pronounce it Ish, right? So Ish itself, is it GPL or what is it right now? It is GPL, GPL version 3, which does probably a pretty good job of protecting against anyone who wants to try and sell it. And you're touching a lot of Linux. So that, I mean, I think that all makes a lot of sense.

So you are open to the idea of maybe making this a bit of a larger project. Of course, I mean, this is a really tricky thing you're doing here, Theodore. This is some seriously low-level stuff. I'm really impressed. Yeah. I mean. I don't know if everybody could pick that up. You know what I mean?

Like, that's not something that most people are going to roll in and be able to solve those problems. Yeah. Really, like, again, I say somebody at Apple ought to just hire you up and make this an official project. Man, I would love that. All right. I say somebody at Apple ought to just hire you up and make this an official project. Man, I would love that. All right. Well, if people are listening right now and they're excited about perhaps getting a legitimate Linux command line on an iOS device, we've been talking iPad. But, you know, you can run it on a phone too, right?

Yeah. It works perfectly fine there. Or even a touch. Yeah. So how can people support you? The best way to support the development of this thing is to go to GitHub, find some program that doesn't work, and take a shot at fixing it yourself. And that is probably the most valuable thing you can do to help this project out. You can also give me money. I'm not going to say, no, I don't want your money.

But that will motivate me to keep working on it, but not be as motivating as seeing people who care about it enough to take time to write code. Yeah, and I noticed a couple of times it's really just changing how the application does input or output. It's not major code development to get individual apps working. There are so many programs where the fix is one line of code. code. Well, Theodore, please keep up the great work. You're making this, you're making these devices more valuable. And I really appreciate your hard work. Thank you for coming on the show. And let's chat in the future. Oh, yeah, yeah. Thank you for inviting me on. Now you can go get your hands on this if he has enough test pilot licenses still available. I

don't really understand how that system works, but we'll have a link to his website in the show notes at linuxunplugged.com slash, what is this, 278. So just go there and get a link to that. And I think this is a really nice idea that I'd really be bummed if Apple were to shut this down because I think a lot of us have, like, an old iPad somewhere floating around that it'd be pretty cool to build a pull-up Linux terminal on.

Yeah, and he's gone to such lengths to try to be compliant. You know, it's really, it's not trying to jailbreak. It's not trying to flout Apple's rules. It's just something people want to use computers for. Well, and I asked him, why not just use QMU or something? Why did you build an entire x86 virtual machine? Like, I mean, hell of a project, but like why? And he said the answer was so obvious that when I asked it, you see, it made me feel like an idiot.

You can't run QMU on the A12X, right? You're not going to be able to run any virtualizer on the market right now on the Apple A-series processors. You've got to build for those chips, basically. And so that's what he's been doing for the last year. I mean, it does sound like a great project. Talk about a passion project, right?

I mean, that's one of the things. And then to turn around and release it all as GPO. If you take away the iOS, he's been building a virtual machine for a very specialized series of ARM processors that millions of people have. And he's GPLing it all. That's awesome. I think it is really cool.

I think that's a pretty neat story. So thank you to Theodore for coming on the show and telling us about it. Now let's shift gears and talk about Android, but maybe not Android so much in your pocket, but Android in the cloud. Now, Simon on the Ubuntu blog has posted a really interesting rundown of how they are enabling Android to run on Amazon's EC2 A1 instances. Now, those A1 instances are them new fancy AMD, or I'm sorry, it was almost AMD. It was almost. They're ARM processors. Yeah.

This is an interesting story, though, Wes. Why don't you tell us the story? A little side tangent that Amazon just launched ARM servers, but they were almost AMD ARM servers. Yeah, up until 2015, Amazon and AMD were working together on a 64-bit ARM server-grade processor to deploy. I mean, of course, they want them everywhere in their data centers. It's tough to get that to work. It was a big deal. And of course, the project fell apart when, according to one well-placed source, AMD failed at meeting all of Amazon's performance requirements.

Ouch. Yeah, in classic Amazon style, they went out and actually bought someone who had already licensed from ARM a system on a chip designer, Annapurna Labs. And then they just put that team to work building all the things they wanted, right? Internet of Things gateways, the Nitro chipset that they're using. That's for handling network and storage tasks that back EC2 machines. So they've already got a lot going on, and now they've got this team of talented people who are designing chips. They've got it right in-house. So this last week or so, Amazon announced these new rigs, these A1 instances that use the custom AWS Graviton processor based on the ARM architecture.

And when you read through this blog post that's up on Canonical's blog, blog.abuntu.com, they talk about using LXD containers to stream Android games from an EC2 instance to a mobile phone over the internet. And they are running a genuine Android environment and Android applications in the cloud and then streaming the results down to the phone. So I dialed up our go-to ARM expert, Mr. Wimpress. No, Wimpy, you were actually the one that linked this to me originally. I didn't see this until you linked it to me. So I wanted to pick your brain a little bit about some of this. Like this seems really far out there over a mobile network and all of that.

How is this even possible? Yeah, so this is demonstrating a number of the technologies that Canonical and Ubuntu have on offer to sort of model, scale out workloads in the cloud and then deploy that stack very, very quickly. So it uses MaaS, which is Metal as a Service, and Juju, which is the orchestration component, and then LexD containers, which are containers of all of these Android instances. So it's kind of proving out all of this capability

and the applications that run inside those lexd containers are all snaps and what have you so it's kind of demonstrating the whole the whole product suite working together right but the reason that we have this capability is they're actually customers that are interested in running android in the cloud and and this is something that we're actively working on. And it's unfortunate Simon, due to time zones, wasn't able to join us this evening, but he's the tech lead behind this particular group of technologies. Now, I'm struggling to understand the use case for Android in the cloud. I get game streaming, but I don't quite see how

game streaming seems like it would be at its fundamental weakness over a mobile network. There's so many latency issues and variables. There's a couple of things there. So game streaming is certainly one of the use cases. And when you talk about game streaming, most people think about streaming H.264 video, you know, a bit like the Steam Link does that we were talking about earlier on the Raspberry Pi app, for example. The technologies here can do one of two things. They can stream H.264 video or they can stream the OpenGL commands. the OpenGL commands. So in the demonstration that's in the video, in the blog post,

that's actually streaming the OpenGL commands from the cloud to the client device. So it's rendered client-side using the GPU capabilities on the device. So that's one kind of distinction. Now, the other reason is that why game streaming is attractive to game developers is when you're targeting Android, you have a number of Android API versions that you can target. And depending on your generation of Android device that's on the client depends whether or not that client has access to that game. If you have your game or your application deployed in the cloud against a known API version, it doesn't actually matter what the client version of Android is because now you're doing the compute in the cloud and you're just streaming the visuals down to the client device. So you can narrow where you focus your development effort to you know a known api constraint but i think you've also got like ci and unit testing and stuff like that like spin up a thousand android instances to test it against various versions right exactly so you know

if you want to do testing at scale having a hundred or a thousand devices is invisible, impractical, but if you want to spin up dozens, tens, hundreds, thousands of instances in the cloud to run your automated tests, then that's now something that you can do relatively inexpensively. Yeah, that seems like it could be a pretty significant use case is development testing. Yeah. And the other use case is if you have a sensitive application, let's say a banking application, if you have that in the cloud, then it can't be tampered with on the local device.

So if you stream it from the cloud, it's sandboxed away from the users of the application so you can't pull it apart and reverse engineer it. Silent Drifter, were you thinking along the lines of development testing when you said you can see a couple of use cases? Part of it was development testing, but my other thought was it would also be really nice, especially where there's so much hype right now about the idea of convergence, that it would be really nice to be able to put all of that heavy computation in the cloud.

And then you can offer a lot more services on the application of your phone, even though it's all running. You're just kind of sending off the data to have it do the heavy crunching for you. Right. I could definitely see that as well. I think this is a pretty cool demonstration. And if you want to look at the video,

it's about two minutes long, and it's embedded in the blog post that we'll have linked. And one of the nice things is it's a single Juju command to set up the entire Android environment, the containers, and all of it. So it is, like Wimpy was saying, it's a pretty good demonstration of the stack. Yeah, and all of the network orchestration

that goes on behind it. And there's another potential upside there for games developers. If you've got an MMO, if you have that now deployed in the cloud and streamed down to the client, you can have the high-speed backbone network infrastructure that exists in these public cloud providers brokering the network interactions between these multiple instances

of these games rather than over your cellular network. So hopefully less lag and more you know, more low latency game interactions. And I just realized, like, if it's an MNO, you've also got instant anti-cheat, just like with the banking app, you've got anti-cheat because in the cloud and the user can't tamper with it. Exactly. Yeah. Just like the banking application, you can't tamper with the local app because there is

no local app. That's one way to solve the Android fragmentation problem. But I think the thing that impresses me the most is the capability of streaming the OpenGL commands down to the client and then have the rendering done on the local GPU. That's pretty slick. Yeah, that's pretty neat. Simon's a clever guy. Maybe sometime soon he'll be able to come on here and talk to you about this in real depth. Oh, i would love that but at the moment this is this is a developing technology uh we're working

alongside a few sort of business partners in a couple of different areas and watch this space for for more news about this and uh hopefully it'll come to a cloud near you soon have you had this problem with your external GPUs that I've run into where systemd udev consumes an entire core and it just sits there and runs one of your cores non-stops after you boot

if you boot up without your NVIDIA graphics card connected? Have you had this? I have not. Oh, well, I have as a result of trying out the GPU dock and it was causing issues with my audio

because of the CPU. We were really scratching our heads about this. Yeah, it started really with, we thought we were having jack audio issues, where jack was dropping audio. And we started going through this troubleshooting and troubleshooting process. And it wasn't until I got into tech system and was setting up, and I had just bailed on jack audio, that I realized I still had a problem. And then I had to start troubleshooting what was going on. And I discovered systemd udevd consuming 100% of my CPU. Now, there is bugs everywhere about this,

including there is a bug on Launchpad. There's a bug on the Manjaro tracker. I have a couple linked in the show notes. And it seems to be a bit tricky. And it doesn't actually seem to be restricted to just NVIDIA cards. Although they do appear to be the largest offender, Bluetooth chipsets have caused this issue and audio chipsets have caused this issue. And in short, what it really is, is it's system to UDEV trying to connect to devices and get the right driver set up and all of these things. And then it's failing to do so because it can't find the device, so it can't load the driver.

And then it gets caught in a loop. And it just creates log noise forever. And it starts with the NVIDIA core is being initialized, and then it gives it a device number. And then it says no NVIDIA graphics adapter found. And the next line in the log is we've unregistered the NVLink core, major device number. And then it says the signature is not signed. Then it says the core is being initialized. Then it says the adapter can't be found.

And then it says it's being de-initialized. And then it gives an error message. And then UDEV gives an error message, and then the whole process rinse and repeats over and over again for as fast as one of your CPU cores can possibly facilitate it. And it'll make your laptop heat up, it'll drain your battery life.

It was really kind of awful. And I feel like there's probably several ways to fix this. Obviously, one is I could just get rid of the NVIDIA driver. But that's sort of a Hail Mary, would prefer to avoid that solution. I mean, you're trying to use it, right? Because you have the dock. That's the point. Right.

Now, when I'm traveling and I'm not hooked up to the dock at the moment, I need it to be reliable. And it can't be going and mucking up my audio. So I did a temporary workaround just to get systemdudev under control and that was just basically doing a systemctl stop systemdudev and then starting it again. And if you stop and start it after you've booted, it behaves itself.

It's just simply after a cold boot and you haven't stopped and started it, it just goes nuts. And you can see different forum posts on the internet where people are trying to figure this out, like several about Bluetooth adapters. And it's just essentially across every distro that seems to have a 4.15 or later kernel. And it doesn't appear to impact everybody, but it is an issue. And I was really trying to figure out like, why is my audio so choppy?

What is going on? And when I looked at it and realized it was because of systemd, UDAV eating up a bunch of CPU, I thought, you know what, the audience might have a way for me to have my cake and eat it too, where I can still have the driver installed, but maybe I run a script, I don't know, something, some workaround where I don't have to do this hacky shutting down of systemd, UDAV. So if you know, if you got an idea, tweet me at chrisles or linuxunplugged.com slash contact. Do you use the audio interface in the dock? Yes, I do, when it is hooked up, yeah, because I have larger speakers that are plugged into the dock, yeah.

Fine, I shall say no more. Oh, well, I'll tell you what, for the purposes of completeness, in case anyone is listening to this and thinking, oh, this is my problem. I have multiple audio interfaces, and I don't use the one in the dock ever, and I actually have that interface completely turned off, which is probably why I don't see this issue, because I've only ever got one audio interface

that's actually enabled. Well, see, I figure this is going to become a larger problem as Thunderbolt 3 is on more and more systems, and more and more people try external GPUs or external docks. And that was one of the reasons I wanted to get the GPU dock was to actually very much try to discover if something like this would happen. So that way I could come back on here and say, maybe don't do it yet. Or if you do, this is one of the things you may potentially run into. And, of course, I was really cursing that decision one of the things you may potentially run into. And of course, I was really cursing that decision when I was trying to set up for LAN and I had 20 minutes

to get on air. And I thought to myself, why the hell did I make myself a guinea pig on this? But I feel like it's going to be something that gets wider and wider adoption. So we got to figure this stuff out now. And I bet there's somebody out there who already has. So let me know. But before we get out of here this week, we got to talk about Clear Linux, which seems to be getting more and more hype, more and more buzz, more and more people talking about it. And I guess a lot of us have always thought of it as something that's for containers or something that Intel is hoping takes off in the server industry because it's a distribution that has traditionally focused on security and performance. It's got a rolling release model, and it is really built for customization.

In a sense, they look at it as you fork your own version of Clear Linux and build it, or they have some pre-built versions for you. And so I wanted to take a deep dive into Clear Linux for a little bit in this episode. And Wes had a chance to try it out. And I had a chance to go dig into some of their docs and watch a few talks and thought maybe we could talk a little about Clear Linux. So, Mr. Wes, let's start with your setup, your Clear Linux experience. You know, Clear Linux, it's in a very interesting stage of evolution. You're absolutely right. There's a lot of focus on this like well-engineered and architected distribution that works well if you have these complicated container workloads in the cloud.

And I think the technical merits of Clear Linux are some of the things that really stands out. They have a lot of blog posts about it. It's very well communicated. They've even got this great how to clear tutorial up on GitHub that kind of teaches you the unique concepts of things like bundles and mixes that actually build up how Clear Linux forms package sets. So that's one of the key things. The software is shipped in these bundles. And the way they do that is you put everything into a bundle and all of the dependencies for that bundle are resolved at build time on the server, not on the client at install time. So you don't have like a package manager that's figuring out all the dependencies.

When you install that bundle, they've figured those all out server side. It's kind of similar to package groups in other distributions. But then one of the things they do to kind of turn this up to 11 is get ready for this, kids, is they turn on auto update by default. And that is generally because they have a really strict security turnaround policy where they have a team policy of patching a vulnerability within 24 hours of a fix being available. 24 hours to get that fix turned around. So they turn on the automatic updates, which generally works out to be about nine releases per week,

about twice per business day. There's a batch of what are generally security updates for Clear Linux. I mean, it's an interesting sort of posture. It does seem very well engineered. They've thought about it. They want it to be robust. There's been a lot of effort put into that before to even get any of this. If you're going to try it, it's worked as a desktop for a long time. Originally, that was with XFCE. They've been transitioning to GNOME Shell, so that's what I tested out. And they're starting to roll out support for KDE. But actually, to get to that experience,

you had to use their longstanding text-based installer. It's pretty simple, but it required a working internet connection, and it didn't offer a lot of customizations around RAID or encryption or any of that sort of fancy stuff that people might want. These days, there's now a beta, a desktop beta, and, you know, actually it's pretty nice. It boots right up pretty quick. One thing I will say is it works offline, but it is chunky. It's like a solid 8 gigs after you extract it and takes quite a long time to write to that flash drive. Once you get into it, you've still got a text-based installer UI, but honestly, it's a pretty well-engineered Curses interface.

I'm a stickler for how to configure disks, and I didn't find the Clear Linux OS installer's disk management really the best. It did kind of mess up the first time I tried it. And there could just be, thankfully, I mean, it was on a test system, so it's fine. But that could have, you know, I was trying to put it on one partition and it just rewrote the whole partition table on me, which I didn't expect. And it doesn't provide you a lot of prompts for things like, once you've set up a root partition and a boot partition, then it'll let you install. But

I had mistakenly put slash boot slash EFI as the mount point. It didn't tell me anything. It just didn't let me hit the confirm button. There was no error message anywhere. So you have to be a little bit experienced to know what you're doing. So did you lose all the other partitions on that drive? Not all of them, or either that

or just reset up maybe a couple of them. It was definitely a different disk layout when I rebooted and not what I expected. But, you know, it is lean and mean. It works really well. It's pretty easy to go configure your mixes and get everything like what software you want, how you want it to run. That part was pretty smooth. Yeah, their mixes is kind of where I was going with that. Like you get your own mix of cleared Linux is kind of how they look at it.

And so it's fascinating the way they've kind of made that distinguish. Like, once you kind of set it up, it's your own OS in a way. And then they have this stateless design that they call it. And what that really kind of essentially means is they put default fully functional configs for when you install software in slash user. So if you were to, you know, start up that service or whatever, in theory, it should be ready to go, fully functional, not a half set up config. We have to go like define the port and things like that. But what's interesting is those go in slash USR. You can put your own overrides in slash Etssy or like in the home of a user account it's another

place you know you're in your in your home folder that's not too unusual but the cool part is you can go in and just wipe out all of those etsy configs just go rm them just get rid of them and what you're essentially doing is you're just setting the system back to defaults you can just it's essentially like hitting the reset button and it's a default Clear Linux config because they still have all of the original configs in slash USR. That's one of those things is they've really, they're doing things differently

and not really like any other distro. I mean, you know, Arch was this own new wellspring that was building things themselves, but Clear Linux clearly has different goals. It's minimalist in some ways, but in a lot of other ways they're going above and beyond, right? Things like their fancy Clear Boot Manager.

That's long been needed, and no one else has pursued that path. Yeah, this is an interesting piece that I don't really understand a lot about, the Clear Linux Boot Manager. I do remember talking to Ike about it ages ago, and I think he worked on it to a degree. And he may be back there working on it again.

It sounds like. I don't know if this is the role he's playing there. But this was something that was kind of coveted by other distributions for a bit. And I never really fully understood why other than it was supposedly going to make system updates like kernel updates rock solid. other than it was supposedly going to make system updates like kernel updates rock solid. Yeah, I mean, a system layered on top of all the other stuff that's going on at boot to try to make sure that things wouldn't fail, that it was going to have a solid, reliable rollback. And it really also embraced UEFI.

So that's another area where they are targeting a lot of more modern things. They target, you know, it's very well optimized for recent Intel processors. A lot of products benchmarks keep showing that. So that's another area where, you know, people are doing comparisons and seeing like, oh, they've spent a lot of time. They're doing non-trivial extra patches to glibc sometimes. They're really looking at this. And I don't think it gets a lot of attention. You know, it might see some cloud workload, but if you're a power user, that might be something you want. Yeah. So they talk a little bit about how they build

the whole stack sort of optimized for the IA processor architecture, the Intel processor architecture. And they talk about one of the things that they've really focused on is anything that they execute that has a series of dependencies, they've gone down that dependency tree and optimized all of that stuff too. Because, you know, they're Intel, right? So why not make it as badass as possible? Why not make a Linux that really shows your OS? And then they're combining

that with this solid update where they have this clear Linux boot manager that's supposed to protect you from kernel upgrades. They have that stateless OS where it's easy to revert. And then they have this mixes concept where, in their words, you compose an OS for your specific use case. They view it as your own Linux in a way that they're just sort of upstreaming. You've also now got, you know, you've got Flatpak there. So if you rely on applications provided by Flatpak, that's another route for software availability.

Did you give that a go? Did you try installing any apps via Flatpak? Oh, yeah. I mean, it works great. They've already provided some apps that were packaged. Like, it's installed on the live ISO and everything.

Really? Really? So how would you describe the experience of using the GNOME Shell desktop and just Clear Linux as a desktop vehicle? Not bad.

I mean, it's not super far deviated from GNOME defaults. Honestly, though, I think they've done, like, the website looks good. A lot of the experience is well thought out. The background images and some of the theme aspects. Really, I think they've done, like the website looks good. A lot of the experience is well thought out.

The background images and some of the theme aspects. Really, I mean, it's kind of just a plain GNOME desktop. So if you want that, if you just want to kind of get to work, you want a lean, well-architected, technically thought out distribution, I think that's probably the space right now that it's playing.

It's not going to be for someone who wants to try just, you know, any old Linux distribution. I think that's probably the space right now that it's playing. It's not going to be for someone who wants to try just, you know, any old Linux distribution. You have to want to work to get to Clear Linux, but they're clearly making strides to make that easy. My question, I think my main question is, is why? I want to hear more from the team behind Clear Linux about their target audience. Is this for developers who are using it for server workloads and they want to run the same machine? Or is there a wider applicability? You got to wonder if maybe they have eyes on the workstation. Why shouldn't Intel's own version of Linux

be the premier operating system for an Intel-based workstation? Seems like that's a slam dunk. You know, if anybody's going to make an OS that really makes their own hardware shine, it should be them. And Linux is a perfectly reasonable workstation desktop that has been proven successful in many markets that buy high-end CPUs. I kind of see some logic to it, I think. But I could just be kind of hope-casting. I think I just made that up.

You know, I could see it. If we had the software we need, if we got that all worked out, it might make a nice studio machine. Yeah, or like the recording system, something like that. I think so. I think so. Well, thanks for kicking the tires on. I found the deep dive to be pretty interesting too.

I wondered before we went into this episode, I thought, well, what is it that really makes Clear Linux cool? Because I keep hearing cool stuff from the audience, like people love it. And I I thought, well, what is it that really makes Clear Linux cool? Because I keep hearing cool stuff from the audience, like people love it, and I'm like, well, what is it? And there is some really different stuff there. There is some really different thinking. Like, I mean, nine releases a week? Holy crap.

I mean, it's rolling for sure. Like, that's a pretty intense release schedule. You're absolutely right. And I think regardless of if Clear Linux really takes off on the desktop or not, for whatever value of what that means, it's one of those distributions that other people in this space can learn from. They're pioneering new approaches that I think will make all distributions better over the long term. Ooh, I like that, Wes.

I like that. Well, we'll have links to Clear Linux if you want to give it a go. Try it out. They just had a new version. We talked about it in this week's Linux Action News. And one of the quote-unquote features would be this new desktop live image that Wes talked about,

as well as that Plasma version they're working on now, which, you know, it gets us excited here. So, yeah, additional coverage in Linux Action News as well as links to Clear Linux. In the show notes, a pretty neat-looking distribution. Intel's trying to actually do something different there, and you got to respect well as links to Clear Linux in the show notes. A pretty neat-looking distribution. Intel is trying to actually do something different there, and you've got to respect that.

Well, Mr. West, that brings us to the end of this week's Unplugged. Thank you to Wimpy and Popey for joining us. Go check them out at the Ubuntu podcast, which rumor has it, another one will be out this week. Hey, yo! And another great episode from those guys, I'm sure. And go get more Wes Payne at techsnap.systems. Wes and I do a systems network and administration podcast.

That's what the snap stands for. Over there at techsnap.systems. And a reminder, we have our special recordings coming up soon. Our hope casting, my new term, and our Predictions episode the day after Christmas, so December 26th, and the day after New Year's, so January 2nd. We're recording one day later, and we're doing a special edition, and we'd love to have you join our virtual look for that, and get your predictions in and your Hopecasting. But that's all for this week's episode. Thanks so much for joining us, and we'll see you back here next Tuesday. So I'm doing a ugly sweater party tonight, going to an ugly sweater party.

And I, uh, I'm not an ugly sweater party tonight, going to an ugly sweater party. And I'm not going to lie, I took the easy route, and I went on Amazon, and I just searched for ugly sweater. It's a whole product category now, and there's a lot of ugly sweaters. And the one I got is like a polyester thick shirt. I don't really feel like it's a sweater. I'm worried I'm going to get called out. I didn't know because I got it off Amazon. Thick shirt? Is this a technical term now? I don't know how like it's a sweater. Like I'm worried I'm going to get called out. And I didn't know because I got it off Amazon. Thick shirt?

Is that a, is this a technical term now? I don't know how to put it, Wes. It feels like a thick, it feels like multi-layer polyester shirt. I don't know how to describe it, but it's very gaudy. It's very ugly. Oh my God. I rush shipped it on Amazon so that way I'd have it for the trip. So like there I am like just with a bag.

I have the one bag rule, just a carry-on, one carry-on. And not even two, you know, a carry-on and a bag, just – anyways. So I'm stuffing this stupid ugly sweater into the backpack and I'm just looking at this thing going, what am I doing? And I forgot it at the hotel. No. So I run back to the hotel, grab my ugly sweater, and then race off to the ugly sweater party.