Linux Weekly 12: A Visit from Eric Raymond! Open Source, Commodore Phone, and More!

[00:00:00]It's time for another weekly Linux project update. And uh we've been working on projects together.

[00:00:07]This is Eric Raymond.

[00:00:14]>> We're here to have a lot of fun and chat and uh you know, we might start talking about the Linux news, but we might we might ramble. So, it's fine. Thank you for joining us. Let's get started.

[00:00:30]So, he's here because I'm building another Thread Ripper system for him.

[00:00:34]9970X 32 cores. It's going to be a lot of fun. That's going to be in a different video. You always have an interesting way of looking at things and interesting insights into the happenings in the Linux community. And slightly outside of the outline that we put together for the show, the the the forest for the trees thing is uh you noticed that Linux adoption may be accelerating.

[00:00:56]>> Oh yes, absolutely. And it's not happening the way we expected. We thought that that Linux would improve and improve and improve and improve and at some point it would pass the good enough for most people threshold and boom. That's not what's happening. Well, it's partly what's happening, but what's hap what's even an even more factor important factor is that the competition is steadily and shitifying itself, especially Microsoft.

[00:01:25]>> It's it is really nice to see the college students and enthusiasts adopting Linux as their their daily driver or Linux- based OSS um as their daily driver. But it's also worth pointing out, I mean, Linux has already completely taken over the universe basically everywhere outside of desktop.

[00:01:40]That's correct. And now I noticed that before 18 months ago, I never saw videos by people saying, "Hey, I'm switching to Linux." This was just not a thing that happened. Now they show up in my feed every day.

[00:01:53]>> I think also the psychology of wait a minute, I don't want the Windows experience on Linux is helping because the Windows experience has become very bad.

[00:02:02]>> Yes, indeed it has. And so that's helping because it was the goal of Linux was never to give you the Windows experience because once you learn the Windows experience maybe was not the most optimal one, >> especially not when it involves constant badgering. You sure you don't want to move all your files to one drive so we can control them?

[00:02:20]>> I think most people would be okay with it if the operating system was, hey, I'm going to use like a command and control plane in Microsoft that you might pay5 or $10 a year for, >> but everything is actually stored on your own computer.

[00:02:31]Instead, the direction they're going is, "We want your machine to be a a terminal for a giant AIdriven cloud thing that you don't even not only don't you control it, you don't even know what it's doing."

[00:02:43]>> Sa Nadella uh he also made the news because he said that I think the exact words he used was we may have lost our way.

[00:02:52]>> Wow. What was your first clue, Sherlock?

[00:02:57]>> Probably the one drive pop-ups. But let's talk about the economics underneath this. Let's let let's consider why they're screwing up this way. I think um a major reason is in modern management, you only optimize what your metrics can see.

[00:03:12]>> Yeah.

[00:03:12]>> And they can't see users being driven away by these measures. All they see is engagement metrics. The problem is there's this relentless drive for revenue maximization. I mean, which is not a bad thing in itself. That's how free markets work. That's how we all get to have nice stuff. But if you have a if you're given a choice between having a software company maximize revenue from you and an open source product over time the open source product is going to look more and more attractive and and we're seeing that happen now. I think if we wanted to patch it in a capitalistic way, I don't know that like the open source aspect, yes, but the immediate fix is to just make it completely okay.

[00:03:52]If I write a fix that is going to solve the planned obsolescence of the expiring perpetual office >> problem, >> you know, you you can kind of have your cake and eat it too. It's like they want to sell you a perpetual license because everybody's wise to the subscription thing, but the perpetual license is not actually perpetual, >> right? So, it's a five-year kind of a deal.

[00:04:15]>> And there's a wild card here given to us by LLMs. It is now possible to take a binary and disassemble it and get back source code that you can recompile the binary from. And that means that software that was that was shipped as closed source with enough determination and a few hours of LLM time, people can reverse engineer and modify it.

[00:04:39]>> I had a thought about that too. There is a bit of a keyhole in the DMCA for that type of activity. You can't distribute circumvention tools, but an LLM as a tool to create tools. If you create your own tools, >> right, >> there is a bit of a keyhole there in the DMCA for that. What else is exciting in in the Linux world for you right now or in the last little bit? Not necessarily news this week, but just fun things.

[00:05:03]>> I have been really enjoying Cosmic Cosmic Desktop. It's not finished yet.

[00:05:09]It's not completely polished. There are bugs around the edges and there are things that could be improved, but it's such a breath of fresh air. I I feel like looking at it in retrospect, the desktop environment world had gotten kind of stodgy and complacent and and overweight.

[00:05:27]Uh, and Cosmic completely changes that.

[00:05:30]Uh, Gnome and KDE have to up their game now. I like the uh the uh the things that the cosmic has done especially around uh window management tiling. You can get the best of both worlds of tiling and traditional.

[00:05:45]>> That's the big draw. I I like tiling window managers and for a long time I was an i3 user. The thing is monastically pure tiling window managers are a little bit too austere. They're a little bit too uncomfortable in respect and and now I have cosmic and I get all the things I I had in i3 and it's also it's pretty that's kind of nice.

[00:06:10]>> It's pretty and effective.

[00:06:11]>> Yeah, >> it gets out of the way. Yeah.

[00:06:13]>> Uh we will have like there's a full uh video on your build and the software and hardware choices that went into it. Oh, Commodore. Uh Commodore is releasing a flip phone. We talked about that a little bit earlier. Commodore. You should probably a lot of the audience may not understand Commodore's importance in um the early rise of computers as an enthusiast platform.

[00:06:40]>> But we're not here to talk about nostalgia. Not really. This is really a dispatch from the department of you can just do things because Paraphraractic, a a Commodore revivalist, he just did things. He went to the people who owned the IP rights for Commodore and and were the successors of the company and said, "We want to license or buy or whatever the right to use the Commodore brand on these these revival products." And they got back to him completely reasonably because they were holding a wasting asset. They weren't manufacturing anything. Uh and said, "We think we'd like to sell you the company." So, he ended up buying the succession of Commodore. He is Commodore now. So, they have enough people and enough capital and enough bandwidth to do a uh a flip phone. They're resurrecting the flip phone. They call it the Callback 8020.

[00:07:33]Yeah, I think that's it. Callback 8020.

[00:07:36]And has a bunch of interesting features.

[00:07:37]Uh they describe it as a digital detox device. Now this is an important part of the new Commodore brand that uh social media and the internet have become toxic and we need a more a gentler more filtered computer experience especially for children.

[00:07:55]>> Your device should not permit doom scrolling, >> right? Uh paraphring products that that parents can feel comfortable giving their their children, but that's the subtext here. That's the subtext on a lot of their marketing. And so this is a flip phone that doesn't let you uh sideloadad a web browser. Uh it doesn't do Tik Tok. It doesn't do uh all the the horrible toxic social media channels, but it did does give you basic phone functionality and Google maps and and signal and most of the Android store.

[00:08:31]>> It's a really really good idea except ouch the price.

[00:08:36]>> The base model is $4.99. I don't think they're going to sell many many at that at that price because you're going to buy your kid a $500 phone that they might lose.

[00:08:45]>> Yeah. But it's a good idea.

[00:08:47]>> It's a very good idea. And they they need to cost engineer it better though.

[00:08:51]>> But you you've been doing a bit of that with DOSs game revival. It's like Oh.

[00:08:54]>> Oh, you want me to talk about that?

[00:08:56]>> Yeah. Yeah.

[00:08:56]>> Okay. There was a sharewware game, a DOS sharewware game I had fond memories of called Firefighter, and I've I've always wanted to kind of try to revive that. So I found a copy and I uh said to an LLM, decompile this into something I can modify. And it looked at at the binary and it it it I I watched it think it could tell it was originally written in Borland and Pascal from the way the string literal stored. So it decompiled the binary into readable Borland Pascal and even knowing the premise of the game came up with good variable and function names.

[00:09:35]It was amazing. So I looked at that and said, "Yeah, this looks like it works."

[00:09:40]And then I told it, "Okay, I I don't have a Pascal compiler." So, um, transcode this into Rust using this particular screen painting library over here. And it did it and it worked. And the whole process took me less than five hours.

[00:09:55]>> That's pretty incredible. And the I think you reached out to the author and the author was on board with all of this.

[00:09:59]>> Yeah, he was. He was. I I asked him for permission to distribute it as open source and he said, "Yeah, that's fine."

[00:10:06]>> So that's on GitLab and you should definitely check that out. There's probably a link in the description or on the forum. We'll we'll like it's there.

[00:10:12]Check it out. It's a lot of fun. But I think that's a pretty amazing success story and also a microcosm of both preservation and the kind of thing where it's like I don't know that this was necessarily worth the human effort. Like certainly there were probably enough fans of that on planet Earth that you guys could organize and get it done, but this was a much better outcome.

[00:10:31]>> This program I don't know whether it was worth the human effort, but there are applications of this technique that certainly will be. One of them uh them that's on my mind is I know that there's a lot of orphan software running numerical controlled uh CNC mills, things like that. And in some cases, the companies that that uh wrote the software have gone out of business and nobody has the source code anymore. And what are you going to do when you want to modify this software or uh move your your fabrication process to a different software to different software using a different input language you're I mean before this technique of decompilation, you would have been pretty completely screwed.

[00:11:12]>> Uh there's a there's a local company that I sort of help deal with things.

[00:11:15]They make duct work and they have a DOSS program that does a very good job nesting duct work and also running their plasma cutter so that they can cut shapes out of things.

[00:11:24]>> And all of the new versions of the software are all subscriptionbased >> and orders of magnitude more expensive than the software they paid. I think it the for their whole setup I think it was like $250,000 in like 1989 or 1990. and they've been able to keep everything running for the last, you know, they still make duct work and they still, you know, replace the machine, do the thing, but they're not willing to move to software that costs $25,000 a year for subscription when the DOSs version has been going forever. Now, it might be slightly more efficient in terms of like packing the shapes onto sheet metal or or whatever, but this is also a great example of how that kind of um crappiness in the industry, like sort of over over the long-term initification, you might not notice, but here it's it's it's it lays pretty bare. It's like the machine shop would literally be able to be a better machine shop if they had more advanced tools, but the more advanced tools are locked behind glass, for lack of a better way to describe it.

[00:12:23]We seem to have strayed somewhat from from the topic uh you know, Linux news.

[00:12:28]There is actually kind of a lot of Linux news this week. Kernel updates. Uh you noticed the IOPS improvement just from deleting a couple of lines of code.

[00:12:37]>> Sometimes micro optimization is your friend.

[00:12:41]It definitely was like two to 5% better IOPS >> from deleting one mess set. That's crazy.

[00:12:49]>> No, no. I mean the the memory >> puts in a tight loop.

[00:12:53]>> Yeah. Yeah.

[00:12:53]>> So, um that's uh that's one of the problems. Not really problems, but that's one of the architectural things facing ZFS. ZFS was never designed for uh storage that is a significant fraction of your main memory bandwidth.

[00:13:06]>> Yeah. And so it's doing a lot of work to make sure that it touches the disc as little as possible, but sometimes it's actually less work to just hit the disc a lot.

[00:13:15]>> Yeah. Yeah.

[00:13:16]>> So, um, also, uh, XFS, uh, finally got patches to make it work with shingled magnetic recording. Are you read in on shingled magnetic recording?

[00:13:28]>> Oh, that's where the the the the zones in the magnetic material are tilted.

[00:13:32]>> No. Well, they're uh it's uh dealt with in blocks and so you have um the readr head on the mechanical hard drive can't be physically positioned uh accurate enough to only erase what you want to erase. So you have to deal with it 100 or 500 megabytes at a time.

[00:13:50]>> And then there there are padding zones on either side of that. So when you have to rewrite something, it doesn't rewrite something. It looks kind of like an SSD where it'll try to write it over here, but it's a mechanical drive and so the fragmentation really really bites you in the butt when you're dealing with that.

[00:14:06]And so you need to do some kind of accounting to deal with that. And the good kind of shingled magnetic recording is when the operating system deals with it because the computer on the drive will never be sophisticated and complex enough to really truly deal with um with that type of recording.

[00:14:22]>> Man, does that sound like a huge pain in the ass. Yeah, but in this flash shortage, you know, I would think the shingled magnetic recording will bring ahead the um uh will bring forward the uh uh flash revolution, but then the flash shortage is kind of like I don't I don't know. Um uh you get better density with it as well. So shingled makes a 3 and 12 in drive that can store 40 terabytes a little more accessible. There's a there is actually there's a lot of Linux adjacent news that we will probably talk about in his build video because uh Wayland and speech recognition and local AI and like all of those things factor a bit into stuff that we worked on this week. So that was a lot of fun.

[00:15:09]>> Yep.

[00:15:10]>> So IKEA is selling a temperature and humidity sensor for around $30. That's what I I brought with us. I forgot I forgot to grab it. But it uses a matter protocol. It's an open protocol. And so even though it's not like weekly Linux news, we can talk about how the open source influence people are not willing to subscribe to things and that is also finally influencing internet of things things. People experienced the uh Google Nest thermostat mysteriously. It's like oh this is going to stop working after two years. Who replaces their thermostat after two years? Um, and so now we have sensors and smoke alarms and things like that that >> function as they always have since time in memorial, but can also integrate with home assistant and smart home and this this sort of thing. And so >> at some point as people figure out that the that go through the vendor servers for everything is a is a toxic swamp, there's actually going to be market pressure on the vendors of these edge devices to go open source.

[00:16:09]>> That's happening. Yes, that's with IKEA like I think it's a watershed moment that IKEA is selling something that uses the matter protocol and is doesn't depend on Apple's protocol and doesn't depend on some third party's proprietary protocol or portal or anything like that. And I think that is kind of a watershed moment for open source in that like IKEA it's like people are not willing to buy the thing that requires the proprietary cloud.

[00:16:36]Maybe the subscription vampires will turn out to have been a transitory phase in the evolution of the internet. That would be nice.

[00:16:43]>> That would be nice. Yeah. There are things that kind of resemble a subscription that um seem to work out like Steam for example. It's like you buy your game and it's like you you trust Lord Gabin to be benevolent, >> right?

[00:16:57]>> I think that is the the willingness of people to tolerate a subscription like it ends pretty far or pretty close after what Steam does.

[00:17:04]>> Yeah. Yeah. that people don't want to feel like they're in the wrong end of a of a predatory extraction machine.

[00:17:10]>> Yeah.

[00:17:10]>> They'll pay subscriptions, but they won't pay to be used.

[00:17:13]>> Yeah. And more and more people are noticing that they are used because I would say that a lot of the market success of of some companies depend on customers not noticing that they're used, but there's been kind of an inoculation in popular culture. is like, "Wait a minute. This is not the most optimal arrangement cuz people were willing to enter into that when they didn't realize."

[00:17:35]>> Yeah.

[00:17:35]>> Some of the Steam Machine benchmarks have surfaced. Those are probably going to be expensive. Like a It's got the They've got the uh the Commodore problem. I bet those are going to be north of $1,000.

[00:17:45]>> What? The the the new Steam Box.

[00:17:47]>> Mhm.

[00:17:48]>> Uh the last price I heard quoted was $6.99.

[00:17:52]>> I think that was way before the shortages.

[00:17:55]>> Oh, >> we'll see.

[00:17:57]>> Yuck. I don't have any inside baseball.

[00:17:59]I don't know on that. And Steam Valve is very good about keeping all of that um keeping all of that under wraps.

[00:18:05]Graphine OS supported Android 17. Maybe I should take a fresh look at Graphine OS.

[00:18:10]>> I would have I would have loved to move to Graphine OS years ago. Unfortunately, it's only it was only supported on a handful of very high-end flagship phones, and I'm not going to spend that much money.

[00:18:21]>> A very kind fan sent me a um a Pine phone, which if you're watching, thank you again for that. And that was really a lot of fun to experiment with. I never managed to get a video out of it because there were always showstoppping bugs and I felt like I was really crapping on something that I wanted to encourage.

[00:18:38]And so it's like this is great. Keep going. But you need to keep going like kind of a long way.

[00:18:46]Sorry.

[00:18:47]>> That's the heartening thing about Cosmic. It's already usable.

[00:18:50]>> Yeah. They've come a long way in very short order and they seem to be accelerating.

[00:18:54]>> Yeah. So that is it is that that is exciting. That's kind of a microcosm for Linux on the desktop. It has come a long way. It has a ways to go, but it's accelerating and it's very nice to see that it's accelerating.

[00:19:06]It seems not to There was a time there there's a there was a brief moment where it kind of looked like that um the Linux experience, for lack of a better way to describe the ecosystem, was going to get bogged down with really silly calorie burning distractions that no one should care about. It's like, yeah, I mean, that's fine philosophically, whatever.

[00:19:26]I'm just going to go over here and fix the window manager.

[00:19:28]>> Yeah, >> if you I think the audience knows what we're putting down with that way of describing it, too. I think >> I have to be a little oblique here, but you know, whatever. It's good.

[00:19:39]>> Uh, do you have any parting words of wisdom?

[00:19:43]>> Keep on doing what you're doing, guys.

[00:19:45]It's working.

[00:19:47]>> We must continue. We must continue the relentless reasonleness.

[00:19:50]>> Yes. Eventually, Relentless Reasonableness will win if we will just be uh un um we won't take the hint.

[00:20:00]All right. Well, thanks for hanging out.

[00:20:02]We'll see you next week.

[00:20:04]And then through the magic of editing, uh if we forgot anything, we can just dump that in. Make sure the camera.

[00:20:12]>> Was that what you wanted?

[00:20:13]>> Yeah. Yeah. No, I think it's good.