Tech Over Tea - The Linux Distro No One Talks About | René Rebe

Episode Date: June 7, 2024

Today we have the one and only René Rebe on the show, the developer of T2 SDE one of the very few standalone distros that is severely under represented in the media alongside running 2 youtube channels, Code Therapy and Bits inside ==========Support The Channel========== ► Patreon: https://www.patreon.com/brodierobertson ► Paypal: https://www.paypal.me/BrodieRobertsonVideo ► Amazon USA: https://amzn.to/3d5gykF ► Other Methods: https://cointr.ee/brodierobertson ==========Guest Links========== T2 SDE: https://t2sde.org/ Code Therapy: https://www.youtube.com/@MoreReneRebe Bits Inside: https://www.youtube.com/renerebe Blog: https://rene.rebe.de/ ==========Support The Show========== ► Patreon: https://www.patreon.com/brodierobertson ► Paypal: https://www.paypal.me/BrodieRobertsonVideo ► Amazon USA: https://amzn.to/3d5gykF ► Other Methods: https://cointr.ee/brodierobertson =========Video Platforms========== 🎥 YouTube: https://www.youtube.com/channel/UCBq5p-xOla8xhnrbhu8AIAg =========Audio Release========= 🎵 RSS: https://anchor.fm/s/149fd51c/podcast/rss 🎵 Apple Podcast:https://podcasts.apple.com/us/podcast/tech-over-tea/id1501727953 🎵 Spotify: https://open.spotify.com/show/3IfFpfzlLo7OPsEnl4gbdM 🎵 Google Podcast: https://www.google.com/podcasts?feed=aHR0cHM6Ly9hbmNob3IuZm0vcy8xNDlmZDUxYy9wb2RjYXN0L3Jzcw== 🎵 Anchor: https://anchor.fm/tech-over-tea ==========Social Media========== 🎤 Discord:https://discord.gg/PkMRVn9 🐦 Twitter: https://twitter.com/TechOverTeaShow 📷 Instagram: https://www.instagram.com/techovertea/ 🌐 Mastodon:https://mastodon.social/web/accounts/1093345 ==========Credits========== 🎨 Channel Art: All my art has was created by Supercozman https://twitter.com/Supercozman https://www.instagram.com/supercozman_draws/ DISCLOSURE: Wherever possible I use referral links, which means if you click one of the links in this video or description and make a purchase we may receive a small commission or other compensation.

Good morning, good day, and good evening. This is actually take two of this podcast because last week, last week was something, but yeah, how about you just introduce yourself again now that everyone's actually going to hear this? Yeah. Thanks for having me. Welcome, everyone. Absolute pleasure.

My name is Rene. My name is Rene Rewe, coming to you from Germany. And I'm a 25-year Linux developer, started around 1998-ish, with Rock Linux of all Linux distributions, and basically did touch most major open source projects since then. So Linux kernel, GCC, and still run the successor of this Linux distribution today, T2 system development environment, T2 SDE. Linux that is not yet another Linux distribution, but actually a very flexible build kit or some call it meta distribution.

And probably as an introduction, what is so special because so many people are like, why yet another Linux distribution? Like, hey, we're doing this for 25 years, but also we support all CPU architectures from ARM to x86 over RISC-V, PowerPC, Spark, MIPS, some collection. And yeah, that's basically, I guess, the introduction.

And thanks for having me, and let's see what we talk about today. Well, when people say this is just another distro, I think people forget how few distros really were around back in, so that would have been 1999, 2000? Yes? Before that, yeah. I think Rocklinux was started by Claire here in Austria. I think 1990 is like summer of 1996 or so, and I got there relatively early. So I'm for sure contributed to Rocklinux 1998. And at that time it was already in magazines and stuff and international. We had international developers and stuff from America, even Singapore and so on. So 1998 was already, it was going.

So it was before the year 2000, for sure. So before we can get into your stuff, can you give just like a brief explanation of what Rocklinux was? Because I'm sure most people probably haven't heard of that one yeah i'm sure i'm also also a little bit more popular back in the day um so in the 2000s there were like instead of like i always joke we have 2000 or probably more linux distributions so back in the early days there were like maybe only two dozens or so um many of the early ones you have never heard about are DLS or, I mean, of course, Slackware and so on.

I think Reddit was even based on some other, probably, whatever that was. And so Rocklinux was one of the few unique Linux distributions, and it was one of the early source distributions, similar to Gentoo.

I think Gentoo was probably invented some month after, also 1996, somewhere there. So it's because people were like, why don't you just Gentoo? First of all, it's entirely different, and it's as old, right? So these were the early Nuke distributions. And many of the others, like just last year, the things with Red Hat and people forking it even more, even the Oracle and SUSE and alliances and so on.

Most Linux distributions are actually simply forks, right? That's just a copy of another one. And Rock Linux or Slackware and Debian and others are actually those original inventions of even in the case of slackware and debian they technically were a fork of um i'm blanking on the name now but or sls or actually yes yeah yeah i even i mean it's it's i mean i was super young i mean i was at school right it's like running Linux distribution from school. People always don't believe me because like,

hey, I started when I was in school. So like, you're not doing this for 25 years. Like, yeah, sure. And yeah, even those details, I mean, 2000 distributions, even I start to like, this starts to decay. Even I would need to open Wikipedia for the details. But yeah,

it's one of those true original distributions. And the reason I found this, so I actually started, so I grew up with DOS. So we, as a funny thing is we had totally no Unix. I don't know how, probably how it is for most normal people, like my neighbors and like my parents or neighbors, like nobody had Unix, right? I didn't even know that existed. I grew up with a 286 and then quickly 386 of my father's. And so I did everything on, I started majorly on the 386SX25 with two megabytes of memory and initially not even a hard drive starting off as floppies.

And I've not even seen the greatest graphics because 386 didn't do much graphics. And only neighbors had like an Amiga 500, right? And I only knew like the highest end graphics I knew were the Amiga 500 graphics of our neighbor. And so I didn't even know that Unix existed. So when I only found this Linux through magazines and like I tried some

because back in the day, the magazines sometimes came with some Linux CD on the cover, right? And like I tried Debian and like in 1996 or I don't know, 95, 96,

it wasn't that amazing. And it's like coming from DOS and Windows. It's like, it's interesting,'s like coming from DOS and Windows. It's like, yeah, it's interesting, but like, do you, is that hard and cursed? And then I even purchased the SUSE back in the day, the big German distribution. And this was like better, but still, nah.

And this is how I found Rocklinux. I thought like, it's an interesting concept, but it could be so much more. And so that is how, and also from a magazine, I found rock Linux back in the day and started contributing there. Like it was basically because it was a source, you could build it because you couldn't just build Sousa or you couldn't just build Debian. And I wanted to change some things. And this is how I came to the source distribution that is

rock Linux and started there and basically did the desktop stuff there. So basically did the desktop Rock Linux since then until it continued to in 2003, I believe this continued into this more professional targeted T2, which is the only successor of the Rock Linux line of source distribution. So what was it that initially drew you to Linux? Was it just this cool thing that you saw in this magazine? It's like, oh, I like DOS. I like messing around with computers. Was there something specific about it

that really grabbed your attention? Yeah. So the thing is, so my friend and I, so it's, I mean, it's crazy that, I mean, today everything looks so easy, right, you just Google some stuff and Wikipedia and OSDev, org and stuff. Yeah, yeah. So back in the day, we had to teach all our stuff as books from the library. And there were only two computer geeks in this town back in the day, my neighbor and

I, right? We basically started game developing and the demo scene stuff. When we were 13 or so, we did high-performance assembly on 386 and tried to get the most out of VGA graphic. And we wanted to start an on-rope ring system because we were developing on DOS, and DOS obviously sucked except for gaming. And Windows sucked even more. Basically, in the 1990s, every advanced thing in Windows crashed. You did something

advanced, it crashed. And that is why I am not touching Windows anymore. Yes, it's 25 years since then, but nonetheless. So this is why Windows was for us a crashing experience. And we wanted to do 3D graphics in Windows 95. And surprise like every API call you did like crashed. Yes, those were the early days of 3D graphics but like, and my friend and I like, how can this suck so much? Like why does it crash like all the time? You can't even, like how should we do a 3D gram versus crashing stuff?

And so we were longing for a more stable operating system and we were on the verge to write our own. Like basically we were sitting and it was in Intel, like we called Intel Munich in Germany. It's like 14 years old. We would like some hardware reference

specifications of your processors. This time when Intel sent you some thick book and protected mode. It's like, okay, how do we program this stuff? Then we discovered we were sitting in the basement starting our system and then because we wanted something more stable than Windows, and we didn't even know, we never probably heard of Unix, right? Like with all the books we read.

I mean, I think we have not even seen a Macintosh so rare, this expensive stuff was at least in Germany, right? Not growing up in Silicon Valley. So I did not even knew probably that Apple existed from books and the library and stuff. So we were thrilled to find like, hey, someone else has done this protected world system that's not constantly crashing. And

like, hey, let's use this as a base and see what we can do with it. So you're like, Windows sucks. It's made by this multi-billion dollar company we're just going to do it better yeah or to be fair like back in the day microsoft wasn't that big right back in the day multi-million dollar company was that yeah i mean okay they were yeah but i mean windows 3.0 was really not that sure sure but yeah i mean obviously like a decade fast forward a decade later obviously

microsoft became yeah but given the business practices and and stuff and yeah marketing and so on yeah but like i mean but we grew up with DOS, right? I mean, and subjectively or objectively, DOS is not a very sophisticated operating system, right? So we mostly knew DOS and the graphic front end and that like even to 14-year-old look like, I mean, I can understand Steve Jobs was like, like, dude, what are they trying to copy there?

Microsoft, like this window DOS extensions, like, what are they trying to copy there? Microsoft, like, this is Windows extension. It's like, yeah. But sure, I mean, Windows NT, Windows NT 3, 4, I mean, obviously were more stable than Windows 3.0 and 95. But I mean, to be fair, I mean, obviously everyone knows from history how unstable Windows 95 was, right? It was just what a true protected mode or fully protected mode. Obviously, everyone knows from history how unstable Windows 95 was, right? It was just what a true protected mode operator,

or fully protected mode system throughout. But I mean, it's like Linus Torvalds, right? It's like Linus Torvalds, like, hey, this stuff is expensive or sucks and stuff. It's like, let's do our own, just basically some five years later, five, six years later, chiming in there on the big Linux movement. So you found your way to Linux, you found your way to Rock Linux. How did T2 come about from that? T2 basically, as often with projects, basically, so I did the desktop, and I became stable release maintainer. And as some who saw me on YouTube and Twitch last year probably see how much time it takes to put in there

to get everything working and do a little bit of marketing and so on. And so something that for Rocklinux became a little bit frustrating, like Claire wanted to do some web, basically what is GitHub today. She wanted some subversion frontend and all patches go through there and stuff. And I was like, do I have... My to-do is so long and I can't... Each patch... Some people would say, hey, continuous integration and quality and stuff, but there were also some personal things that I wasn't even aware of, right? Like in retrospect, I know some things that were going on. I only learned this later. So the frustration was probably accelerated due to some other things going on that I was not even aware of.

And so that is why eventually I and many professionals who used Rocklinux for embedded work said, okay, we better fork this. And like most developers continued with T2, which fun fact was only a temporary, like T2 stands for like try to, or technology to, it was only meant as like, okay, it's all placeholder name, but within a year, everyone used it. Everyone who knew about it used it. It's like, okay, let's keep this name. It was only meant as like, okay, it's all placeholder name, but within a year, everyone used it. Everyone who knew about it used it. It's like, okay, let's keep this name. It was only temporary, but tech people are not good

in marketing. And within a year, probably Rocklinux, because within a year, nobody was left at Rocklinux, I would probably say, which is a little bit sad. I mean, now Claire, of course, does super successful things. Claire, for those who don't know Claire Wolf, she reverse engineered the FPGA bitstream

stuff. Like maybe it was even, I mean, I just wish I had known some things back in the day, which I only learned a decade later, I learned some stuff, right? But maybe in the end, it's even better for everyone because Claire became highly successful in FPGA. She and her team and friends and everyone in the open source sphere did amazing work with reverse engineering all the FPGA stuff and writing the whole open source, Yoast, FPGA, tool chain, or maybe some more than FPGA, probably ASIC maybe also, like all the open source, yeah, silicon toolchain and stuff. And probably it was meant to be, but in retrospect, of course, it could be different. I wish it would be different,

but it is what it is. I also only wish, for example, that domains would not have been let slap and now probably some domain grabbers have the domain, which would have been cool if they said, hey, we have done this now. Hey, do you want to have the domain? I always like this when you open some historic stuff and you get to the right things and not some advertising site. Yeah. Yeah. Which is similar. I still need to try to contact the T2 Linux people because

the open source people who did the Apple T2 I was going to bring that up because now if you search for T2 Linux you don't find T2 SDE. It's only the T2 stuff for that. Yeah, that's a little bit unfortunate.

I mean, I see why. I mean, they came much later, right? Probably like a decade later or so. I mean, the fun thing is like every big company used the T2 name, which even Apple T2 and there was even Sun Microsystems,

Niagara Falls T2. There were even digital video and there was even Sun Microsystems, Niagara Falls T2, there were even digital video broadcasting chips. That's a little bit unfortunate. I only hope, like I see already in the future, one day the T2 Linux people will stop their Intel work and maybe they could pass a domain on. That's always would have been nice. If we don't rename this, maybe we should rename it someday

to something more flashy. Well, you did say it was a placeholder, but I think that... Yeah, it was a placeholder 25 years ago. Yeah, I mean, then first some microsystems did a CPU, then Apple did some SOC, some system on show, and who knows what, like it's

yeah, maybe eventually. The problem is if you rename it, like, you have to then think of a name. Yeah, and like all the decades of marketing, and then it's even more confusing, like Zetzing, formerly known as T2 and stuff. I also thought in the end it's not important, like name is name, it's like we have so what, it's people who want to find it, find it.

And I don't know, but maybe that's always so much things to think about, including the name. always so much things to think about including the name so when the project continued off of rock linux was it initially just basically a continuation of the project or has the goals of t2 sde evolved from what rocklinux was trying to do um so initially we tried we forked it with the initial goal to make it more professional for more embedded development because what most people also don't know, so it's basically

because people often don't quite understand and so what it's basically it's like build route on steroids or like more Linux from scratch, more automated or like Gen 2 scratch, more automated, or Gen 2, but for more cross-compilation. So that is what, when we forked this in 2003 or so,

the focus was more professional embedded development, like for your next ARM board, right? Like ARM or whatnot board here, right? My desk is full of embedded stuff. Like cross-compiling, basically, that's also how most people with T2 made a living. They did embedded projects like satellite receivers. We did carrier grade baseband stations there. I think I heard that even some hotel chain used it in IPTV in Austria, but that's not me, it's I only heard this. A Swiss company used it for archiving, even a company used it for some telephone. That's also not something I've done, only other developers. Because T2 supported Blackfin DSPs, basically we support everything. So because we were the only ones supporting

Blackfin DSPs. Basically, we support everything. So because we were the only one supporting analog devices, Blackfin DSPs, it was used for some telephone voice over IP PBXs. I can't even know where it's used. It's like build root and people use it for embedded projects, right? And that's what it's mostly meant to, what the main target of T2, but we continued, always continued to support server virtualization desktop. So we we all like all major core developers always have run it on the desk and i'm live streaming from t2 right like um today so that's fully featured desktop to embedded distributions that you can also run on your server and desktop right okay okay that makes sense so well one of the things that you did mention early, earlier on

is the hardware support. And I don't know if you want to get into that now, because I know you've got a lot to say about the way the Linux kernel doesn't handle hardware support. I know you ranted a lot about this and I've seen it on at least two different streams so far so if you want to yeah some years ago i got to to youtube so the funny thing is like so some years ago i got into youtube live streaming and and even like live streaming development and commentary and now i'm of course like after five years live on YouTube and Twitch like more people know me for being the Linux influencers for my Linux distribution

but running Linux distribution for so long certainly I mean not only I've seen most things if not everything but most things I also have most hardware that I run on a B-daily basis like SGI and Sun and stuff, PS3.

That gives me a pretty deep insight into how things really are and what most people only know from the surface, from news sites. so well one thing i've seen you bring up before is the whole the kernel cutting out drivers which from your experience just continue to work and it seems like yeah they don't they're cutting it but it doesn't seem to make sense why they're cutting it seem to make sense why they're cutting it. Yeah, that happens a couple of times. So the biggest recently was of course Itanium

support, which of course we still support. I mean, it started with 386, the minimum is 486. They also deleted already Spark version 7, the initial Spark, the earliest commercial Spark from Sun from 1990, which I also have in my collection, like all those vintage and

retro YouTubers. And I mean, dropping 386 and Spark version 7 probably makes more sense than dropping Itanium, because Itanium was sold until three or so years ago, which makes it just sold and still probably in operation in banks and insurances. And probably some NASA space telescope stuff. And Itanium makes completely no sense to me. I mean, this 1990 stuff, I can somewhat understand. The Spark version 7 is certainly more cursed. Spark version 7 doesn't even have hardware multiply and divide. Spark version 7 only has some multiplication step functions.

If you need to multiply 32-bit, you need to run an S multiplication, like mal S or so, like 32 times or whatnot. I mean, yeah, Spark version 7 was limited. I mean, 3.86... I mean, I understand the kernel developers don't want to maintain super old support. And I guess for 3.86, it's more like atomics, probably had less wide atomics and stuff for modern algorithms.

But like, I mean, that kind of thing's okay. But I mean, itanium, it was just sold recently. Yeah, I'm looking at the other media article. It says 2019, it discontinued. Yeah, and it fully worked, right? Like, when they deleted it, like, friends of T2, they even mailed, say, like, hey, it works, and hey, we want to maintain it.

It wasn't me personally. For example, Frank and Thomas. And they wrote, like, hey, it works, and, like, we can maintain it, and, like, they deleted it anyway, right? It's not, like, and, like, it worked, right? The only, like, on all the hardware we have, like, and that's the crazy stuff, like, we are only, like, we can maintain it and they deleted it anyway. It worked. On all the hardware we have,

and that's the crazy stuff, the Itanium team is a team of three that I'm aware of, my Itanium circle, and we have maybe six machines or so, or seven, and all except one worked, and the one that didn't boot, they fixed I think Thomas, Frank or Thomas, and all except one worked, and the one that didn't boot, they fixed, I think

Thomas, Frank or Thomas, one of them, they fixed it in an afternoon. It was literally one uninitialized memory thing, and it wasn't unfixable. The only machine that didn't boot was an afternoon job to fix for uninitialized memory. And so all the other machines worked the days they deleted it, and we keep maintaining it. Until now, it's just rediffing this patch. It's probably a three-megabyte patch of just restoring it.

The most minimal things, which as hobbyists, as DIY influencers at home, it's like it takes no time. You drink a coffee and it's rebased. It's not like people say it takes

so much effort to maintain. It's like, no, it's just work. It barely changed anything. It's there. We run the latest kernel 6.9. It puts on the tanium fine. And we will support it. We've written this on the website. We right? And we will support it.

We've written this on the website. We will support this another decade or two just because for fun and because we can for fun. I have actually a way to do so. I don't have... So many machines have very long, like many I've collected over the years, like the O2, SGAO2, SGA Octane, PS3, other PowerPC and SparkGear, many things I have for a decade.

The Itanium I only have since last year. It was working, however, so other people were using it, but now it's deleted. Like, hey, do you want this Itanium? Sure, of course I want to have this Itanium. And I have on my list gaming on Itanium. You will see this on my YouTube channel.

We will play some games. We will run desktop Linux. We have Firefox to restore support. I mean, yeah, okay, Firefox doesn't have Itanium support anymore. But sure, but that's not the main kernel. So everything else works. And there's other thing, for example, what else was recently? All right, PCMCIA. So, I mean, they deleted or even older Wi-Fi cards.

Yeah, yeah, I think the PS3 was on the list of the chopping block for that initial list. Yeah, and so this PCMCIA, I mean, they also just work, right? I took, of course, I don't use them daily, but of course I have an Apple, I mean, given I maintain PowerPC stuff for 20 years, of course I have, I mean, I don't have the most, even I have only I have three Apple, oh no, okay,

I have four or five Apple PowerPC machines left, like a cube, like maybe a cube. And these are super nice machines, right? Like, the difference, sure, you can't, like, triple A game on them, but you can, like, they're very high-end machines, sure.

It's 20 years ago, but there are people who just like their vintage cars, right? That's my favorite comparison. It's like, hey, drive some vintage Porsche, Ferrari, and stuff. And I just, like, it's like a museum, like a living museum. And theoretically, I could daily drive them, except internet is a little bit slow, or rendering websites is a little bit slow.

But as always, like, I could, 90% of my software, I could develop still on these machines. And I do once a week just for YouTube content and fun. And I was surprised. I mean, the problem is most people don't realize this until it's deleted because even I sometimes hear this in the news, I don't read every Linux kernel mailing list message. I took the Apple PowerBook out for testing and I was like, where's my Wi-Fi driver? Right? Why does Wi-Fi not work anymore? It's like, oh, it's deleted. Cool. And I restored

the patch and it worked. Right? So sure, it doesn't have the latest and greatest crypto, which is not ideal, but for some testing and stuff, it's the same for the P3. It's basically the same. And it's not like, I mean, yeah, for most, I mean, but there's still even P3, right? The fun fact, the most new users T2 currently gains are through all the Vintage and Retro support, right?

All the Apple, Power, Macintosh, Itanium, P3, obviously, which is also the last people who maintain this is NT2 and the funny thing is now that I finally with my YouTube content creation have

a little bit side income financing this because otherwise I couldn't sit down a few months and write a 3D driver right so irony is like now that I finally have like some free time and YouTube finance stuff of like hey we could finally have like some free time and YouTube stuff of like hey we could finally write

the 3D driver which we could have ever written and now they want to delete it right it's like like dude you can still use it it's sure a little bit low memory but yeah the irony is every Itanium and PowerMic has more memory than a PS3 but

a 256 megabyte is, yes, it's very compromised. It was compromised in 2006. Yes, it's more compromised today, but it's still a fun system, right? And on the other end, they add totally superfluous

support for Nintendo 64, right? Wait, what? I didn't know about that one. Yeah, someone at home has written support for Nintendo 64 and they send it upstream, like, hey, here's support for Nintendo 64, and they were like, okay, cool, we apply it, right?

And I mean, that's cool. I mean, that is, in my opinion, that is the way it should be, right? And I would also, like, my stand would be, as long, like, just leave the drivers in there, as long as it doesn't take a full-time job to maintain them. Like, if it's just there and just lives there, just let it live. And like, as soon as, like, sure, if it becomes super annoying to maintain, which in my opinion, it is not, right? Say, Bali, like, many things they delete, they didn't touch for 10 years.

Like nobody touched it in 10 years. Like, yeah, but it's working, right? Like nobody touched it in 10 years because it's working. Yeah, I don't understand this because what I also usually say is like, I grew up in the Linux world when we had no hardware support. For me, all the stuff, people had to put hundreds, if not thousands, sometimes even a team of five people had to put thousands of hours in to get the stuff working.

And now, besides it's working and there's no real technical need and stuff like they just deleted, it makes no sense. And especially given all the work it took to put in there. There's, of course, the other thing that maybe you have on your list, which I would also have not engineered the Linux kernel the way it is because for me, I would prefer a stable API, which the Linux kernel has not. And so if, like other operating systems,

there would be a stable API and you could have these drivers in binary form staying there and have this supported for 5 or 10 years, that would also be cool. But given the fragile and intentionally breaking nature, the Linux kernel argument is like, we want to optimize and we want clean code. optimize, and we want clean code. And the only reason those drivers break is because the news kernel people constantly refactor the code. They sometimes just rename, they rename put to delete or delete to put.

Sometimes, I mean, usually they say the name is better, right? Sometimes it's a rename, sometimes it's like they don't need an argument anymore. It's often such tribal stuff, right? Like an argument like put something or initialize whatnot. And the last argument might be Boolean like delete or initialize or whatnot. And so it's like often someone sees no driver is using that except one or whatnot. And it's like, yeah, let's

remove this argument because only this one driver used it. And this is how usually stuff breaks. And sometimes the names even go worse. Recently they renamed something where it's like, they renamed remove to put or so. It's like, how is put disk? Or is it like some block layer stuff like a year ago?

I was like, how is this? This name is not even better. To me, this name is worse. But that is how the stuff constantly breaks and why the P3 support every quarter. Yeah. The last year, every quarter upstream broke P3 support, right? With some rename and refactoring and API and API. And the irony the PS3 support is upstream, and currently it's broken, and it was... I think for the last decade, PS3 support, like Sony, PlayStation 3, I think it was for 90% of the time in the last 10 years broken. Like, I or others usually fixed it, and it's even currently broken because I sent something upstream and it's like, yeah, it's whitespacing,

but it's like, okay, whatever. So what I sent upstream is not even in there because currently the disk piece where disk driver corrupts the data took me off. And so it always takes me like usually a day or a weekend to find such things. Why does my PS3 not work anymore? Because they changed some,

they cleaned some stuff up and now it's corrupting the data. And, yeah. The last year, every quarter, they upstream broke once was Elf ABI, they changed PowerPC, the Elf ABI, so that broke

all the hypervisor interface, they broke the storage stuff, and it's like, and previously other stuff broke it, like open device tree stuff and um yeah for the greatest like yeah and that's in tree which like i'm i'm thankful that it's in tree like please don't delete it because if they deleted it's even more work for me to maintain but that is upstream but itanium which was working it's's deleted. So, yeah, it's... I would just let it in there.

The argument also is there is no maintainer, but I would simply, like, as long... Like, stuff that's stable, I would say doesn't really need an active maintainer. It's not like it... Like, it's feature-complete and working, and it doesn't need someone to touch it every month. And if someone changes some API,

it's some cool project, and replace it's fine um i wouldn't so quickly delete this stuff but it's basically my personal opinion so unless it was i guess known to be broken you probably would just say, don't touch it. The thing is, so if, if the news kernel would be my project, I would basically leave it kind of sort of in there forever until someone says like it's majorly corrupting data or whatnot, or like if it totally gets in my way. So the other thing is like every quarter that's visible since I do the reason I started to YouTube live streams, because I was doing all the stuff here in my office and nobody has seen this right now I saw all the YouTubers I saw like although and that's also like I basically wanted to do Linux influencing of between all the Apple

and Microsoft fans and so now it's visible on YouTube how much work all this is and the irony is like even all the stuff that's upstream I constantly have to fix, right? Relatively modern stuff, for example, the fiber channel, either SCSI or fiber channel controller QLogic, I fixed now three times in six years, that is in high-end workstations, right? Like they broke it, I fixed it. Usually I often only notice a year later because I like have 40 vintage machines here, Unix workstations. And so every other month I turn something on. And so it usually takes like a year. I turn something and it doesn't work anymore. Like, and so the QLA driver I fixed, they broke it three times. I'm sure it works for some Intel machines by pure chance, but at least the way it's set up and configured on this big Endian Spark workstations, it's three times broken.

And that's upstream. And that's with a maintainer because the maintainer constantly touches it. They constantly improve it and they constantly break it again. Other things, the SGIO2 framebuffer is currently broken. I've also not sent it upstream. Basically, I'm so busy fixing this. People always ask me, like, why'm so busy fixing this. People always ask me, like, why do I not send this stuff upstream

more often? The point is I basically don't have enough time because sending it upstream takes like 10 times more time and my to-do is still 600 lines long. each time, because the problem is in my opinion, I also find this upstreaming process.

People don't respond, and then there's white-white spacing, and then there's like, my email client broke it. Can you resend it with GitSend? It's like, oh, I resend it. For example, I want to make a, it's on my list of YouTube videos. So I fixed GCC for Apple G5 and Power PCs, IBM Cell. It's broken since four years and I sent it in three years. So I sent it upstream three years. I fired a bug seller.

I pinged them. I sent it to the mailing list. So what's happening is you compile your Linux stuff and you optimize for G5, Apple G5, or IBM Cell, which is more than PS3. There's also Blade servers and stuff. And it's optimizing for Power 7 instead of Power 4.

So this is Power PC ISA level 4, but due to bug and GCC, it's a one-liner, right? It's basically one number. And what many people don't understand, it takes most experts, me or at IBM, like,

okay, maybe at IBM, maybe it takes them one hour, but because I don't look every day into GCC, like, it easily takes me an afternoon to understand, like, what's even happening, analyze, debugging the compiler. And it was, like, one line's, like, and they changed it, right? Like it broke through cleanup.

It worked for a decade. They cleaned the code up for cleaner code and they broke some early PowerPC optimizations. Like basically everything before Power7 when you have alt-evac turned on. So I filed the bug, nothing happened. I pinged them, nothing happened. I sent it to the mailing list, nothing happened.

I pinged them on Twitter, nothing happened. I pinged them, nothing happened. I sent it to the mailing list, nothing happened. I pinged them on Twitter, nothing happened. And so eventually someone's like, but have you filed a bug? It's like, yes, here's the bug. Okay, have you pinged the mailing list? Yes, I have. Okay, can you ping again? So like a month ago, I was like, hey, by the way, it's a three-year-old bug. Can you finally apply it? Just send a letter to

Linus' house. Yeah, send a letter to IBM or the Free Software Foundation. But the funniest thing is, you will not believe, so now last month, I was like, can you please finally apply this patch? And then the maintainer of PowerPC at GCC is like, he replied, yes,

this one line, I guess that looks valid. Can you write us a test case? It's like, no, I should also write a test case for you. It's like, yeah, I will probably do live on YouTube because the problem is, I mean, that is a little bit procrastination, right? Like I should have, the problem is i'm probably not good at saying no what i because now four weeks passed because i thought hey i now i can do it live on youtube but because like basically i have so much to do and it's like like do i should do it i should do

it but what i mean i should have said realistically i should have said yeah that's a good idea um please write the test case i can review it for you so because because i i understand now like now doing all the stuff so open on youtube i understand and reflect much more and so in a way it's of course super stupid that someone from ibm asks some external contributor to write the test case for them and puts even what's basically like burning out open source developers and all the free like asking free contributors to contribute even more for free which had I not this

YouTube channel I should have said yeah this is great idea please please do it and I review the patch I review your test suit or I test your test suit for your stuff right in a way it's stupid that I I didn't response like so much work okay I do it on YouTube, but now four weeks have gone, right?

And this is also how those in the open source community who already do so much, like overload them, like basically wanting to do perfect, and then because they want to be perfect, stuff falls through the cracks and people get burned out having too much to do and Yeah It's not ideal

See I can tell that you're used to streaming because I can send you down a topic and you talk for 20 minutes and I Don't have to say a single word Yeah, it's I yeah, I mean I wasn't very could a time. And I was actually thinking, if it's a successful format, we could do this sometimes because I guess like once a quarter might even be popular to share some. Yeah. I'd be more than happy to do this at some point. Yeah, for sure.

There's a lot to say after 25 years. Well, I guess we probably should talk more about the actual project because we've gone like down random tangents. One thing I did definitely want to ask about is, so T2 SDE, the SDE is System Development Environment. What does that actually mean in the context of this project? in the context of this project? It means to develop firmware or embedded systems.

And it's basically, what we did was we took this popular IDE, like integrated development environment, and applied this to the whole Linux system. So basically, it was meant to be the whole IDE for integrated system development, because an IDE is just your one program, but that's basically compiling all your system. And what T2 does is basically T2 bootstraps everything from out of nothing,

cross-compiles, for example, PowerPC RISC-V and Spark and Itanium, which is also why with so few developers we can support everything. It's also very low code, right? It's recently realized it's actually a very low code packaging environment where

if you package stuff like Pipewire and Firefox, you usually need to write much less code and in much more systematically sorted way also with patches, like unbreaking GCC. We try to patch stuff as little as possible,

but because of all the constant changing environments and libraries, of course we have to patch stuff for new GCC, for new libraries, or RISC-V, or itanium support. But we don't patch graphics, we don't patch vendor wording. We only patch it when it's necessary, technically. And we also have all those patches nicely sorted in the source tree. Other distributions sometimes tar all the stuff up. So I find it much easier to work with them

with many other systems. Others tar all the stuff together. Then you always, as a developer, you need to untar this archive and pack it back together. Some other distributions also don't version there. So everything in T2 is versioned. Also the patches. Some other distributions and build systems

maintain patches not versioned, meaning like GCC Linux patches, they come from unversioned, not in version control, like mirrors. Like it's coming from distribution XYZ, mirror patches, whatnot. And I in general don't like this because as a developer,

I would always need to grab those patches first and they need to be on a mirror and stuff. And I, in general, don't like this because as a developer, I would always need to grab those patches first, and they need to be on a mirror and stuff. So in D2, this was what Rocklinux has been 25 years ago and still is today, that everything is nicely sorted, as simple as possible. It's also scripted in shell. Some people find it low-tech,

but most developers find it maintainable and readable. And it keeps shell. Some people find it low-tech, but most developers find it maintainable and readable. And it keeps everything. It's basically a very simple, minimal build system, kept as simple as possible to build all the 5,000 packages we have. So if I wanted to set up T2SD on my system, what am I getting myself into? Like, where do I start and where do I go from there? Yeah, so you can just download a pre-built binary ISO that's like a gigabyte. It's relatively minimal, but it includes these days all of XORG, Firefox, XFCE and Wayland. That is probably pretty compact for what we deliver, including, like, I'm basically live streaming

from such an ISO. You can just fully work on this. You have Firefox, and then use it. Like, it's super simple. It's not that much. I mean, given it's a text installer, and you need to know a little bit about the inner workings of a computer, it's not as simple

as macOS, but it's as simple as a BSD, right? If you can, we probably need to make it more simple. We recently started to make it more simple, but it's not super challenging. And from then you can just install packages from source if you want it. Like if you want to install some terminal or some shell or LibreOffice or other KDE and GNOME stuff, you can just as a BSD or a Gen2. Some people, by the way, many call T2 the BSD of Linux distributions. That is something what pretty many call us. And so you don't need to, like, it's basically as easy as NetBSD, OpenBSD, FreeBSD, or Gentoo.

And you don't need to... Obviously, we don't expect normal users to rebuild their system. And the benefit is users get an up-to-date system, and it's optimized for their CPU. Unless you have a super slow vintage system, if you have something halfway decent, because there's always a discussion, like some distribution switching to the next x86-64-version 3, 4, 5, 6,

AVX 512 enablement stuff. So with T2, there is not such a discussion. You just e-merge all your stuff for your native CPU if you want to and get the highest performance even on PowerPC, MIPS, and titanium. I'm curious about that BSD comparison because I'm not really sure what you meant by that, like calling it the BSD of Linux distributions. Yeah, being simple and this install from source, right? You have this package source tree and like it's basically you have this pre-built minimal system that's quite similar to install. It's also like minimal and light wide,

and you just emerge the open source world there from your package source tree, which is pretty, it's, of course it's not derived from that, but it's quite similar in concept. Okay. Okay. So I guess the obvious question is if it's source space, you know, the main source base distro that people go with nowadays is Gentoo. What is it about the way that T2SD does things that provides... Words are hard sometimes.

What is... What is... We're going to try this again, slowly. We're going to try this again, slowly. What does T2SD do that would convince someone who is a Gen2 user to maybe want to use it? For users, it's a tough call, given that there are 2000 new distributions. That is true. It's more for the professional way, right? If you want to ship a product, security product, a virtualization, or orchestration,

your virtual deployment, it's then probably more suited for developers and all those modern virtualization. You want to create a firmware um okay satellite top satellite boxes certainly are like here like like an ai box or all the modern stuff some firewall appliance some some storage server all the different like some some linux cube for for the office or stuff right then or any other like all the embedded stuff right like you have any other, like, all the embedded stuff, right? Like, you have any embedded project. I mean, we have even people using it for environmental controls, right? And once we even did some project with UPS, right? So, like, even UPS is this parcel service

they are using. It's, yeah, so for end users, it's tough. That's true. So basically for developing systems, I mean, sure, I'm biased. But I think the more simple and organized structure and the cross-trues, there is, in the meantime, like a decade later, there is some cross-compile supported Gentoo, some cross-Gentoo, whatever. I don't follow all those details. So there is some cross-compile supported Gentoo, some cross-Gentoo, whatever. I don't follow all those details. So there is something.

I don't know if it's a mainline Gentoo or if it's some... So there might be something for cross-compiling. But that's, of course, the main strength of being able... So what might shock most people is that all those ISOs, we have 36 releases. So we might have the most official supported CPUs and built variants. Built variants meaning alternative C libraries like MUSE, the more minimal and potentially faster MUSE or alternative C compilers like LVM Clang. So we have 25 CPU architectures in 36 variants like GL G-Lib C, Musel, and stuff.

And that might be the most of all Linux distributions. And what is probably shocking most people is that I cross-compiled all of them on just one single Ryzen 1750X, not even, I mean, sure, relatively high spec, but I can build all 36 ISOs on T2, cross-compile them, Itanium and stuff, everything, in, what was it, 24? I forgot. Either 24 or 48 hours.

Less than two days. Whatever it was. Maybe it was two days. Maybe it was 48 hours. But on one single machine, right? And the reason many other distributions struggle, even like Debian, who want to support more architectures is because they mostly natively compile everything. And then they can't support MIPS or PowerPC anymore because they simply don't have fast hardware. And certainly it took me also

quite a lot of work to make even Firefox. We can cross-compile Firefox, meaning make even Firefox. We can cross-compile Firefox, meaning compile Firefox for RISC-V on x86. It's cross-compiling on one CPU for another. It just took a lot of work. We do this for 20 years, like we did

20 years ago, 2003, 2004. We only could cross-compile the base stuff. In the meantime, we could cross-compile LibreOffice and Firefox, Rust. And that is certainly, it took 10,000 hours of work to keep all this, like set it up and package and test and debug.

And that's certainly quite some achievement, including even OBS Studio, right? I cross-compiled OBS Studio for RISC-V. And that's probably the biggest flagship feature of T2. So I know it's hard to compress tens of thousands of hours into something consumable, but what sort of work has gone into this cross-compilation toolchain to make doing all this actually possible? It's mostly all the scripting and debugging, right?

Because the problem is most packages are not written in a way, because people only develop on Debian and natively, they often are written in a way that they run binaries and stuff. And even other packages like OpenEmbedded is also one competing project for purely embedded stuff. They often do a little bit of a shortcut. They run those intermediate binaries that are often used. They run them through QEMU or other such emulators. And the problem with that is we don't do this.

And the reason is, first, maybe three reasons. The first, I find it a little bit a hacky solution. The second is it's slow. And the third is we don't have QEMU for architectures. If we would do that, we couldn't support Itanium because there is no Itanium QEMU. We could not... And many others. There are also other architectures that don't have a stable or not at all QEMU port. Sometimes they also remove it right even in qmo it's not supported um so yeah and so it was mostly i mean i should also say it's easier today like basically all the hard work was basically 20 years ago and like spreading this more throughout the open source ecosystem. Because 20 years ago,

there was even less notion of having proper cross-compile support. And even GCC, for example, like until some years ago, like maybe like even we and T2, we didn't cross-compile C++, for example, because even it just always takes like dozens of hours to debug why it doesn't work. The problem is also it often fades in totally obscure ways, right? You see this on my YouTube channel.

Sometimes I sometimes have live streams, like four hours just like debugging, like why does it not work? Like why is this so cursed? Why is random stuff happening? What is even going on here, right? And even, yeah, so, yeah, it's just a lot of detail work, like, and making C++ work. Even Rust and Cargo, like, nowadays,

like, it didn't even exist five years ago, but now more and more stuff needs Rust and Cargo. Just integrating Rust and Cargo to be somewhat properly integrated and cross-compiled and working, this was probably a dozen YouTube live streams. So it's easily hundreds of hours. That is what people don't see and why I turned to this to be also a YouTuber. Because people never believed this. But it's working. It's like, no, it's not working. It's only working because people put hundreds of hours in there making it work and debugging it and stuff. Is that the reason you feel like, oh, don't?

I only wanted to say it's also recently... Maybe you have this on your to-do list there. I also stopped desktop Linux for a decade. I don't know if you have this on your list. But it took me, like the last year's live on YouTube, it took me it took a thousand hours just to get all of KDE and GNOME updated and working again, right?

Just recently, like just two weeks ago, I live streamed 11 hours making most of KDE 6 working. And this is how much work it takes, right? Like this dependency, this new package, K opening hours, K indoor maps, one more dependency.

And it's like one hour, another hour. At the end of the day, you've worked 11 hours just updating all of KDE. And that's what people don't see and probably don't really value mostly for downloading linux for free i think because most people they use a distro like ubuntu they use a distro like fedora or even arch or gentoo where they don't really have to think about anything that went into it if you download ubuntu you don't have to think about most of the packages you install because it just comes as this big thing right like most of the drivers you don't manually select i want this package i want that package at least on something

like art you at least think about the individual things so there's at least some concept there that maybe there's some work that goes into making that work. But when, like, I've got the list open of all of the ISOs, and just having this in front of you, just this by itself indicates just how much work goes into getting something like this actually functioning. And I don't know, I hope that goes at least some way to making people appreciate that that kind of work um but i i did definitely want to expand upon the the kde and gnome stuff because i my my understanding is soDE, with their library stuff,

they used to have it as one, like, big ball of libraries. So it was like a 500 megabyte library package. And they split that out into individual packages so that you wouldn't be installing every single KDE library just to install one application. But now that means there's's i don't know how many different library packages 200 different i don't know a lot of them yeah so yeah if you can expand more upon like what dealing with that is like and then just the general issues with

getting these big desktops actually working. Yeah, I think it wasn't that large. I said this while I was doing that, and I remember it probably was more like 30 megabytes, like KDE so-so like 20 years ago, like KDE 3. I hope it was only 30 or 50 or whatnot, hopefully not 300, but my memory might be wrong. But it probably wasn't that huge. KDE dev gave you that number, so I don't know.

They might have been exaggerating as well. But so I... The problem is, for me, it doesn't even have much of a benefit. So I understand, like, fine-grained dependencies and stuff are great. But the problem is, they are hard requirements anyway. So I only packaged all of this. Like, hey, opening hours, it's probably some like parsing email or like map, like business listings or so, I guess. So many dependencies, I don't even know what they're doing, right? It's only from the name. And there's even indoor maps, right? So I only packaged all of this because I couldn't otherwise

compile KMail. All I wanted was to have KMail, the mail client building. And I don't think they're optional, right? So if they are not optional, I would rather have some PIM, like personal information manager, like KDE PIM or whatnot, like some 30 megabyte thing instead of like 20 different. There's also like parsing this, like parsing MIME. So, I mean, so KDE, we have probably 400 KDE. I mean, I could check. It's probably 400. The logic with the individual packages isn't necessarily if you're using KDE.

It's if you install a KDE package outside of KDE. So if you wanted to install something like Dolphin and you're not on KDE, you don't have to download every single library or Kdenlive or Critter or something along those lines. Yeah. I'm not so sure. So even in D2, we have currently 445 KDE packages. That includes all of

Dolphin and stuff. So it's like probably 200-ish libraries or whatnot. But I mean, so I had to package all of them because they were hard dependencies. I don't know if you could somehow disable this maybe um but to me it looked like you need you need some anyway um at least most of them um so i don't know i mean also i mean there were really really many right like it's um like plasma there's even like active activity and activity stats and i needed them, right? It's not like I packaged them for fun. I needed them because I couldn't build stuff. Like, I couldn't even build KWin, the window manager, or some desktop stuff without all of them, right? Like, without activity stats and stuff. And this feels really, I mean, it's, it's really excessively little small stuff. And for all the 2000 distributions,

there needs to be some maintainer who goes through all of them, package them, make them build and manage the dependencies. And like 20 years ago, I was actually a KDE user. And so they're like 20 years ago, they were like 20 source tables, like KDE base, KDE ellipse, KDE PIM, and KDE games a KDE user, and so 20 years ago, there were 20 source tarballs, like KDE Base,

KDE Ellipse, KDE PIM, and KDE Games, KDE Education, and then some, like probably 10 major source tarballs in like 10 languages, like KDE I18n, DE, US and stuff. I mean, 20 tarballs, like half a language, that

was okay, like now we have 400 or whatnot. With KDE, I mentioned this also in my stream. I'm not the most amused by the packaging because they have like three different version numbers. So they have some stuff that it's, so some stuff is 6.0, some stuff is 6.1 or now 6.2 since this week.

And some stuff is 5 or 4. They have three different versioning schemes. This framework are 4 and the Plasma is 6 or whatnot. And they also have three major locations for downloads. Even for me, it's three times more work because each time I need to look, is it there? Is it there? Is it there? And this is...

I mean, not only did it take me three times the extra time, you always... Then you miss something because there are three places, and then you miss some update, and it's very suboptimal in my opinion.

But, I mean, maybe it works amazing for them as a developer but as a distribution maintainer it's very verbose well i they're not gonna change that but there is one thing they're working on that should make things slightly easier uh probably not for you but for distros like ubuntu so gnome has their release cycle basically lining up with the Ubuntu release cycle, where every six months they have a new release.

KDE is working on getting their libraries releasing that same sort of format, so they do these big releases every six months, and that should make it slightly easier for them to package, but it's not going to deal with the fact that you have 400 different packages you have to sort out

and make sure you don't forget any of them, make sure they all get updated and all compile correctly. Yeah, I mean, now it's packaged, right? I mean, now they can just like, maybe just don't change it because now we packaged all of it. Please don't touch it. Now, we package it. Please don't touch it. No, it works. But as a developer, I mean, even as a, I mean, also software developer, professional software developer, right?

Like even as a, like it's really many small libraries, right? It's, it fits a little bit like says, I don't know if you have seen the Rust cargo micro dependency hell. Yeah. It's the same thing that that web has where you have these little libraries where it's like yeah 10 line long libraries like that doesn't need to be its own library yeah yeah i mean k opening i was probably can't do so much stuff right okay indoor maps right like it's and i also i forgot i forgot to point this out in my live streams i would also software deploy i would say this is not in my live streams. I would also say, this is not, in my opinion,

not the best way to implement it because I would say that this stuff should be more optional. Like, if you build KMail, it shouldn't require this on building. It should have some flexible plugin infrastructure that, in my opinion, all those opening hours and stuff

should be more like add-on plugins so that you build KMA without all this stuff, like basically the core, and the other stuff should be plugins. Like having it external libraries, but being maybe hard dependency?

I have to say maybe, because now we have 24 build systems. It's also my fighting windmills here. A decade ago, we had 10 build systems, and they all sucked, and everyone invented another build system, and now we have 24 build systems that suck even more. Because nowadays, I have to say maybe hard dependency,

because 20 years ago when I was young, you did configure, and usually, yes, the auto can suck. I know. I'm with you, but now we have CMake and Mison and Ninja and stuff. But now, in my opinion, it's super stupid that usually the CMake stuff, the superior CMake stuff, I don't know if it's hard to even, because as a packager, I run run this stuff it tells me like hey opening hours isn't there right and then i would need to do extra work to look up is it really like could i disable that i i didn't check i mean the stuff told me it's not there so i packaged it it is

possible that maybe you could disable it um which 20 years ago it was always like it only built what was there what was optional. Nowadays, most packages I've seen, they hard code this and most things are on. And in CMake, for example, and you would need to manually disable it, which I find stupid. If I configure the stuff I have used, it was how it was 20 years ago. And nowadays it's like some arbitrary choice of some developer and most stuff is on. So maybe it could be disabled, but I would still find a core K-mail and optionally dynamically loaded plug-in stuff

for advanced indoor maps and other goodies and opening hours. Preferable, but that's how it is, you've you've brought up the whole thing about there being like 2 000 different distros before and how they all need to worry about packages i one thing i will say is most distros are just not real most distros are basically just a glorified post install script they they just use the debian repos or the ubuntu repos or the arch repos like one thing you brought up one of your streams which you got very confused about and just hadn't heard of was uh garuda linux which is relatively popular amongst desktop users but you got really confused i saw

you be like what is what is this what is dragon eyes mean i don't understand this um but like that all that is is just a post install script for Arch in the form of an ISO and all those packages. I think they might have their own little repo, but the rest of it just comes directly from the Arch repos. Yeah. I, yeah, that's, yeah. That's probably another thing, like remember last week's call when things happened? What we also wanted to talk about is like getting PR rights. It's basically coming back to that. That is so sad for us.

I guess our website is not the forefront of web technology because we spend all our time developing our distributions. But that is all those themed distributions getting more PR than like hey we support all the CPUs and build all the stuff and nobody cares but hey someone made a dragon themed distribution

they also have an eagle their main one is an eagle it doesn't matter at all but hey you open the website you just see a giant picture an eagle what so yeah um yeah yeah we last week wanted to also talk about getting pr right so when that's also when i sometimes i mean obviously i make sometimes funny and frustrated youtube because it's just it's so sad right like when you you put so much work into that and

then you get less PR than someone making a theme post install script right and that's and especially this year like this year I and now I show all this live right especially this year, like this year, I,

and now I show all this live, right? Streaming this stuff and like you, you write documentation and you write press releases and you don't even know what you should do anymore. You, you, you, you try to do it like a big corporation each year, better and better. And, and still nobody writes about your project just because nobody knew about it. It's this chicken and egg, right?

Like big media only writes about things that are big and known and you're not getting big and known before like doing all the stuff yourself and starting your own YouTube channel to do your own PR to fix it. So that's with all the 2000 versions like yeah that's i know you specifically pointed out like pharonix and like a lot of the articles being like oh here is um we added support for some controller or it's like here's a GCC release candidate or just like random

little things like that why do sorry what I was going to say I love the delay on discord it's great what I was going to say is why do you feel like T2SD

T2SD just doesn't get like any sort of media coverage. Even from outlets like Foronix who seem to just cover basically everything. Yeah, Foronix, I don't know. I think the author has some preference for some other news distributions, but it's not only Foronix.

I criticize, I mean to give him credit, he wrote about us once a year or two ago. So there's that. But it's not only for Ronix though. Although I would say I would

appreciate if they write about, because like every second or third post is like some release candidate or stuff. I think his news site would... The space

would probably be better spent on some actual news and not some release candidate or other rumors or other discussions. But it's also others. And it's even our big local news here in Germany. I mean,

the biggest thing is we're not yet big and known. I think for the most part, it's a chicken-egg problem before you are big and known. Then it's also not sexy enough. If it would be flashy with a dragon, which is also why I keep...

I mean, not only I want to use my P3, but because I think it's fun to sometimes use, and we will definitely do more interesting stuff with it. But it's also why I keep some of the systems maintained. Certainly, having P3 maintained, having Attenu maintained, having SGI MIPS maintained

brings all those exotic users and developers to us and gives us at least some extra attention. Of course, otherwise, it's like two developer and a build system, too technical. Certainly, we need to improve our website. We worked a little bit on that, but it's like, yeah, developers trying to do website.

But, yeah, everything also, after all, it's... I mean, my company is somewhat behind it, right? I have an IT company. Without my IT company, probably T2 would not exist since a decade anymore. I mean, so I have my business behind it, right? Otherwise, it would already not exist since a decade anymore. I mean, so I have my business behind it, right?

Like otherwise it would already not exist, but like even there's only so much 10,000 hours you can pour into your side project in your company. Sometimes you need to do some customer projects and stuff. But yeah, gaining, I mean, it's probably each probably each startup right like each startup visibility getting into the media um it's difficult probably everyone would agree and um it's and also like if it would be brand new maybe we would get more news coverage if it would be brand new but it's 20 years around. And probably

in some vicious cycle of like, people use it, it works, it's great and stuff, but because the media sees it's around 20 years and there's not much media coverage, then it's probably not that newsworthy.

Something of that sort, I guess. It's just my analysis of the situation. But we're working on that, right? Now that with live streaming and stuff, now we can put even more resources in this. We made using T2

easier, we improved the website. Next, we will improve the website more. And that's a boring thing, right? We want to do code, but then we need to improve the website so that news sites write about us, right? So it's not what we want to do. And of course, users are like, yeah, but the website sucks. It's like, dude, but have you seen what our stuff can do?

But that's why people buy Apple stuff, right? Because it's PR and marketing and yeah. Please don't make your site like the Apple site. I genuinely hate Apple's website. You have to scroll down five different images before you get anything literally of value.

It's really bad now. It's nothing but marketing, marketing, marketing, marketing. Oh, marketing. Okay, now we get to something useful. Yeah, that's true. I also recently discussed

like why packaging KD and Konoma. I even thought like the KD and Konoma websites are not the most ideal if you open them. In my opinion, not the most useful marketing like some IRC chat. I would say that KDE and Gnome probably should copy some Apple marketing, maybe, at least as a start image. I can't speak for when you looked at it, but right now on the Gnome website, they have a button

near the top that says, get Gnome. There's like a picture of a laptop and then under that, get Gnome. I don't know what the a laptop and then under that, get Gnome. I don't know what the heading is. Yeah, but it's not, so I just opened it, but it's not very flashy, right? Sure. Like they don't have like some, I mean, I meant like both websites. I mean, sure. I realized I'm a code guy and my website is not the best, but like, hey, Gnome are bigger, they have some big foundations and stuff.

And they, so as someone who wants to see Linux desktop thrive for 25 years, I would say the start picture should be more productive. Well, the KDE picture is cluttered as hell. I can't even tell what's going on in this desktop. The KDE picture is cluttered as hell. I can't even tell what's going on in this desktop. They don't show flagship productivity stuff. The GNOME screenshot is useless, in my opinion. It's like it's some web browser

and app store thing. And the KDE, they show... I even had to Google what the apps are. The KDE website shows some Mastodon client and some IRC chat, right? Yeah. And they should show there's some amazing graphic applications, some offers. It's like this is not really inspiring, right?

Yes, my own website needs to be better, but I also don't have marketing experience. I better, but I also don't have marketing experience. I mean, maybe they also don't have marketing experience. Yeah, and even it took me a while, like packaging this stuff, it took me a while to click around there to even get to the latest source list, right? So, but

yeah, yeah. But yeah yeah our website needs to be better and that's also but that's what people forget right like everything

is so much work and at the end you don't only need to do all the code for like 25 architectures 5000 packages making KD Chromework

Rust Cargo like nowadays see like GCC LEM Clang Rust Cargo

200 libraries Firefox and stuff and at the end of the day, also have a pretty website. And best, a handbook and a wiki. Yeah. But we're working on it. We are maybe matching

to this. I'm not sure if you had this on your list, but I also had the last before, like five years this, I'm not sure if you had this on your list, but I also had a 10 the last before, like five years ago, I had one decade of not using desktop Linux. Not sure if you had this on your list, which is also why our website is more outdated, right? I basically didn't touch the website for a decade because maybe matching to this KDE and GNOME, I thought it probably fits right now. So I used Linux desktop, exclusively with Linux since 1996, right?

Which most people find mind blowing, right? Like this was KDE one and GNOME one and so on. And I, well, Pentium 200, right? Or stuff like, and with Rocklinux, of course. And I used desktop Linux for everything. And most, at that time, like the year 2000, most people have never heard about Linux, right? I was using Office.

I was, everything was perfect, right? We had eventually KDE, KDE 2, GNOME 2. Everything was mostly good. But then somehow the desktop, like every year worsened for me, like KDE broke stuff, GNOME broke stuff. Every year we reinvented graphics stacks and stuff. And to a point where even I,

as a hardcore kernel developer and distribution maintainer, like then around 2008, like basically eight years later, like 2008, for me it was like, there was like compits and 3D desktop, and then they wanted to write stuff like rewrite Xorg and Wayland and stuff, and the audio broke. Yeah. Yeah.

And then I run a distribution, sure, I run an embedded distribution and stuff, but in 2008 it's like, okay, enough is enough. I need my desktop working, right? So we had these power PCs and stuff. We did some Mac software, so I switched to macOS for desktop, like many people. And so that is why... But I still kept maintaining T2. Of course,

we had huge virtualization projects. We had embedded projects. So for a decade, I actually didn't touch KDE and GNOME at all, which is also why we had to catch up with all this stuff now. And basically rediscovering the Linux desktop. But that is also, in my opinion, a sign of how lacking and maybe frustrating the desktop Linux experience is, right?

If people like me moved away from the news desktop for a decade. So in T2, I kept maintaining the core, like the kernel, the C libraries, compiler, even X, of course, X server. Like, for me, give me Blackbox, give me a terminal, I can work, right? Of course, I updated Firefox mostly, so of course Firefox was working. But with a decade of KDE and GNOME not updated, of course it was thousands of hours of work. But that also shows how lacking the Linux desktop,

in my opinion, probably still is. Like I am now back to Linux desktop since three years. And I even live stream with OBS and I have like most things working. But realistically, I have to say, I would wish by now the Linux desktop would be in a more mature, everything perfectly working state,

which in my opinion, unfortunately, I have to say it's not really. I might even go as far as say 15 years ago, like 2006 or whatnot, 2006, 2008, maybe stuff was working more stable. Now, I mean, the world also became a more difficult place, right? With all the crazy incompatibilities with smartphones and everything. What proprietary formats for sharing files and whatnot, Apple's pages and numbers and stuff.

But sure, we have LibreOffice. But sure, if you run, of course, you can run. I mean, I do. And you can run most things with Firefox or Chromium. You have a PDF viewer. Sure, you can get stuff done, but I would argue the Linux desktop leaves a lot to be desired. And I think we would be in a way better state if we would not constantly rewrite stuff. If we would not constantly rewrite stuff, like if we would just focus for a decade on perfecting everything,

making every smallest corner just work, audio for example, and not reinvent everything every five years, I think this would greatly improve desktop Linux acceptance. Well, speaking of things that just should work, I'm going to tell you about a bug that me and one of my mates discovered in KDE because it's really stupid. So you know how KDE has all these fancy desktop effects, like their overview thing and they've got an alt tab and all of this fancy whatever they want to call it so um turns out that they don't load these effects into ram every time you use one of these effects they are read from disk so if it is on a slow hard drive that can cause your entire desktop to lock up for multiple seconds. Yeah. And I mean, it was also like 2008 when I left the Linux desktop for a decade, it was also even then that stuff became like basically the KDE 4 and GNOME 4 or whatever

rewrite. They were so slow, right? I mean, back in the day, I had good hardware, but not the state-of-the-art hardware. But basically, everything also... I didn't want to buy every year's latest computer just to have KDE and GNOME snappy, right? And the craziest thing is, I mean, the running gag is, of course,

audio, right? And which the irony is, I mean, not only that last week we had some problems, but the irony is, I mean, not only that last week we had some problems, but the irony is like for decades I had working audio, right? I never had like, I never had so many audio problems like today. And for basically a decade, I was joking, like for a decade, everyone came to me like this problem with audio, that's not working. And I was like, oh, it's like, what is the problem? Everything is working for me. Turns out, I was not using this modern audio stuff. I just was using

raw ALSA. So for me, everything just worked for 20 years because I was just using ALSA and all the other people used this modern audio daemons, which I never liked. And now, that's also so stupid, where even there, the bugs are, in my opinion, like even I have

to debug things an hour here, an hour there. And even today, I was using Pulse Audio. And I got that to mostly work and behave. And then we switched to Pipewire. And now, and only after our last call, last time, I thought, I actually forgot about this. I knew about this bug. So on my Ryzen motherboards, Pipewire doesn't root any default audio. And only last week I debugged. Why? Because somehow this either driver or whatnot thinks there is no headset plugged in

when it is. And that's also so stupid, right? And then, I mean, not only that this is stupid, but then instead of just rooting it there, when that is a primary audio device, it's deciding to root it to some dummy null thing. It's like, how is this the same choice? And this is basically Linux desktop was working when like true headphone detection, one thing, Linux desktop working when, like, through headphone detection, one thing, maybe it's that, whatever, but not routing it to the primary audio device, but trying to disable it because there's nothing plugged in. I mean, besides you could have some line out speaker plugged in or whatnot, but I mean, these are these crazy decisions that I don't understand. Like, how is this the same choice to stop rooting it and root, like, actively making the decision to stop rooting it to the only audio device? And it's like, this is just so silly, right?

Well, the other joy of Pipewire is because on Wayland, Pipewire is the way we do video capture as well. So I have this fun issue where if I'm capturing my main monitor and then I change the input of my monitor over to my HDMI, if I'm trying to capture my console and I want to see it on my screen, my if i'm like trying to capture my console and i want to see it on my my screen that causes the capture i have in obs over pipewire to basically have a a null uh object attached to it this causes pipewire to constantly seg fault every time i change my audio level yeah fun bugs it's yeah but it's also because we have to rewrite. I understand rewriting and cleaning stuff up and people coming there and saying, like a decade later, the design wasn't good,

let's redesign it. But there's also a limit. Maybe we should do it half as often. I mean, I have not a gold solution, except do it half as often or design it better from the beginning. Or every other company would retroactively fit it

in there, like add video capture to Pulse Audio. And yes, maybe that would suck and whatnot, but the breaking of compatibility and it's like for drag and drop, for windowing systems, for every small little thing breaks, right?

I have to say so. I mean, except for the decades that I used macOS a little bit, which I always run RYT2 on servers and virtualization stuff, but I otherwise use fully desktop Linux, and I have to say I

used macOS extensively, and it also isn't perfect, right? It also becomes increasingly buggy. Now that Steve Jobs is not there anymore, and there's no oversight, and they put everything in there, want in there, and whatever. And Microsoft as well. Since I never used Windows, like really for me

and my stuff since 1996-ish or whatnot, most people people like, I only start virtual machines for reverse engineering or like having to touch some project code and visual studio shortly, but that is like nearly never like people really need to give me painkillers like for days for me to touch windows, even in the end. So it's not like windows is amazing.

Like, like I could not use Windows. And the crazy, I mean, but I understand that normal people are used to this. And some things sometimes work maybe a little bit better. But like, certainly Windows, I mean, Windows for me

sucks. And the craziest thing is that macOS, I mean, not only the modern macOS, for me it's a nightmare because everywhere there's a security nightmare. Do you want to load this? Do you want to load that, and bugs and stuff, right? Like SIFs, Zumber filesharing is super slow for strange reasons and whatnot. But also multimedia is even better. I mean, multimedia OBS runs better on T2 Linux than on macOS. On macOS, you can't even capture desktop audio without a third-party plugin for security or whatnot reasons.

It's not like on macOS, it's everything shining. It's like I was shocked. I wanted to make one live stream on the go for macOS. It's like I couldn't even record desktop video. It's like, what the heck? It's like, yeah. Or other, I mean, then you capture your Mac desktop and you pen your desktop and it follows the penning

and then it glitches strangely. So it's not not like it's bug free. It's increasingly buggy, but it doesn't make it better that the Linux desktop could be better. What do you actually run as your Linux desktop? What desktop do you run? Currently I use Simply Sway. Okay. With a la critty. I mean, most people see me, I mean, thisway. Okay. With a la critty. I mean, most people see me, I mean, this is so funny. The YouTube comments, like, now that I have a growing YouTube channel, like, the comments

are going, right? All the people, like, like, Duo is not using syntax. How can you program without syntax highlighting? It's like... Wait, do you not use syntax highlighting? Hm? Do you not use syntax highlighting? Sometimes not.

I mean, I don't play... Okay. I mean, sometimes I have it on, sometimes not, but I don't need it, right? I don't even realize when it's off. I disabled it recently because the latest Vim version colored divs so strange, like patch files, right? It's like pure color cancer and so strange low contrast, like light blue on pink or whatnot it's like it's like it's barely readable right that's like i just disabled it because like this patch was so full

of color it's like like do you wish it reads this and i didn't even notice that i turned it off because of this patch thing and then people mind like i can just code and people like how can he code without syntax like who cares but, the comments are really fun. And then also like all the people like, which desktop, which editor, which... I'm sure people are like, hey, why don't you try out Hyperland? Why don't you try out this one?

It's like... Yeah. I mean, basically, so that's also how I work on all the vintage systems, right? Like some of the vintage systems don't even have either no graphic or non-working graphic or not supported graphic. And it's like, yeah, serial console and just give me a, give me a terminal. And I can even work on a un-estimated frame buffer in Vim, right? It's fine. But so for a decade, I actually, like since KD and GNOME became too resource hungry for me,

I just give me X server. And for a decade, I run Blackbox. Now I run Sway. I tried Hyperland. It's fine. I mean, I tested KD and GNOME. Our default ISO ships with XFCE for end user, feeling at home and pixel art.

Yeah, Firefox. Yeah. end user feeling at home and pixel art. Yeah, Firefox, yeah. That's the thing, right? Then people desperately want Chrome. We have a Chrome package, but it's too often broken. Most core developers don't love Chrome the most it does. It's funny that so many users love Chrome so much when developers don't love Chrome the most. It's funny that so many users love Chrome so much when developers don't. Some areas still need some.

The next 10 hours live stream making Chrome fully work. I need surprisingly little, but I'm aware that most people need all of KDE and GNOME. But then even in KDE, even the PF viewer was broken. Even that i had to debug like

why does ocular not load its rendering plugins and stuff right and where you have to wonder like why does this stuff not just work but sure i mean people in debian and suza fixed that too but like why does it not work out of the box? Actually, one thing I want to ask, because you are using Sway, why didn't you just go and use some X11 window manager? Why did you bother using Sway? I did until recently. I mean, only because I was using Blackbox.

Even on my early YouTube videos, I was doing all in Blackbox, I believe, because that's all I need, right? I need to move some windows. That's fine. I only started because people, like all the black box, I believe, because that's all I need. I need to move some windows. That's fine. I only started because all the people were bugging me, like, can you support Wayland better in D2? We want Wayland support.

Wayland, Wayland, Wayland. Like, okay, okay, fine. I do Wayland. So I only did it because everyone was whining to me they want the latest and greatest. And then I obviously needed to test it. And then I switched to it, and it was eventually working. I mean, I I switched to it and it was eventually working. I mean,

I got used to it. It's mostly fine. You need to test the latest and greatest, right? Fair enough. I basically have the first most stable compositor there, Raylan compositor.

Sway worked for me with OBS and Hypalent too. I need surprisingly little. Just recently I cleaned up some vintage window managers, so we have dozens of all the vintage stuff right from i3 to Awesome and Red Poison and i3M. And IceVM, so if you, if someone wants to use T2 as a desktop, you will feel at home no matter what. And patches welcome, right? If we have over 5,000 packages,

so if something, I mean, surely there, I mean, we have FPGA toolchain, YOSES and KiCad and OBS, but certainly I'm aware that other distributions have more than 5,000 packages, although the numbers are skewed, right? People always like only 5,000 packages. Yeah, but Debian also splits them, right? Like we don't split them. So Debian has devil and dog and whatnot. So like 5,000 true packages with all artificial death, dog splitting.

One thing really to appreciate with uh your package listing is you don't have some annoying search to find packages it's just a giant list of packages on the website yeah yeah it's old school um i mean once upon a time this google custom search was working but google stopped the service somehow i need to still delete this custom search. Oh, you're right. You still have the bar there. It's like, yeah, we have Itanium to support. Yeah, but I mean, yeah, the website will significantly increase in the next months, is my promise. And yeah, I mean, we can,

we do all this, all our work at ExaCode with T2, right? Desktop, FPGA development, YoSys, KiCad, there's an electronic cat. I'm gaming on T2 since the year 2000. Yeah, that's this next, like, what's next? Next, we will improve the website. Yeah, that's this next, like, what's next? Next, we will improve the website. We will probably have to fix... We will iron out some last... With 5,000 packages, there's always one last thing that doesn't build right now, like Chromium.

We will fix some build issues, update Chromium, because people are asking. And we will probably also... Probably what most people don't know, because people are asking. We will probably also what most people don't know, we have not even mentioned, theoretically we support building Homebrew on macOS and BSDs using the T2 build stuff to build on other systems.

We probably should expand on that. Yeah, sure. We also use this in production. On our Mac development systems we build like stuff we need on Mac OS we build with T2 and like libraries and stuff and QEMU like we

for developing Mac software we sometimes develop like we run this on QEMU built with T2 homebrew on Mac OS but it's not like it's proof of concept so we probably should build this out a little bit more. So basically you could do all the things with

T2 Homebrew embedded desktop. And maybe that's also what most people don't understand, because, like, it's just so much, but it's this really simple code. Most people will be surprised how simple the code is.

Readable shell code and not obfuscated, thousands of libraries and frameworks. And the package is also really low code, right? Like zero, if a package is not totally cursed, it could be built with T2 with zero code for packaging. Most distributions have a lot of code for building the RPM and stuff. And if it's not zero code, then at least it's low code. If you need to quirk something to make

a package build, it's very, very, very little var assignment and hooking stuff or patch. It's extremely simple. Stupid simple. So what do packages actually look like on T2? So the source packages are some really simple key value format for the metadata. I mean, you could click... I mean, I can't easily share my... Because I'm using Sway. I can't instantly... I can't

just share. But if you... You can just direct me to somewhere on the website I can go to? Or something I can go to? If you go to Wim or so and click on the T2 source listing and click on the Wim desk there. Maybe Wim desk is... You find that? Sorry, what was that? On the T2 website in the package listing, if you go to WIM.

WIM. Oh, god. Uh, WIM. Am I hearing you? Or any other. Oh. Okay, yep, sure. And you see this, if you see this package listing and there is at the middle somewhere

this T2 source stuff and there is like Vimdesk. It should be some key value thing. Like it should be some brackets. I see. I see. Like IT, like it's like header, like information header text. And like basically what T2 does, that's basically the package format. It's a bit old school.

It's like basically like Tommel and Jamel. Like it's just we invented this 20 years ago before there was Toml and YML. But we're basically not changing it because it works. It looks like array assignments. Something like what you could have done nowadays with Toml and YML and stuff.

And so basically T2, that's basically all that's needed. T2 will basically figure out, it will unpack the sources, that it's like the DS of download stuff with a checksum. And so the T2 build system will download this, extract this,

and figure automatically out what's in there. Is it a makefile, CMake, Mason, Ninja, and stuff. And usually it will automatically just build. Unless, like if it doesn't build, like if it needs some code, you can add code or patches. There might be,

there's even a hotfix install patch. Like if something else is so stupid. Yeah, I mean, yeah. I don't know if you want to click on the hotfix install patch. But so yeah, if you need a patch, just drop it in there. And if you need to add some code

to set some configure options, it's just adding some other package install dependencies and add, disable, enable, whatnot. So it's either zero or very low code packaging. And T2 will do all the stuff automatically. Only packages like the Linux kernel, GCC, LEM, Clang, Rust, they have some more complex scripting that is shell code, like in Gen 2, for bootstrapping for all kinds of conditions.

What is your Linux kernel package code? The Linux kernel, That's just Linux. I mean, yeah, there we have some 60 packages because we have some architectures to maintain them. I've got to scroll through this big list of things to find where Linux is because Linux is a keyword in a lot of things. Okay, here we go. Found it. Awesome. Yeah. I mean, okay. And you got quite a few patches here as well. That is probably the most patched. So usually we patch nothing except like the Linux kernel. Yeah, there is some stuff going on. Yeah.

Yeah, there is some stuff going on. Yeah. I mean, I would love to not patch stuff, but there is hotfix. I mean, sure, some stuff I implemented newly. There are some completely new drivers from scratch for me in there. Like, including DDI, I've written live on YouTube. Like, if you want to see my stuff live,

you probably can drop a link to my channel. Like, I've written, like, there's the DDR5 driver for reading the SPD EEPROM and temperature. Currently T2 is the only distribution that ships this because I've written it. I should upstream this soon. I've also recently written live the Apple Intel

XServe front panel controller. And we also have a restoring Itanium. Of course, 0.0.0 IA64 is 3MB restoring Itanium support. But otherwise, if you delete all the patches, it's still built.

You just don't have all the hardware working. Oh god, there's a lot of... Okay, right. I... Yeah, I'm sure the kernel is a fun one to deal with. Okay, right. Yeah, I'm sure the kernel was a fun one to deal with. I mean, it only the last year's got out of hand. Like, since I started to fix everything that isn't working live on YouTube,

like, since then it got a little bit out of control. But on the plus side, it, like, everything is working. Like, SG Octane,-O2, P3. So some stuff is also not finished, right? Like, for example, the DDR5. So if you want your DDR5 temperature, don't delay installing T2Links today and you have your DDR5 temperature sensor working. But otherwise, some stuff is also not...

For example, for UltraSpark, I recently debugged my 8GB UltraSpark. My highest-end UltraSpark isn't booting Linux. It's crashing very early. I debugged a whole weekend, and I have a workaround for that. So if you have an UltraSp with 8 gigabit of memory you probably want to install t2 linux because everything

else crashes early in memory initialization but I've not like in one weekend I have not found the root cause I have a workaround for that that isn't pretty but like I still need to fully understand the special high mem like huge page

bits there and there's probably mem, like huge page bits there and there's probably some serious huge page confusion going on. So that's why some stuff works but isn't pretty. Like last night I fixed finally the Apple Cinema

displays to work. So some stuff is new and other stuff I really should send upstream. Also, the SGIO2 frame buffer I probably did last month. So X didn't start anymore. It's also like a decade old bug. It's like, yeah. If you have an SGIO2 and PS3, then you're

busy finding regression and fixing things. I mean, you're busy finding regressions and fixing things. It just... It's also not fun to send some stuff upstream, right? Yeah, you bring this up a lot. It's just like 10 times... Yeah, I mean, it's just so much work, right? And because sometimes people criticize, I should send more stuff upstream,

but at the end of the day, like, I'm tired and want to go to bed. And then it's the next day and new stuff is on your desk. I just maintain so much. Then it's sending SGIO2 framebuffer and then you're fighting GCC maintainers for three years to finally import the patches. Realistically, nobody else supports, like who else is supporting SGI 02, right? Or Octane? Like what is the point of upstreaming if no other links distribution is there anyway? So. Yeah.

This is because it's the main criticism, criticism people bring up. It's like, why is it not upstream? It's like, because nobody else cared and i i didn't have more time that's it's basically that is a really good point though like if if even like debian's dropped it by now for certain pieces of hardware is there really a reason to upstream it if you're the only one who's actually going to make use of the patches yeah um that's also why I really turned to YouTube. Because I want... To be honest, my company is... My company is

extra code. We do code stuff and stuff. Since I founded my company in 2006, my company is pouring way too many... Realistically, every controller manager I've already long said like like stop this like stop this stop this money draining business like i only do this because i want this stuff working and it's my company and like i can like like every

everyone else like would have x this project long time ago of course we have successful virtualization security like like the reality reality is everything pays better than maintaining this distribution. We're getting paid for code, we're paid for code review, we're paid for consulting. Like nobody, okay not nobody, very few people pay us good money to maintain distribution. Literally everything else pays more than distribution. And with YouTube, it finally allows me to put so much more work into that, making really everything work. And I should say it probably is worth upstreaming because then it prevents the stuff from getting deleted because then they see, hey, someone cares. And there was really a bug.

And many of the things I only did the last years due to all this new video streaming business. And I will do increasingly sensitive stuff upstream. And also, and this is a set thing is coming back to this vintage and retro hardware. I just like this hardware, right? It's just like people collecting cars. I don't pay crazy prices. I mean, some of the stuff I got for free, I mean, very little I got for free. I mean, some of the stuff I got for free, I mean, very little I got for free. I mean, sometimes people donate me like, hey, you want to have some stuff?

Like, sure, I want. I mean, some stuff I waited five or 10 years on eBay searches to get it affordable. I'm not paying $1,000 for an S-Jerk 10, right? Right, right. A lot of the stuff, like, I'm just lucky to watch the stuff for five or 10 years to get it affordable. And it's just nice collectible pieces. And there's still so much stuff I would do

eventually now with this YouTube thing, 3D on a P3. We could, since 2010, we know how to do 3D on a P3, just nobody had the time to do it. I did some proof of concept stuff. Proof of concepts, it's just hundreds of hours of work, right? So that's the fun stuff. I will eventually do it. Just a matter of time. We could do like only recently, the SGIO2 is new.

I got this probably last year only. We could even do a 3D on SGIO2. new. I got this probably last year only. We could even do a 3D on a SGIO2. I got it new. I will probably do it also eventually. A lot of the stuff is hardware acceleration. And that is the code that's being deleted, right? Like people hating XOR, but like, sure, is it? I mean, it's museum collectible kind of thing. I get it, but it's still fun.

And this is also like deleting the stuff is, in my opinion, also like taking the fun from this hobby DIY people who just want to enjoy the hardware. What else is here? I mean, theoretically, I have writing a microkernel on my to-do list. You might have heard this. I did briefly hear a bit about that, yeah. But yeah, a lot of this stuff is hardware. It's 3D.

Let me... Yeah, I mean... Yeah, there's also... But the 3D stuff is also fun. And I have done this, right? Because I've never written a 3D driver from scratch, which is why I tried this live on YouTube. Some four years ago with S3 Verge,

this infamous de-accelerator. I just wanted to see how difficult is it. And it turns out, at least for me, it wasn't difficult. And obviously, it's compared to modern graphics, not difficult. And I will expand on that so i will other people want to delete this but um sure the xorg and mesa card was old school see i get it um i would i just would not have deleted it it's like security and stuff it's like yeah you don't need to run state-of-the-art desktop and security on 3dfx Voodoo.

You just keep this code around there and don't touch it much, but okay, fine, it's deleted. And so it's also educational because with this old stuff, it's so simple. You can teach people in a YouTube video how to do it. With this modern stuff, like even I wouldn't, like probably in a year, like a year full time, I would probably not be able to write the. With this modern stuff, even I wouldn't. Probably in a year, a year full time, I would probably not be able to write the driver for latest AMD stuff, right? So complex as modern stuff is. Yeah, you need the whole department of AMD and Intel

of maintaining the latest and greatest stuff. And so that is also, from the educational perspective, deleting code that at least is so simple if theoretically so simple that people like hobbyists and diy and and and young people getting into computer graphics could still understand so and uh so we will do some some stuff with old school graphics there for sure the next years that actually doesn't really cool you've got a lot of stuff on your list for sure one thing i will say about the micro kernel is look as long as you can get it done before canoe herd's done you're doing pretty good yeah it's i mean it's also for educational reasons i

mean i wish i wouldn't have to do it i mean but in retrospect's, I said this a couple of times, like in retrospect, instead of, but I mean, you never knew, right? Like when we were like 16, there was this Linux thing. It's like, hey, let's do this Linux thing. Like, I mean, in one way, like Linux is one of the most successful operating systems, but on the other, not on the desktop, right? Or like even on smartphones versus Android thing,

not so sure. It's basically not Linux, but Google Android, right? so sure. It's basically Linux but Google Android. It's basically if Linux is there or not, it's even cursed vendor drivers. Maybe not really. In retrospect, if I had a time machine,

you never know. You don't know before a decade passes and you've done all this stuff. If I had a time machine, I probably would go back in time and tell myself, don't do this Linux decade passes and you've done all this stuff. If I had a time machine, I probably would go back in time and tell my younger self, don't do this stuff, do your microkernel. Because this is also, I mean, sure, I get it, performance, performance, performance, yes, yes, yes. The old performance argument. I mean, there are good microkernels, like QNX and stuff, right? People are always like, there are no good microkernels. Sure, there is a proprietary one. I just wish it wouldn't be proprietary. There's L4. I mean, a lot of people do stuff with

microkernels. I mean, there is a security enhanced L4 stuff, SEL4. But this is also hidden from plain sight with some, I think even one one Australian company is very big on that, right? I'm not sure if they still exist. They did a lot of this formally verified, highly secure L4 microkern stuff. But this stuff is just not so popular. Allegedly, it's running in a lot of million devices somewhere.

I've personally not seen them, but that's their marketing. But I want exactly for this kind of stuff, like driver compatibility, maintainability, and for research, even if it will not be successful just for research, I will

continue on this. I will probably do some low-level 3D graphic first, so probably more voodoo. And I'm not sure if I do the P3 first because it's just so much more complex. And the hypervisor,

like each time the hypervisor doesn't do something, it crashes so strangely. It's a little bit annoying to develop if each time something, if the hypervisor doesn't do something, it throws some GPU exception and you need to reboot. That's a little bit annoying.

At least I have not yet found how to recover every one exception without rebooting. So maybe there is a way. Probably there is a way, but I have not found it yet. But

that's the thing. I said before Linux kill OS innovation and it's like I wish instead of forking the 2000s Linux distribution, people would do more genuine research from scratch. Instead of just like... I mean, okay, to be fair, most people probably couldn't do it

because they just changed some theming and package selection and some script. But I wish... I mean, there is some, right? There's a lot of stuff like HelenOS and there is this German GeodeOS people here in Dresden. I mean, people do some fun stuff,

even with microkernel. They're both microkernel-ish. Dragonfly BSD is a little bit hybrid, microkernel-ish, they say. I've not particularly looked into that, but they say that at least. It's not like it's not existing. And I think it had,

for me, 25 years in Linux, I wish stuff would be more driver stable API, more compatible, more stable. I mean, even as Linux kernel developer, I don't like my Linux kernel crashing when I write a USB

driver and it flubbers some stuff and you need to reboot. Sure, these days you can develop in a VM, but at the end of the day if it's in production, it's a customer, it still can crash the same way out of bounds if you overlooked some condition.

I thought for decades already, like, stuff should even, I mean, there is no perfect code, right? As much as we all want perfect code, there's always even in Rust, there's always some because what people don't understand, like, Rust is only memory safe, not logic safe, right?

You can still write even if in the safest right? You can still write even if in the safest Rust code, you can still create logic errors that cause deadlocks or send the wrong packets or whatnot, or even the

hardware stops responding on the PCI or USB bus or stuff, and then your Rust code deadlocks and stuff. I mean, sure. A micro kernel cannot fully protect against this, but at least it's,

if it's 90% better, it would already be better. Right. Right. Like your keyboard can crash your kernel. So it's, it's not ideal. Yeah.

That's not, that's not the best state to be in. You know, I think at some point we need to do like a full episode just talking about the whole microkernel thing because I feel like you've probably got a lot to say on that considering that you've, you know, wanted to do this for a long time.

Yeah. We sure can. I didn't probably we're probably two hours in, right? Yeah, I don't want to do that episode right now. We're almost two hours in. It's about hour 55. No, I mean, it's been a long day. Yeah, it don't want to do that episode right now. We're almost two hours in. It's about hour 55. No, I mean, yeah. It's been, it got long today.

Yeah, it's about how long I normally am to do it. So I think it was a fun episode. I think, I don't know. I hope you enjoyed it. Yeah, definitely. I hope the viewers like it and learn something. Like I was saying, I hope you enjoyed this and learned something. And copied this from Lewis Rossman with his hardware

tinkering. So I hope viewers enjoyed this and took some new takeaways and things to think about. Certainly not low-level assembly stuff like I do at times, but broader concepts of Linux and the ecosystem. Also

that the main opinion on Linux kernel and mailing list proper is not like what all the rest of the world thinks just because some big name people dictate it like it's like everyone is happy with it and still keep Itanium around

so I guess before we end off then just let people know where they can find T2, where they can find your YouTube stuff, anything else you want to direct people to. Yeah, T2 you find at t2sde.org, and I'm otherwise active now since recently, since the last four years on YouTube, youtube.com slash Reni Rebe, or more Reni Rebe. Ironically, my MoreLife channel took recently more of than my good old content channel.

It is a time sometimes. Yeah, it would be amazing. I mean, T2 is open source and fully free, right? So if you want to download that, play around at home, run FireUp or VM. Even on obscure hardware, PowerPC and NIPs and stuff,

or latest RISC-V. And let us know if something doesn't work or send a patch. The most amazing thing, send a patch, like some typos, some package, whatever you find and want to tinker with. Awesome, awesome. And it might get fixed quicker than a year later when you eventually cycle to that

machine and realize oh wait something doesn't work if you maybe get a message about it like hmm yeah that actually doesn't work now does it yeah recently i mean we got no much better um with with all the community um the last four years we spent up modernizing all the desktop stuff and fixing obscure stuff and all the new Rust stuff and audio stuff. But T2 is in really great shape now after four years of full-time development. And we are now much quicker. Like if you report something like some obscure stuff, like we usually fix it within a day or week. Like unless it's like crazy big stuff like Chromium or so. Certainly, we have, for example, what I also

also with your views on YouTube, you directly support the development. For example, we still have to debug why Firefox is crashing on PowerPC. It's working on Little Engine, so of course, we have modern open power, little-endian power PC. There, Firefox works on T2, but on old, good old, big-endian power PC, it's segmentation faulting in strange ways.

And this stuff, like, I debugged this for eight hours already, but, like, it's like, yeah, that's what all those views support these days. Fixing all the stuff, like, even the Muscle Up Foundation wouldn't even take a look at. We didn't even, I just realized we didn't even touch on the Linux Foundation, did we? We can save that for another one.

Yeah, we can make this a day. Yeah, we save this for next quarter. But yeah, I mean, there's really, my to-do is, no kidding, my to-do is 666 lines long. There's like, yeah, we need Firefox on Big Andean, on Itanium, but that's a bigger thing. There's a lot of stuff, but now we increasingly do the fun stuff. Also,

we have not even mentioned Hewlett Packard precision architecture. We have not even mentioned Hewlett Packard precision architecture. But the future is, of course, in RISC-V, and that's also where we are at. We have to do stuff, right? We have Firefox and RISC-V already, but the next future of the NixDisc group is in RISC-V, and embedded, and so much to do. Yeah.

Is there anything else you want to direct people to? Is that everything you want to mention? No, that's basically it, I guess. Okay, awesome. Yeah, I'll do my outro then. So the main channel is Brody Robinson. I do Linux videos there six-ish days a week. They're going to be a lot less technical

than what you'll see over on his channel. So if you want to see something, you want to see a lot see over on his channel. So if you want to see something... You want to see a lot of code? Go over there. If you want to see me rant about Gnome devs doing stupid things, come to my place. The gaming channel is Brodeon Games. I stream there twice a week. Right now, I'll be playing through Sekiro.

And I'm doing a Pokemon stream with Ren. So check that out as well. And if you listen to the audio version of this, you can find the video on YouTube at Tech Over Tea. If you want to find the audio, it is on every podcast platform. Search Tech Over Tea, you'll find it. Chuck, an RSS reader, you'll

be good. So, give me your final word. What do you want to say? How do you want to end this off? Yeah, thanks for having me. It was a fun time. And for the viewers, have fun on your devices. Go fix some nice stuff or play with some Nintendo stuff, embedded DIY hobby stuff, right? And looking forward to hear from you. Pitch, email, comment, and hope you enjoyed this

and learned something. Awesome.