Flock 2026 Lightning Talks

No, I thought do you not have food to throw at me? >> [laughter] >> Yes. >> [laughter] >> So Adam, you literally can just pull me off. Hello, testing. >> [laughter] >> It's very tense in here now with the time. >> I say I I don't I'm very bad at doing talks in time, so this is going to be this is going to be fun. Okay. >> All right, go. >> Woo. All right. Hey everybody, Mark Pearson from Lenovo in Canada and I basically do the Linux enablement program on the Lenovo laptops. So this is to give you a very quick update on where we are with the program. So apologies, a super busy chart, but I wanted to put everything on. We now do over 60 platforms with Linux enablement, which is kind of a lot of work. Um color coding, so if it's in orange or red, that means it gets a Fedora preload. The black ones are Ubuntu, which is the base for our thing. Um blue ones are new this year, so platforms that got added to our program. Um you're going to cover up the really important So uh just thought the bits that would be interesting. We've got a couple of the red ones, those are Fedora exclusive. They will be only shipping with Fedora preloads on. As you can tell, we basically got a couple of platforms where we're doing two different CPUs. So we do one with Ubuntu, one with Fedora. Uh there's I was really hoping to be able to talk about it. There is one that we are doing which is experimental. We're seeing if we can get Linux running on it. If it works and we can get through our QA team, we'll be doing that. So, and that will be with Fedora if it happens. So, to come. The other piece of news, which I'm really excited about, but we have a Yoga platform. So, Yoga is a computer as a consumer-level business unit. So, we have one Yoga team. The Yoga team came to me and said, "We want to do Linux." And I'm like, "Hell yeah." So, we're going to do that. I can't tell you what it is yet, but it's coming soon. So, I'm really excited about that. Already, business update. So, the good news is Linux demand is growing. Uh it's growing steadily for big reasons. Price increases, I know it sucks. The DDR, the SSD, CPU prices, they're all going up. The silver lining is that business units are looking for reasons to sell laptops cheaper, and Linux is a great option cuz you don't have Windows license. So, that's silver lining. Uh data sovereignty and geopolitics, I'm not qualified to talk about this with my Lenovo hat on, but you can all see why it's a very important to have data sovereignty. I'll leave the rest to you. Um AI is also driving a lot of Linux adoption. AI developers, I'm seeing a lot of systems that are targeted for AI development. And uh so, it's really driving uh a lot of Linux demand. And the other thing, which is where you guys come in, the quality of Linux these days is awesome. I run it. You guys run it. We know it's good. People are leaving Microsoft, coming to Linux, and asking for it, which is great. So, I think a year or 2 years ago, I promised I would try and say share sales numbers. I can't share the actual numbers, but I can give you trends. So, guys, so, Fedora is on the left, Linux, which is basically Fedora and Ubuntu, is on the right. As a rough guide, Fedora is about 10% of our sales. Um treat this as trends. Very important point, most Linux sales are not recorded. We have privacy, and Linux is great. I don't know if you bought a system and put Linux on it. I have no way of telling. Uh and basically, a lot of people buy with no OS, install Linux. A lot of enterprises will buy with Windows, and the users that need it convert. So, these numbers really as a trend it's more important than the actual numbers, but I will say we're not they're still low, but they are becoming interesting from a business perspective. I still, you know, making business cases. I apologize web availability is still bad. I think this will work. I No, it's not going to. I did have the the the X1 Carbon 14 was just released with Fedora and I had the page up, but I can't switch to it. So, there you go. I'm in 5 minutes. Contact details, look me up. I will put these slides online. Woo! Thank you, Mark. >> [applause] >> Oh. And those bears were the ones I mentioned during the candy swap. >> Okay, we need Adam Williamson. And yep, Christian Lee. Thank you. >> Hello, everybody. I'm Adam Williamson. I work on Fedora QA, Open QA, but also Fedora CI. >> And I'm working on the Fedora CI and TMT. >> Christian >> Christian Lee, everybody. Um so, really quick, some updates on cool things that have happened in um Fedora CI, Open QA this year. We have some CI tests running on ELN now. Great. Yay, ELN. Um we killed Zuul. When I say we, it's almost always Christian, but we. Zuul is gone, which is great. Um you don't have three systems reporting to your pull request anymore, just one. Um and you don't have like three separate things posting comments on your pull request anymore. Um RPM dep lint is 10 times faster than it was. It was 20 minutes, now it's 2 minutes. Yay for that. I wrote a thing called RPM dep check, which tells you if you broke anyone's dependencies, and it's accurate and it runs in 30 seconds, so there's that. And somebody asked me, I can't see him, a couple of days ago to make it so you could run open QA tests on pull requests, and now you can. Type {slash} open QA test. It works. I'm hooking things up, but it works. So, there's that coming soon. And I'm going to hand over to Christie now. >> All right. And I also need these notes. Uh so, we're starting to export some subtest results. So, the idea is for RPM inspect to have the individual inspections be exported and like get some of those. Uh we're moving some definitions around and like reusing lots of the code, which we didn't do before, surprisingly. And we're preparing to deprecate the whole Jenkins pipeline. Finally. Uh the body links should not point you to Jenkins anymore. If they do, please let me know. >> It's a bug. Tell me. >> Yeah. >> If you ever see Jenkins, we don't want that to happen. >> Yeah. Uh a few tests, like installability, RPM deblint, and RM debcheck, run once on the whole update, not for each package. Um we dropped the CIFMF. If you don't know about it, perfect. You don't need to worry. Uh we added the testing farm action for the forge, and feel free to use it. >> And one more. >> And the last one, uh experimenting with other forge workflows, for example, the docs. So. >> Do you want to mention the shared test repository? That's not on there. >> Okay. I need to explain this to you one more time. >> more thing? >> [laughter] >> Yeah, and we're making a shared test repository. So, if you have any kind of test that you want to add, just go in over there and add it. Yeah. >> So, all of this, what kind of tests are you doing? >> Well, things like RM dev check, RPM dev lin instead of having five different sets of three repos for each one, we have one repo with all of them in for the pull requests right now, not for body testing though. >> One challenge for me like if you can figure out this like uh if any seek has an idea of a test that would be interesting to figure out how to do. >> Yeah. >> Okay. >> Yeah. >> [applause] >> I hope it will work. Ah. Okay, and I know I need to find out where it is. Come on. My desktop. Okay, so hi everyone. I'm Mirko Konecny and I work on I worked a long time in the Anaconda installer uh but right now I'm working in the AI world. Which is fancy and new. So, what I want to show you is what we basically spend a lot of time a last few months here or something. And the thing first thing is I want to show you basically the benefit of having goose and the second one of uh using a Linux uh MCP uh server. So, the thing what I want to point out here, goose is in comparable level with the open code cloud code and similar similar tools, so you can use it basically in the same way. The reason the the difference is we spend a significant amount of time to get it to Fedora so you can avoid these installation methods which are the standard these days everywhere around the world. You don't want to do that. So you can really type DNF install goose and it will work. I will try to show you the demo for this and hopefully it will work. So I have here basically goose running as you can see one specific we are working we are packaging just the CLI and even that is hard project to do but it works so this is connected to our 6i whatever and the MCP Linux MCP server is there. When I run it it hopefully should investigate my virtual machine I have there. So the Linux MCP server for information is a MCP server we designed which has a read only access no write things it will not mess your server and it basically can take a look on analyze your server and trying to find the issues on the server. As you can see there's a tool calling allowance and everything and it will okay I find the failed service it's engine x service I broke in myself. Uh then it's getting service logs getting service status etc etc looking on the logs and then it should file the issue and yeah. It was a header give me a second yeah. Who done it? This is configuration file. Configuration file of the engine x is break it completely so this is my demo if you want to use it Linux MCP server is not packaged in Fedora but it's easy to install it by standard ways and honestly I'm not sure if that's even beneficial to have it packaged but we are working on that. Uh so, you can use that. Feel free to use it. >> [applause] >> How impressive is that that they're working demo in a lightning talk? That that's impressive of its own. >> Okay, then. What do we have here? HDMI >> Yes. Yes. >> Plenty of time. It's fine. Hello. Uh I'm Miro. I said this sentence like 100 times this 3 days. Um Uh I do a little bit of Python packaging in Fedora, and what we did with Carolina uh the last week or a week before that is that we rebuilt everything almost everything Python with Python 3.15, and uh normally we use a koji side tag that uh release engineering creates for us because historically doing it the self-service side tags way through Bodhi was too dangerous. So, uh we go to release engineering, ask them to create a side tag, and then we build everything in it, and then we bother Adam in the middle of the night. Uh which usually is his day that's that's cool to run open QA tests, and then we merge it, and it skips Bodhi entirely. And And that's what we do every year. And uh we try to make it as short as possible because despite our instructions for packages to not to do anything crazy, they always do. I know there are a lot of instructions, and I I I I don't blame them. I'm just mad at them internally, but yeah. Um anyway, we did something new this year and I wanted to share it. I know I don't have the time to explain all the details, so you don't know how Koji works, I'm sorry. But, if you do, it might be interesting. So, this is the F45 Python side tag that inherits from Rawhide and we built the stuff in it and that's pretty much what we did every year, but this year we also created a Fedora 45 Python bootstrap side tag that inherits from the Fedora 45 Python side tag. And uh why we are doing it because we need to build all the packages in order and their order has like million loops in it. So, that's impossible, so we need to break it. And um previously we did it by committing these stuff to this git and that's the perfect point for the packages who didn't read the instructions to go and build on top of that. So, it actually happens for us that we disabled something, for example, tests, and they updated it to a new version that obviously would fail all the tests, but they never noticed and they fast forward merged it to stable Fedora as well for good measure. So, we don't like that and for packages that have to be con that's actually called bootstrap, we just build it in the bootstrap tag and it kind of works. And what I learned after 15 years in Fedora is that there is a target in Koji and a tag in Koji and they are different things and this is actually quite useful. So, we created a target that sees the bootstrap tag, so it sets the bootstrap beacon, but it tags the build directly to our normal tag. So, we only do it the only thing that we do is if we want to build something in bootstrap, we change the target and it works transparently. The only problem is that then you need to Koji beta pop both of the tags if you want to build something. But, the worst case scenario is that it doesn't build because the dependency is not there and you try again later. And the other thing that we noticed that when we fired off the first 1,000 builds, those are usually the the packages that have no tests whatsoever, so they don't depend on anything and they always build. Most of them are no arch and then Koji does this thing that it arbitrarily selects the architecture for the no arch build and it likes to select the crazy architecture. And then it waits for builders because there are no more builders. And I'm like, you could have selected something you have. And I actually have my Koji admin privileges, so I played a little bit and I said, you know what, crazy architectures, go home. So, uh for the rebuild, all the no arch packages that we built were only built on those two architecture except for a couple at the beginning. This is not a healthy thing to do for the project long term, but if you want to build something really fast and you don't want to bother with with stuff that you don't have and nobody uses and I don't know anything about it, this is really helpful and and it works. By the way, in in Fedora proper, we also exclude some of the architectures, but not all all of them. So, for for us, basically we excluded power because the the other ones are already excluded anyway. Um I am not saying this was the reason, but it went really smooth this year. So, yeah, that's it. >> [applause] >> Hello, my name is Miro. Just joking. >> All right, so my name is Clement um and I want to make super easy for anyone to contribute to Fedora and in particular agent. So, I've started to work with the AI and ML SIG to create a library of skills for agents. So, what are those skills? Like we're just getting started. So, this is a call for participation. If you want to to participate, come and share your knowledge. I also use open code and I like to do curl bash install. So, for example, this morning I just tried to do a quick skills to look at like to make it easy to look for like Fedora composes. So, I don't know if you're like me, but I never know where to find information. There are like so many repositories. You need to talk to so many application, koji, body, and so on. I I get lost. And the good thing with this is we can actually teach the agents to have this information and the agents are are very good to go and find that information for you. So, yeah, it's like going through the logs, going to read the logs of the of today's compose and try to find like what failed into it. So, you see the all of this like you you can do it yourself if if you probably Adam Williamson, if you're not Adam, it's a bit more more tricky. Yeah. I've done it this morning. So, obviously the skills can be can be improved a lot. If it works well, we should find that some Python dependency were not rebuilt with Python >> [laughter] >> 3.15 and uh that is actually why some spins are are not are not building currently, but Yeah. But yeah, but we don't have to wait for for the end, but yeah, the the initiative is starting. So, if you if you're interested to to contribute some skills, to share your knowledge, and to work, come to the AI ML sig metric channel, and uh look for Siyana, like ping me, and I'll be more than happy to to chat with you. All right. >> [applause] >> Hi. So, I'm Owen. This is June. We work on a Fedora downstream called Ultramarine Linux, and one of the main focuses of it is hardware enablement, specifically uh the Chromebook platform. So, while I talk, we're going to try to get this live demo working. If it doesn't work, we're just going to show you guys the screen. Um basically we uh partnered with Linux on Chromebook specialists and projects in the ecosystem to make the process of running Linux on a Chromebook from installation to package support to drivers and everything as easy as possible. So, we made a bootloader. Oh, cool. Uh we made a bootloader based on u-root to load a full Linux image from stock firmware, so you don't have to flash custom firmware or anything. You just flash this little 8.5 MB bootloader. You boot right into that and then it loads full Linux. So, you don't have to risk bricking your Chromebook or anything. So, we're going to see if this works. >> So, you can see the >> You just boot to the internal disk from the Chromebook recovery screen. And then you select your kernel entry. >> For some reason this specific install seems to take a very long time to boot. I have no idea why, but usually it isn't black screen forever before it boots, but it does eventually boot, I promise. >> Yeah, while this boots I also have uh someone in the audience's Chromebook. Um this one is running Mr. Chromebox UEFI ROM coreboot firmware. Uh so, this is if you do want to actually flash custom firmware, you basically just get a full UEFI laptop and can just install a distro like normal. Uh yeah, this is not going to boot past this. Yeah, it works. >> [laughter] [applause] >> And this is Ultramarine. Oh, and an application crashed. Cool. >> [laughter] >> Um If you have any questions, uh come find us or Ellie in the audience here with this Chromebook and we'll talk about Chromebooks for hours. Thank you. >> [applause] >> Hi. So, just to get a gauge of the audience here, since I'm primarily an openSUSE guy, uh how many folks out here know what MicroOS is? Okay, fair number. >> [clears throat] >> Uh, so MicroOS is uh currently in the open SUSE ecosystem. It's a transactional system which does provide immutable features based on butter of us snapshotting. Uh, it currently uses on the open SUSE platform a number of uh bespoke tools that [clears throat] they really work quite well. I developed the Kalpa desktop on top of that which is KDE on top of MicroOS. Uh, I am going to have the tooling to provide transactional systems into Rawhide in the near future uh, using all DNF tools to do it. Uh, there won't need to be any of the custom tooling that we're using for the zipper back end on open SUSE. It will all be handled through DNF and through one small uh maintenance tools library that I'm writing for doing a couple of things that just don't belong in a package manager. Uh, this it it it gives you a nice upshot uh in between the regular Fedora point releases and the Fedora atomics because you're not having to do a compose up in the cloud anytime you want to change anything. You can all do it locally. I can just open a new butter of us a butter of us snapshot, install RPMs if I need them, reboot the machine and I've got it right there. And it's a much faster uh process than doing RPM OS tree as currently exists. Uh, I don't know how boot C's handling that because I haven't messed with it. But I just wanted to kind of get the word out there that this will be coming to Rawhide uh fairly quickly. And actually what it will do for people that like to run Rawhide on the ragged edge, it's not going to eat your face anymore because you can always have the ability to roll back to the previous snapshot when something goes wrong. So, that's all I have. >> [applause] >> Oh, yeah. Okay. Okay, let me start. Uh so, my name is Michal. Oh. My name is Michal. And as you can see, there is only one screen on this. I will be talking actually about this. So, uh this is for about the end of Open ID in Fedora. For those that just log in, nothing actually changed, but there was plenty of changes in background. Uh this login screen for some time had, if you used some service that used Open ID, had a red banner that was saying that this will go away. If you saw it and the service didn't move to Open ID Connect, it's no longer there. So, it will just throw you I think 400 or 403 or something like that. Uh so, this is actually Epsilon that is doing the authentication. And what we did to uh actually separate Open ID, we created a new instance of Epsilon that was just trying to serve the Open ID requests. So, if you get to that that one, you get the red banner. If you were using OIDC, it was okay. We helped a lot of projects to actually move to OIDC during that period. When they saw it, they actually knew something is wrong. There was a link to a ticket where this was being uh um discussed. And we tried to move them. And it end on 1st May. I think we actually moved it because the 1st May was uh holiday in Czech, so we moved it. And uh it was removed. And we didn't get any complaints, which is nice. It was It was kind of screen test because uh in case of OpenID, you don't really know who is using this uh endpoint to actually authenticate itself. In case of OIDC, we have much better overview because the projects need to register for uh to have uh Fedora account system as authentication. And yeah, I think that's that's it. We are um out of OpenID, which was not great to use, and we didn't really knew who is using that. And we now use OpenID Connect, and we are happy for it. And hopefully we will replace Epsilon in future as well. And that was all from me. >> [applause] >> Hi. I'm Miroslav, and you can call me Miro as well if you want. Um I was talking you about SPDX licenses for past 4 years it's been already. >> Sorry. >> Uh yeah. And during these 4 years we achieved many things. Like we have federal license data package and a repository where it's formally um described what license is, what is allowed, what not, under which conditions. We have a lot of amazing documentation. We get a hundred of uh new licenses to SPDX upstream. Um we converted 30,000 license tags. Uh we get CI. Now you have in pull request license validate test which tells you whether you made typo or not. Uh we have only last few packages remaining. Two packages plus Linux firmware which is full can of worms be licenses which uh the the lawyers, the Richard Fontana, and Julian Lie Joy try to avoid because it's like ugly. Uh so it will take probably months to like properly resolve. Uh and that's it. And I have to say I'm really tired of this. So so I have to say uh I'm I quit. Um so this I will probably stay in a maintain mode uh and take care about the federal license data repository and do something if someone uh ask me. But I will be not pushing any new things. So there are plans what should be the next things. We have We have s- uh scanners in Fedora, several of them. Scan code uh most the best one probably. Uh and it would be nice if uh if we have some tooling that will scan new tarballs and let uh tell you what new licenses are there so you may add there to the license tag or remove some licenses which are there not anymore. But I have to say that somebody else works. So So, it's depends on somebody of you. Uh So, if you want to uh jump into that, I will happily uh guide you, tell you uh pieces of wisdom I have, but it depends on on on you. Uh and that's all. Thank you. >> [applause] >> Hello. Amir from France. The French version of Amir. Um yeah. So, in the world there is a lot of languages, about 7,000. And I was unhappy about the situation in open source. And with the help of the Fedora community and the Hack Stube community from Strasbourg, we got some numbers and here's the situation. There are only at the at the bottom four languages in the world that will reach 100% of translation someday in the coming century. So, there is a full talk this morning that is not too long. It's 25 minutes long. That explain how we gather the data and it gives more details. But, the learning is the way we are doing organizing the translation in the open source ecosystem is not efficient for almost every single language in the world. And I assume that the desktop operating system is meant to be used by humans. So, most humans in the world are not able to use the the open source stuff that we do and they cannot be empowered by what we do. So, that's an issue. Um What if we change the way to do translation? What if we do it downstream and create tools dedicated to communities to collaborate, to agree on how do we translate each words? How do we create new new flavors of languages? I want French to be spoken um as me as I am a guy or a girl or to be inclusive or I was I want a French that do not translate all the technical words because this is something I don't like. Or I would like to have some translation in Klingon. And I would like to have five single contributors be able to create a new language in the open source ecosystem. I would like those people to be able to ship the translation to any users. And if they want to to translate in a big language or a small language, something that exists or they create, I I would love to see that possible. So, there is a lot of technical challenge that I hope the Fedora community will will help with the different ways to solve that kind of issues. And there is a lot of challenging questions about how do we bring collaboration with upstream because if we do translations in one place, we want to share it with the rest of the world because we don't want to split this the community. We want to make it easy for language community to to contribute. So, that was my 5-minutes talk. Thanks a lot for your attention and this morning you can have a a better overview. >> [applause] >> Hello everyone. I'm not a stool. I'm not on a stool today, so that's an improvement. >> I don't remember which side the HDMI thing is on. >> Ta-da. Okay, hello everyone. Um I would like to announce that I'm starting a software composition analysis tool SIG. Um it's funny that I was almost after the second Miro because he was just talking about this and pretty much the goal of this SIG is to encourage the usage of license scanners, vulnerability scanners, and other tools like that in Fedora and also package them because currently I'm the only one packaging Scan Code and I would like not to be. So, if you'd like to join me, please do that and I'd also love to continue exploring ways that we can integrate this into the build process. Um as part of the Go change, we when we switched to vendoring, one of the things we did is we added a mandatory macro that has to be run in check with like RPM check scriptlet that checks the license tag in the spec and then does a license scan and compares the expected value to the actual value so we don't have outdated or incorrect information in the license tag. And there's also ways to kind of override things or change things if the scanner is incorrect. And this is in use in all new Go packages. Um and also I've experimented with using it for Node.js stuff as well. Um yeah. Do I have more time? How much? Okay. Uh well, I guess the other thing that I wanted to talk about, also unrelated to this, is as you know, I'm the one who sends those annoying um orphan packages report emails. Um and and the current script Oh wait, no, that's the wrong thing. Um the current Here it is. The current script takes a very long time to run. Um I can't even Where does it say? Yeah, it says Wait, doesn't How long does it say here? Where is this thing in this Okay, fine. This time it takes 5 minutes, but uh when there's a lot of packages, it can take like 20 minutes. So, basically I'm working on this new tool that's kind of based on something that Adam, who's sitting right in front of me, did um to use repo closure to figure this out instead of using a 20-year-old script, which was written in Python 2 and then rewritten to Python 3, and then rewritten to DNF, and now needs to be rewritten a fifth time. And also we need to pull the Pagure API stuff. Yeah, it was also rewritten when we switched the package DB to Pagure. So, it's just been like fixed too many times and also it takes forever. So, I'm working on a new thing and with a fresh cache, it takes 18 seconds, which is pretty nice. Um so, we have this repository where I'm now working on um improving the way we actually report this information, cuz currently the email report now is pretty long. So, if you have any ideas or would like to collaborate, um come find me. And also the tool is called endorphins. Like but e n d orphans, which I thought was funny. So, thanks everyone. >> [applause] >> I'm only up here to show you my cool desktop background. Um So, hi folks. I am Angelo, uh systems engineer at Meta, and I wanted to talk to you about hardware qualification. Uh so, it's one of the things that uh we've been doing for about a year now or so as um like one of our main pain points when we qualify new platforms and we go through release qualifications and stuff like that. Um the compatibility of hardware is one of the important things, and there's always issues. Nvidia. [clears throat] And so, we developed this customizer, which is based on the Fedora everything uh netboot. We have our kickstart that configures uh the basic setup things like we care about disk encryption, how we partition the drive. Um oh, and this is a boring wiki. Obviously, that's not external yet, but I do plan to share it. Um this is a structure of the customizer. Um contains this kickstart. It will configure certain things in a way that's um fairly conventional. Uh we inject a post kickstart script, which runs some additional installs of certain applications that we will require afterwards for tests. Um we have an Nvidia driver script because why not? And um in the end, we're going to go through the user setup, create a user account, and um include a script which will run through a set of uh a suite of tests, Um stress tests, benchmarks, CPU, GPU, disk IO, and all sorts of things that will put the system through quite a few backs and forth, which is basically emulating what the user experience would be if you're a daily user, and if your system you use it for, I don't know, games or whatever, and you're going to put it through a lot of stress, and you don't want it to catch fire in your lap. So, this is the output. Let's see. Okay, this looks horrible, and this might look better. And the final output will be this HTML report file, which will give you the score and what all the tests did, and the final full log of everything that run, how long the test took, and the reason we're doing this is what we would like to make this some form of a standard, which the testing is done by, let's say, a Fedora ready program, maybe, and we would like to work with you guys and make this more of a conventional way of testing. We've worked with Lenovo as a great vendor that has qualified Fedora extensively, and their QA team tested the P8 Workstation on our behalf this time around and has dumped it, officially qualified the first desktop that had Fedora out of the box, and we would like to extend this to more hardware and more vendors, and generally make this a let's say, a concept that's a little bit more universal across Fedora, and that we know that all the nice hardware that's out there gets a stamp of approval, that you now have a device that works perfectly with Fedora, no driver issues and stuff that were tested directly by the manufacturer. And yeah, I think that was the idea. If you are interested in hardware and testing like this, let us know, reach out, and let's partner up to make this official and public. That's it. >> [applause] >> Hi, I'm Jens, and this is >> I'm Miro, in Polish it's pronounced Bishek. >> Um So, yeah, we're talking about Nix in Fedora. Um for a long time I used to maintain a a copper repo, which was quite popular. Well, it was one of many Nix repos in copper, but Um but then during the Fedora 44 cycle, um we managed to add it to um Fedora to make it an official package. So, yeah, it's more sort of stable having it officially packaged in Fedora. Um does anyone here use Nix at all? Okay, not many. Um so, yeah, for those who don't know, Nix is a is the official is the package manager used by the NixOS uh operating system. Um And also maybe you've seen that Nix is getting increasingly used in many uh upstream projects like in GitHub. Um so, yeah, I think it's good that we have it available in in Fedora. >> [snorts] >> Another attraction is that yeah, the Nix packages repository has maybe well, there are different counts, but sort of upwards of 100,000 uh packages, so there's a lot of software available. Of course, the other big attraction with Nix is the reproducibility of builds. Um Nix is quite a complicated package to to to package, partly because of the file system requirements. Um the Nix daemon requires various user groups and so on to be set up. Um, and some systemd configuration. Also, it has some slightly exotic bleeding-edge dependencies and so on. So, it's has been a challenge. Um, also the package tries to support the different modes of Nix, both the daemon and the single-user mode and rootless and so on. So, there are different subpackages you can install depending on where you are. And yeah, so the menu who helped a lot also both to review the package and also most importantly to get I never thought this would be possible, but yeah, to get the slash Nix uh exception approved uh in in by FESCO. So, the Nix package is allowed to use this top-level uh directory. Um, and yeah, so now it's available in all current releases and also Apple 10. So, yeah. Um, maybe you want to do the demo? >> A demo? But I uh so uh yeah, um so basically yes, it is packaged. Um, I mean, I already downloaded it. Uh so, the the the the thing that Jan matter mentioned is that fact that we have the this Nix thing. Uh Uh well, maybe like this. Nix is a a top-level directory here. And the thing is that uh people didn't want this in Fedora because it's in the root directory and they need people will put it there by mistake. Uh and the the mistake is easy to fix. You just have to modify the 100,000 packages to refer to a different directory. And they didn't want to do it. So, that's So, this was a bit of contention, but we well, now we have we FPC approved the exception to have this slash Nix thing in root. >> [snorts] >> Uh so, when we have that installed, uh uh we need to do sudo systemctl enable now. Uh is it like this? No. Nix daemon. Uh okay, it was enabled. Uh this is needs to be done once. Uh and then we can use Nix. So, uh Nix uh shell like I had a I had a cheat sheet somewhere here what I wanted to do. Uh okay, so Nix shell Nix packages uh lolcat and maybe cowsay. Uh and so, the thing is that this uh mmh uses a package that is defined defined in the Nix packages um repository and every time this is run anywhere on Earth, in principle, the same things are downloaded and executed. Uh so, um so, this started a new shell. I mean, the prompt is the same, so this is hard to see, but uh I have lolcat here. I have So, cowsay Hello, Fedora. Uh some errors from Pearl and it works. So, I I I will press control D now to exit and now I'm back in my original shell. I did not have lolcat here. Uh how many seconds do we have left? >> Four. >> Uh thank you very much. >> [applause] >> Oh, no. >> It seems to work now. Yeah, so hi. I'm Artur. Yeah, so I'm very happy to be here at this event and since I like going to free software events, I thought that maybe I will invite you to an event that we've been organizing. Which is Pivo, the Poznań Free Software Fest. Yeah, so Pivo is a free as in beer event focused on free as in freedom software located in the Poznań, Poland. Just a tiny bit of history. So, this is so this event actually pay dates all the way back to 2004. It was organized by various different students associations at the Poznań University of Technology. It was basically like passed down through a couple generations of students. And well, in 2019, the event didn't happen. So, they figured, oh, we're going to do it in 2020. And then COVID happened and the the event basically died. Yeah, and in 2024, a hacker space was founded in Poznań, which is called Kniffte. And if you're wondering about the name, it's actually a portmanteau of kniff, which is Poznań slang for a clever solution, so basically a hack. And figatel, which is once again Poznań slang for neighborhood, so basically a space. So, yeah, it's basically a hacker space hacker space. And yeah, and since many of us had fun memories of Pivo, we decided that yeah, like well, why not try to reactivate the event. And yeah, so we did that in 2025. We did have some major road bumps including the venue telling us no 2 weeks before the the date, so we had to scramble to find something new. Uh but uh yeah, we managed to get almost 200 attendees, so I think that was a success. Uh >> [applause] >> Yeah, so for 2026, we wanted to go bigger and we knew we'd need help. So, we reached out to the uh universities in Poznań and we actually uh managed to get two students asso- uh students associations on board. So, that was uh WeBrains at the Adam Mickiewicz University and Linux Academic Group at the Poznań University of Technology. And uh yeah, so like preparations for the event started in October 2025 with some loose meetings. In January, we started having regular biweekly meetings. Those switched to weekly in mid in mid-March. And the event was actually a little bit over 2 weeks ago in on May 20 on May 30th at the uh Adam Mickiewicz University. The schedule was quite busy. We had three lecture tracks with 25 talks, two workshops uh tracks with eight sessions, a LAN party track with tournaments so we could win prizes, and also a quiz and lightning talk session at the end. Uh during the event, Fedora Project was represented by Zbigniew who gave two talks, uh Matt Holmes who served as a photographer, Kacper Skrzynski who's unfortunately not here in the photo, but he's there in the audience. Please raise your hand. Uh and me who well, I guess I looked nice and smelled nice. Uh yeah, the event was also granted some budget and a nice shipment of swag from the Mindshare Committee. So, thank you very much for that. Uh yeah, it was a huge success. We had 452 attendees, so way more than we expected. Uh as part of the event tradition, we have a lunch break with uh free pizza. So, this year we ordered some 62 square meters or 710 square feet of pizza and well, it didn't last long. Uh Yeah, so the event was also like a a historical moment because uh it served as an opportunity to announce the creation of this alliance of friends of open and free software, which is basically an organization that tries to bring together various NGOs as and well as well as commercial companies so we can petition the government more efficiently to use free software instead of proprietary solutions. Uh all right, what's next? So uh Yes, so the plans for the next year are in very early phases, but these are some possible dates. Uh yeah, if you'd like to learn more about the event, then here are some links. Uh we would very much like for the event to grow like both in terms of having more attendees, but also in terms of like making it a uh a brand that's that's also recognized abroad. So uh yeah. Uh go to our website, uh stay up to date, uh please submit uh proposals when the CFP opens, and uh if you have someone who has too much money and would like would like to sponsor us, that would be great. Thank you. >> [applause]