 Um, so yeah, hi everybody. I am Matthew Miller. I am the Fedora project leader, which is kind of a weird title for a project like this because it is not the kind of thing where I say, here's what we'll do. And then everybody does it. It's the kind of thing where I try and talk to everybody who's in this room and not in this room and everybody who works on Fedora and try to figure out what we're all collectively interested in and working on and try and sort of establish a coherence so we can all work together and better get to where we actually want to go. I've been doing these state of Fedora talks for a little while now. You've probably seen one before. I usually have some graphs and things in them. I've got graphs in this one as well. Um, I also am going to try and do something dangerous, which is make a point with some of the things I see in those graphs. So we'll, we'll see how that goes this time. Um, and does the clicker work? It does when I turn it on. Okay. If I want to laser pointer, that's the one to laser pointer. Okay. Uh, so first, um, the last couple of releases, Fedora 25, 26, have gotten really good press and a good, good uptake. A lot of people saying good things about them. So I like to take a moment to show these quotes, which I will not read because they're written down, uh, but they're nice. And I would like to congratulate everybody who worked on these releases. Um, that's basically everybody in this room, I think, uh, and lots of people again, not in this room. Uh, congratulations. Good job. These awesome releases, the last few releases. Yes. So I was also looking back at press from, you know, previous years to see like what people said, you sort of five years or so ago. And, uh, some of the things back then were so mean that I decided not to put those quotes up on the board. And instead here is a, um, pretty version picture of the eclipse from NASA public domain. So that's, um, yeah, uh, we've done really well in the recent few years, but it's not necessarily the natural state of Fedora that everything is awesome. Like it's a lot of hard work to make an awesome Fedora release. And we've done a really good job of it recently. Um, but we've had some times where Fedora has struggled a little bit more. So it's good to remember those things as well as we congratulate ourselves. Um, but back to congratulating for a little bit here, this is a mirror statistics, uh, from the most recent couple of releases, basically, um, a number of times are a number of IP addresses every day that we see connecting to our mirror servers. There's a whole bunch of caveats with this, but that's basically, uh, we feel like it's pretty good for comparing release to release, even if the, um, access numbers are maybe not tell it completely representative. Um, but we can see basically we've had pretty good growth since Fedora 20 each time, uh, maybe a little bit depth of 24 Fedora 25, very high on the charts and Fedora 26, um, just coming up. And actually this is a seven day moving average. Um, if I look at the raw numbers, you can see that Fedora 26 is actually crossed Fedora 25. So it's now our highest peak release right now. So, um, that's pretty cool too. We have had, um, very fast uptake. Um, I think it took, uh, 40 days for that to happen from the release, which is, um, a new record for people coming to the new Fedora. So, um, again, awesome. Good for us. Um, this is my dinosaur picture. Um, because these mirror statistics, I, um, promised Smooj who helps me gather these that I would, um, present them in the context of dinosaurs. Um, it's dangerous to draw too many conclusions from these because, uh, there's very sensitive to network kind of things. We don't do any invasive tracking of Fedora. There's no ID tracking. There's no per machine tracking. So we're just seeing machines hitting the mirrors and there's a lot of reasons they wouldn't show up. Um, machines behind that, if there's five machines behind one, uh, one gateway, that's counted as one. If there's a thousand machines, it's counted as one. On the other hand, if you move around to like 20 coffee shops in a day, you might be counted as 20. As long as you do an update at each one of those coffee shops, maybe you do, maybe you don't. Um, some of us definitely do. Uh, so, yeah, um, dinosaurs are involved in these statistics. So, um, there, there's that to, um, consider when taking too much from this data. Um, all right. So, things are generally pretty good in Fedora right now. Very popular. Um, releases are solid. I'm very proud of them. I'm proud of everybody. Um, meanwhile, on, on Devel List, um, we had this quote, uh, feels like everything is on fire and not in a good way. Um, you can go, go look, you said that if you want. But, um, it wasn't just one person. There's kind of a general feeling that things are kind of crazy. So, if everything is so good, uh, right now we're in the midst of having a cycle where, uh, we decided to stick to the schedule of having releases in October, even though the last release was in July, which means we've got, like, three months to put together a whole release, which seems kind of crazy. Um, it is kind of crazy. And, not only that, um, in order to make that happen, there's basically only one good way, which is to say, we're gonna defer some of the things we wanted to do to the next release. We're not gonna try and cram everything in. Um, which makes sense. Then, the things that we are trying to do on this release are big, crazy, lighting things on fire, destructive things. So, why are we doing these big, crazy things this time around, uh, rather than saying, okay, this is a short release. Now it's time to be conservative. Um, I've, uh, hopefully some of these dinosaur numbers will help, um, think about why we're doing that. Um, I strongly believe, and I think that a lot of people in this room feel like the things we're doing are important right now. And I think, um, this flock will, uh, hopefully feel like we can bring everybody who's feeling uncomfortable and developed into consensus on that, so we feel like we're doing the right thing here. Um, so, this is back to more dinosaur numbers here. And this is basically, um, from, um, all time when we have statistics here. And I've kind of lumped them into random groups here. Uh, just sort of some of the groups, the, the releases lumped together basically arbitrarily as I kind of felt they fit, fit together. Um, the red here, which is actually colored red by coincidence, but, um, is a perfectly good time because you can see that's when things started going downhill there a little bit, where we are, where, um, the numbers, you know, instead of going up like they had been suddenly or going the wrong direction. Um, and this was a time of a lot of change for Fedora. Um, the system D entered the distribution. Uh, GNOME 3 came out. There's anaconda rewrite. Um, DNF's probably in there somewhere. So these are things that were pretty scary at the time. Um, but I think that, um, as we got to the Fedora next thing, which is this green area here, those are some of the foundational things that made it possible for Fedora to be as awesome as it is today. So, um, there was that time of decrease, but, um, then we had this gigantic growth coming up. So I think that was generally good. It would be nice, um, this time when we're lighting things on fire if we can do it and keep growing at the same time. Um, but sometimes, you know, things don't naturally just always go up. Uh, and so, uh, one of the things I really want to point out, the laser pointer is supposed to go on this screen over here. Um, so does it show? Yeah. Uh, we've got this growth going up here. This little dip is kind of a weird anomaly. Um, this seems to be a lot of Fedora 21 and Fedora 22 systems that were 32-bit that, um, instead of upgrading to the next release, they just went away. We don't know why. Um, it may coincide with a lot of our chatter about, um, you know, we're not quite able to support 32-bit on Intel very much anymore. So maybe some people are like, okay, we're gonna switch to CentOS or REL or something else for those systems. Don't really know. Again, we just have observational data. Um, it could be that all those systems just finally died. And so, who knows? Uh, they're gone. Um, and then things do, you know, keep going up from there. But if we look for basic, oh, that's going ahead, uh, for basically this last year here, it's actually pretty flat. It's neither growth nor decline. And, um, that's kind of concerning to me because this, um, we should be seeing growth. And you can see in the last bit here, um, which is basically since the Fedora 26 release, there's higher numbers. But you can also see on the very end of the chart that's actually kind of back down that higher numbers might just be noise. And this is one of, uh, can be one of the dinosaur artifacts, which is if you have a network which has Fedora 24, uh, Fedora 25 and Fedora 26 machines on it, if it's all Fedora 25 machines, that gets counted one time because of the, the NAT problem. If you suddenly have a Fedora 25, uh, and 26 machine on there, now it's counted twice. So that can cause the data to be higher than it is. So I hope this is not true. I hope it's going to keep going up. I hope, you know, maybe at least it's a new level. Um, but I would not be surprised to see this come back down to be flat for a while there. So, um, that's a little bit concerning, because I don't want our growth to slow down. Uh, and so here is another, another look at this data, um, sort of a way to corroborate, uh, what's going on with that. Uh, this is looking at a completely different thing that we have some metrics on, uh, which is hotspot check-ins, which basically, um, my laptop, a lot of your laptops, uh, every five minutes do a little check to make sure that they're not behind a captive portal where you're at the coffee shop or the hotel and it makes you log in. So it's the thing that makes it pop up, that convenient thing that says, you're not on the real internet, you need to log in. Um, this is a pretty common thing that basically every operating system does this, what, does these days. ARS does this by hitting a specific URL on the Fedora servers. So we're just counting the number of times people, uh, hit that server there. Um, and again, you know, it's not invasive tracking, it's just we can see that there are machines hitting this, uh, this URL. Uh, but unlike the other one where it's once per day, this is every five minutes. So we're actually counting the average number, average per day of five minutes here. So you can see it's about 90,000 systems every, uh, every day. And so we can say, you know, like the five, the average five minutes. So if we could basically say, let's say, um, laptops are on a third of the day, eight hours a day. Um, we can maybe assume that the actual total number is something more like 300,000, something like that. Um, which is okay. Um, and not every machine is counted. Obviously servers don't do this. Um, both, uh, GNOME and KDE use this feature. I don't think any of our other desktop environments do. Um, some people, of course, may have turned it off or changed it to hit another server. So I wouldn't say that 300,000 is the total number of Fedora machines out there. Um, on the other hand, it's probably somewhere in that ballpark, maybe half a million Fedora systems. Uh, which is really, I think, a lower number than we'd like to have in this world. Um, there is a Stack Exchange has a survey of developers. Um, Stack Exchange is a gigantic, uh, website that's very, very popular among software developers, and they did a survey of, um, like 50,000 or 60,000 people every year, which is, um, you know, it's a web-based survey, not necessarily scientific. But anyways, uh, about 20% of those, uh, what developers are using Linux as their main desktop environment. Um, if we say that means 20% of developers in the world are using Linux, um, that should be a lot more than half a million people in software developers alone, not, uh, not even worrying about all the other people who are not software developers who are using Linux on their desktop. So, um, basically, I think that, although we're doing fairly well, and, uh, we've reached a plateau, and we really should be growing a lot more in order to have the impact that we'd like to have in the world. So, um, that's kind of why I'm talking about lighting things on fire a little bit, because fire is what propels growth. Um, so, switching graphs a little bit. How many people have seen a curve like this before? This way, over this way. I'm not on the screen. Okay, yeah, good. About half the room maybe. This is the diffusion of innovation curve, um, and this particularly, uh, has a chasm in it. The first time I saw this was actually at FUDCON 2005, or Michael Thiemann stood up in front of us at Boston and talked about this. Basically, this is the idea of, when you have got a new idea, how it reaches the marketplace. And so we've got basically about, two and a half percent of the market are over in this innovators area, and maybe, uh, 12 percent total in the innovators and early adopters section here. And then the late majority, like that's when people are like, okay, this Linux stuff is really catching on, um, and then eventually, you know, down there. So, um, this stuff, the early majority, late majority, and all that, like, this is what enterprise distributions are for. People who are in that area need long-term support and not just long-term maintenance, something like, um, distributions like, uh, that, that, um, have a lifetime for three years, but don't necessarily give you full support, don't necessarily, uh, cover what people need. So this is something that, Fedora, as a community, like that, that big area is probably off limits to us because we don't have the resources to cross that chasm. That doesn't mean that we don't want to, um, succeed, but, um, as part of our charter of fast-moving and innovation, basically, uh, we live in this space here. This, uh, innovator is an early adopter's space. And one of the things I really like to push, because it's something I've heard from a lot of people here in this community, is that Fedora isn't just the bleeding edge. Like, we often hear bleeding edge and I kind of would like to cross that off of things we say about Fedora. The bleeding edge is right up here and it's only a few people and things don't work very well. Fedora really wants, we want to cover some of that, but we don't want to be in this early space where there are people who are actually using it because we want to have an impact on real people's lives and we want to make sure that technology gets tested and used and actually is doing things for people. It isn't just the bleeding. We want to be in the leading area, but not completely bleeding. And we also would like to, you know, go as far that way into making, uh, a larger market as we possibly can, given the limitations of basically being fast moving and having limited resources in doing what we are. So the thing about this is that this leading edge doesn't just sit there. We can't just say, okay, we've invented Fedora and now 14 years later, we're doing the same thing. There are a lot of pressures on this from both sides. Basically pushing in from, from that bleeding edge, we have things like CoreOS doing containerized operating systems. You even crazier things like RancherOS, which runs the whole operating system in a container. We have, you know, things like Raspbian that are basically, all the IoT innovations happening on a different platform from Fedora. On the desktop, there's Solis, which is a very popular enthusiast desktop. There's Arch, which really does try to stay on the bleeding edge. Arch people are awesome. Their documentation is wonderful. There's a lot of change going on in how desktops are put together right now. We've got FlatPak and Snappy and FlatHub and all those things. And those are pushing the innovation curve on things that we need to keep up with. But that's not the only direction that that change is coming from. So coming in sort of from the early adopter side and those side, traditionally, we've been able in Fedora to rely on the fact that a lot of people will use Fedora because our downstreams are too slow. People who are using RHEL can't get the latest stuff. So they finally say, okay, well, I'm going to make the trade-off. I'll go bleeding edge and take Fedora stuff because I know it's new or even if it doesn't give me all the things I need, I just need that newer kernel. I need that newer GCC. I need all those things. But Red Hat has tried to solve that problem for their customers because they need that with things like software collections which are basically meant to bring newer things into Fedora, into RHEL. So we don't necessarily get for free just because we have newer things, people consider us to be innovative. So we need to do something a little different than just say, okay, we package stuff faster. And again, along those lines in RHEL 7.4, that has an updated version of GNOME. So a lot of people are like, well, I could never run CentOS on my desktop. I'd never run RHEL on my desktop because the desktop environment is just so ancient and crafty. That's not going to necessarily be the case anymore. So we can't necessarily rely on, we just have the latest things again. So basically, it feels like we've got this big machine that is Fedora that puts things together. And we work on this crunch kind of cycle to try and put Fedora together. And we have this idea of we've got this big machine of Fedora that we're working on. And so we have to figure out how are we going to get past this flattening, how are we going to make sure that Fedora stays relevant, how we can stay in that early adopter innovation niche, and how we can scale all of that up. And because taking machines and just making the machines we have bigger and bigger is basically not going to get us there because it can scale to a certain amount, but just as we get bigger and bigger, we've got whatever 17,000 packages in Fedora and we're using infrastructure that was meant for thousands of packages and the lower ends of thousands. And so we have things where like a compose basically putting together a release that can be tested from all the packages takes 12, 14, 16 hours to do as long as everything goes right. And a lot because it's so old and meant for things that meant for a smaller set of Fedora, it often breaks. So this big machine is not really doing what we need anymore. So basically what we need to do is set things on fire in a good way. We need, this is for Steve Gallagher here, this slide. We need to make friends with the robots, not just because the graphic is cute, but we really need to find a way to automate what we're doing in Fedora and we need to make big, awesome changes that will put us into where we need to be for the next decade of Fedora. Rather than, there are some things like someone on the developer said, why are we doing all this blowing things up when we've got this basic things we need to do to make Rust packaging better. Now doing better Rust packaging is awesome, but it is just baseline in the kind of things we need to do. We need to really focus on these big changes and the Rust packaging can wait because actually the Rust people using Rust don't even care that we package things. So let's focus on stuff that has big robot size impact rather than the incremental changes because it's not going to get us very far. So some of those things, we've got a couple initiatives that have a lot of talks here at this conference. One of them is Project Atomic and specifically the Project Atomic Continuous Integration Initiative. So Continuous Integration is a specific kind of testing where instead of doing validation testing or in addition to validation testing, you actually test every change as it's made in an environment that is like the production environment. So this is something we've never really done in Fedora. We do a lot of awesome QA and a lot of testing and a lot of people here are deeply involved in that but generally that testing happens either at the like I made a package, okay, it's basic test pass, kind of unit testing kind of things or at the very end when we put all to get everything together, the big compose, the big compose comes out the end and then people pick it up and start looking at it to see if it works. So the idea with Continuous Integration is whenever we make a change to a package, we automatically get an artifact at the end that gets run through robotic automatic testing and if your change would break something, you get told about it right away and in fact, we want to get to the point where your change is actually rejected. So if you submit something that now Fedora doesn't boot anymore because of the thing you did, you can't do that thing. It will say, sorry, please fix this before it gets into Fedora. So that's kind of exciting and that kind of that level of automation behind that is what's going to let us get to the next level of things. So this is the way we produce Fedora two week atomic now kind of. This is actually the plan I put together for that from people's ideas like two or three years ago now and you don't need to pay attention to what it actually says. The key thing is this was a bunch of new components that require a lot of manual work to happen and this is actually working fairly well although a lot of it relies on people actually manually pushing buttons, moving things around, things like that that it should be more automated than it is and again, this only does validation testing. We run it through automated testing but basically when it's done to see if it's okay to release not at the development side. So here's a new slide from Steph this morning which I haven't had time to actually digest but this is what we're you should go to his talk on Thursday to learn what this actually means. This is basically where we're going with this instead of having something that is bolted on the side but actually integrated into the environment using the factory 2.0 stuff to actually and new pipeline to actually automatically test things as changes are made and I don't even know why there's a ghost on here but it's awesome. So again, more automation and producing things in a way that we haven't before that's going to let us move this faster and scale up and do a lot more things because we're working together with the robots rather than fighting them. So modularity is another big thing and this is what I think may have tipped over the edge for the everything is on fire on develop list. Modularity is basically an idea for taking I kind of like to think it as making the comps groups like super comps groups basically selections of packages groups of packages that you can install it's not just install the web server but you install the whole environment around the web server but with this we can make those artifacts so they can install on different bases and move at different speeds so we can say the web server comps group it has Apache 2.4 and that will install on whatever version of Fedora you want to run it'll install on CentOS it'll install on rel and you can have that same version that'll build on a lot of different things and this is there's a lot of different documentation and pictures on modularity I think this is one of the one of the nicest ones basically again with the automation theme you have one module and then that can be made by robots to deploy in a lot of different environments so you don't have to worry about I'm going to need to make a container version of this I need to make a flat pack version of this we're working on infrastructure to automate that part of it and also to automate the testing of it so that when you make your changes you can make your module and then people users can deploy that module in a lot of different ways so I think that's very cool also related to that this is another dinosaur graph this is something that's important to stress about the impact of Fedora so I was saying earlier something like half a million people using Fedora something like that that's not the limit or impact so this is the Fedora operating system the blue line and then Apple the extra packages for enterprise Linux which is Fedora stuff that we've taken and we've built it so that it'll run on rel and CentOS and scientific Linux and so on about five years ago that started exceeding the Fedora base operating system in popularity and then over that's has basically gone off the charts I don't know what's happening at the top we'll call those dinosaurs with that bouncing up there but basically this is gone off the charts in growth and continues to go that way so this is another area where Fedora as a project has a gigantic impact that is outsized from the people who are running Fedora as the base operating system so this is one of the things where I think modularity has a big ability to make a lot more impact because right now these packages are done by a manual process where somebody says okay I'm going to decide to make my package available for Apple and then I'm going to commit to maintaining it for 10 years and all this kind of thing and it's again very popular but it's kind of ragged and is not as great as it could be I think with modularity we have the ability to make packages available to this market that really really wants them in a way that's a lot less work for us we can offer them on different life cycles we can say yeah you can run this on Apple but I'm not promising to maintain this for 10 years because that's crazy I'm a volunteer I'm not going to maintain the same thing for 10 years but I can keep this up as I'm doing Fedora updates and you can keep running it on your CentOS base so you can have the thing you want to move fast moving at the speed you want to move it at even if you have a stable base and so that's a way that we as Fedora have a big impact beyond just the Fedora base operating system and it's also really worth noting that it's not just these extra packages that are impact from Fedora but actually CentOS itself and RHEL itself comes from Fedora and comes from our work so even if we have a small number of individual users our impact is outsized to that so that is also something good and something to be proud of but I still think that we also want to keep growing the base Fedora as it is in order to really fulfill what our mission is switching tack a little bit this is a blog post I wrote a couple weeks ago or a week ago didn't anybody read this blog post? All right like four people read this blog post thank you I would like more people to read this blog post so this isn't about making the operating system itself this is about what we do as a community to bring the operating system to users so we have an organization as part of Fedora called the Fedora Ambassadors how many people here are Fedora Ambassadors? All right a lot more people than read this post please go I posted it to the Fedora Ambassadors mailing list and it is on the community blog basically this is a proposal for what we should do with basically our strategy for spending Fedora money and our effort in the ambassadors to basically show Fedora to the world and traditionally ambassadors have done a lot of things like going to Linux cons and Linux shows and setting up tables and talking to people about Fedora which is good and going to Linux users groups and those kind of things but the impact of that kind of thing is not really gigantic because people at in a Linux con basically know what Fedora is and a lot of people if you go to a local Linux fest like you will talk to people who are like I'm a Linux mint for life person and the conversation with them like maybe you could persuade them to choose Fedora but it's not really having a gigantic impact on things so we've been saying for a while at the Fedora Council and leadership levels we'd like the ambassadors to focus on other things that can maybe bring in new users and focus on things that are going to kind of make a difference for the future of Fedora and the ambassadors have come back to us and said okay that sounds good but we don't know what specifically that means so I have I called it a modest proposal but Brian told me I should not refer to eating babies as my suggestion so I have a regular proposal that for the basically the future when we want to spend ambassador money on events we should pick things that are directly connected to something that's called a Fedora objective so right now a modularity and this atomic CI thing are two of our objectives this is part of our Fedora leadership structure in the Fedora Council that's why I've got the round table picture here the Fedora Council has a mix of people like me who are hired to work full time on it and people who are appointed by different committees and people who are elected and in addition to that we have these objective leads who are people who have we select something that is proposed to us as important modularity and the atomic CI came to this way and we said okay this is something that is important the future of Fedora and that person gets a seat on the Council and has we can basically send the message this is something that came actually from Flock when we were in Prague there was kind of a yeah the Fedora Board says we want to do things but nobody kind of feels like that's an official statement so we wanted to make sure that like when we have something that we select as an objective we empower the people who are working on it to be at the highest you know levels of Fedora leadership and can kind of have a voice for that kind of thing so we've got these objectives so my proposal coming back around to what I was talking about five minutes ago my proposal is we should do things with ambassadors that directly support those objectives so when we have something you know that is automation related we should go to automation conferences we should talk to those kind of things we should talk about when we have mod we have modularity as an objective and we want to you know sell that to people who are using slow moving distributions we should go to where those people are and talk to them and then also as part of this we have the Fedora Editions we have a workstation which is aimed at software developers we've got Fedora server we've got Fedora Atomic this container thing we should go to the audiences each of those has an identified specific audience and some user personas that are identified we should go to the conferences where those people are we should go to developer conferences rather than Linux conferences and talk to people about why they should use Fedora so I think that we should focus ambassador money on those kind of things I would really encourage you to read this post because I put a lot of time into writing it we're going to have a conversation this afternoon about the budget I will be there and I can basically rehash everything I said out loud for people who don't have the attention span for reading which is fine sometimes I'm that way as well and yeah so I think this is going to be an important thing for basically Fedora spending in the future now I've been talking a lot and need a glass of water here's my basic recap we've had a lot of growth but now things are flattening off we really need to live in that innovation space and we need to do the things that make sure that we live there even if it feels like it's uncomfortable like we things are going well why are we lighting on fire when things are going well and then you know that's the time we should light a fire because we need to keep living there if we start feeling like oh this is a comfortable easy thing to be doing we're not really doing what we we set out to do properly so yeah let's light something with fire all right I showed dinosaurs for the mirror stats I have a piece of Swiss cheese here because my next stats are about Fedora contributors and I've gotten these stats from Fed message Fed's message is a data bus that basically when you do certain activities in Fedora that message goes across this bus and robots or humans can react and do things if you've gotten Fedora badges ever most of those badges come from your activity on the Fedora bus that's how the badges system knows hey yeah this person should be aware of this little virtual bling so anyways I've mined this for some contributor stats so this covers if you've edited the wiki it's counted here if you've made a package change if you've submitted updates a feedback update in Bode that's counted here and then also if you're hanging out on IRC and give somebody a cookie that's username plus plus I count you as active here so this is how I kind of counted these are our people active in Fedora so I call this Swiss cheese because there's a lot of holes in this translations aren't counted organizing flock isn't counted all that ambassador effort other than editing wiki pages isn't really counted writing for Fedora magazine like just so much stuff is not counted in the things I'm about to show so here however is that graph so this basically is good data going back to 2015 and so there's some noise in here obviously but basically this is per week the number of people who have showed up and the blue line at the top is people who anybody who's anybody who's username I see regardless of the amount of activity that week the people in the solid colors I only count if that person has also been seen 13 other weeks throughout the year so basically if you if for the last year a quarter of the weeks you've been active I count it there so you can see there there's you know basically froth at the top of people who are dropping in and some of those people maybe actually you know every year they show up and make a big big contribution one week could be that but a lot of it is kind of drive by kind of things but anyways the red at the bottom and I didn't actually mean this to look evil but that's you know it's not it's not bad that's our old school contributors basically that's people who have been seen at that week they've been active for at least two years before that data point so you can see that we've got a pretty good consistent base of around 225 people every week who are also showing up throughout the year to work on fedora so that's pretty huge that's a large number of people and the other thing I think is nice is we can see that the intermediate and the new users the green is people who have been active they're first active just this year but again they've been active for at least 13 weeks in this year so we can see that even over the past two years we do have a pretty good influx of new contributors and those new contributors flow into being intermediate contributors and we don't have a trickling away of people it's not going down there's a lot more data analysis I could do on this and I probably will in the future like you know how where do old people go how long do they stay that kind of thing by old people I didn't mean anything Dan Wall she was offended last time I'm counting myself as an old person here too the dips here by the way are christmas vacation I think that's awesome it also starts here at the beginning as well every year red hat goes on vacation everybody else goes away some of the baseline stuff still keeps going a lot of people are still involved you know like 150 people still doing something over christmas break but there's a clear dip when it's the holidays there that I think is kind of bad so that's not a data dinosaur that's real in that job I would though like to see this graph going up as well I don't think we need exponential growth and the number of contributors we probably can't handle that having a solid base is good but I would like to see a year-over-year you know upward slant to this and this is something again that ambassadors is something I would like to help work on bringing in new contributors and all of us you know working how can we bring new people into fedora why is my next slide blank I don't know that's awesome this is actually the end of my presentation so it's okay that my slide is blank but I had another recap slide here putting everything together but I can go back to my other recap basically everything is on fire yeah that's probably how things should be I know sometimes feels like everything is on fire not in a good way we really want to make sure everything is on fire in a good way here we are at Flock this is supposed to be a do session where we try and actually hack on things this year we're really emphasized actually the actually working rather than just talks so let's find some of those things that feel like they're on fire in a bad way and figure out how we can make them be on fire in a good way because we don't want the fire to cause things to go the wrong direction we want the fire to be propelling us rocket ship robot rocket ship like into the future so let's figure out where those problems are and make sure we get those solved and let's figure out what kind of awesome automation things we can do to make next generation of Fedora be more awesome and continue growing there we are the next missing slide says does anybody have any questions so yeah yeah we went right to a hardcore technical question awesome yeah I will repeat it the question was so we have got a lot of metadata involved in updates and that is basically when every time you want to do like DNF update or go to the software center to update your system there's a bunch of stuff in the background the metadata that tell that describes all of the packages and all the app updates available and that is a gigantic blob of hundreds of megabytes of data and as Fedora grows that grows more as we add modularity modularity has its own metadata and it gets to the point where if you have a small container that's maybe you know just runs a web server or whatever your metadata for package updates is way bigger than the container itself like an order of magnitude bigger than the container and that is ridiculous that is a good question I don't know what we're doing about that that would be something to talk to about the DNF team about that because I think we need a solution a different problem okay different problem I oh in my chart okay sorry not that at all yeah so okay well let me do reset I misunderstood the question although that is something I'm concerned about as well this is basically if people are using containers now they have on their system like five different containers each of which is a separate like little kind of mini virtual machine doesn't that inflate the count the answer is since that's coming from the same IP address to the mirrors it only gets counted once just like if you do DNF update 17 times on your system it only gets counted once so that's at least smoothed out in there on the other hand maybe we should count like look at this all this container usage is going up we're not really measuring that very well smidge right there's some stuff we can do we would like to see in DNF to help us distinguish between running a container running a fedora workstation running fedora server those kind of things DNF people here will corner them and talk to them about that later other questions yeah mo what uh maybe can I go back to a certain graph this one this one yeah yeah fed up yeah so that was fedora 21 or something for 22 so somewhere in the middle of the green yeah yeah so right we we have made right we have made the upgrade process go a lot easier recently which is awesome and so that definitely contributes to more people being on the most recently releases rather than hanging out on older releases yeah right so that actually this this graph which hasn't broken down individually there yeah that kind of shows like that's why the uptake curve for each one is faster now than they used to be you might even be able to see that like I think the the blue curve here for F 21 is before that and I think F 22 is the first one with it so blue to you can see the uptake going more smoothly there okay yeah right yeah yeah so updates the faster updates are a lot because people have worked and made updates more awesome so that is good and that's so that yeah that definitely influences these how popular each new release is but again but the overall total in this graph here you know it's like this is a stacked graph so older releases and new releases all pile on there right yeah yeah the theory is that good upgrades make it easier to say okay I'll just stay the new the new fedora rather than saying oh I have to replace this why don't I replace this with new shiny distro instead of fedora so yeah I agree the upgrade work has been really valuable in making fedora more sticky other questions yeah a world map of users yeah um no we do not have a world map of users and actually that is another dinosaur in this data because this sort of data basically assumes that people are on the internet with reliable connections and checking in every day and people in a lot of places in the world cannot count on that so play parts of the world which are underrepresented in you know always on low-cost broadband internet are underrepresented in our data so there may be a lot of fedora users there that are not very well counted as well and it's sort of for that same reason using this to put together a world map of fedora it doesn't give a very representative world map I think we did a little work on that but it it didn't seem to be showing us very much so we definitely know from you know the people we see and the discussions we have that there are people using fedora all around the world but I don't have good numbers on that next smooth jib more yeah right yeah right yeah you see I made fedora eight be a separate color on this graph it's the light blue because I I still if you go search for fedora on amazon ec2 fedora eight image still comes up first so that's some some reason there's still some fedora eight hanging on but it's also also a mystery yeah so the basic question was if we are if our space is this space here is is that both toe a natural plateau and the limit of basically what we can achieve I think that the numbers we're seeing are too low for that I think that we've got a we've got a much bigger addressable market than we're we're handling basically right so one issue is we're an operating system innovator not an applications innovator the number of people who care about operating systems for their own sake you know some of the people in this room are in that set but it is a exclusive club yeah right so it does like we can't grow forever there's not exponential growth forever I do think that the numbers we're seeing right now are just are as lower than they could be right yeah so the comment is the way we do fedora with everything moving quickly with that huge metadata I talked about with a lot of high rate of change that basically excludes a lot of people who do have poor internet connectivity so that's kind of a something that we can address and I think that again with our mission refinement there are definitely room for parts of fedora like a fedora spin or a fedora a version of fedora which is more tailored at those kind of things I think there's room for that I think also some of the stuff we're doing like batching updates and things like that or going to address that a little bit more hopefully some of the modularity stuff will help Steve right yeah yeah so the question was basically is that strategic should we care I'm making that more cruel than it was but like that's basically the idea is this something we should focus on and I'm going to go back to my blog post about objectives for this because again the point of the objectives is for us collectively and on the council to talk together about things that we do think are going to advance fedora and make fedora grow and we've got room basically buy out the charter to have between two and four of those we've got two right now so there's room for more and basically if you or somebody who thinks this is an area where fedora really could grow put together that argument bring it to the council and we will talk about it and take it seriously and I know people have been kicking around the idea of university students and we should you know maybe that's a area we should grow the way to do this is to put together an objective proposal and then that we'll know that that is something that we could focus on right so what about the reverse if it's not an objective do we not do it this is a volunteer community people are going to do what they want to do and should do what they want to do and do what you're interested in if there's something you think is important awesome you should do that right but I think yeah so I I do want to take it to that step if this is something that is you know in conflict to one of our objectives if we you know if we would say we have an objective over the next 18 months to really make fedora appeal to low bandwidth users and somebody has a change which you know destroys delta rpms or something like that we would we would prioritize fixing that and say you know this is we can't we can't do that right now this is the objective we're focusing on and again this is you know by saying with the spending if it is something that we've identified as an objective we should spend money on it if it's not something we've identified as an objective when we go to do the ambassador budget and things we should say okay that sounds important but we don't have money for that right now because we're focusing on these other things so yeah it does mean if focus naturally means there are things that are outside of the focus which don't get things that they want yeah mo okay more statistics questions yeah so the question is we don't have I didn't do a breakdown of fedora additions and spins and those kind of things do I have an idea of that we do not have a super good idea of that again because the metadata like the data we get doesn't distinguish between those kind of things we have a long standing request for dnf to include some basic information like that which would make it very helpful I think that the best we have is kind of interest in the download pages the web pages like which part you know people look at the get workstation versus get server kind of thing and I from that I kind of get the sense that it's somewhere around 80% desktop and 20% server and atomic with the fedora workstation gnomes spin being the lion's share and the other spins being somewhere around 10% but I again we don't really have solid numbers on that so that's right yeah so right and that's yeah and some of the stuff like I know that a very large company is using atomic for something in production that is pretty awesome and they've got a large scale deployment of it and that doesn't show up in our numbers for atomic at all which seem to be very low so some of the stuff doesn't get counted very well so yeah I wish I knew more but I don't more questions yeah yeah yes yes there are it wasn't really a question yeah a lot of people the question is a lot of people are on old releases what should we do about that I don't know I try and gently encourage people to upgrade I think the work we put into making upgrades easy and gnomes software or fed up the dnf system upgrade really helped because it's not such a scary thing to upgrade and actually I don't have it charted here but if you look at current releases over time more and more people are on the newer releases so actually right now a majority of people are on supported releases and that wasn't the case say four years ago four years ago a majority of fedora users were on old releases which is kind of a sad state but now it's not just a majority it's a pretty solid majority but yeah it would be nice to do more about that but I don't know what more we can do other than making upgrades easier yeah right I we did there is a package which isn't isn't there anymore but we have for all called system autodaf which would disconnect your system from the network after the release was not no longer supported that actually wrote for use at universities because we had a lot of people installing systems and then never updating them and then there were security problems later I stopped maintaining that because I kept getting angry people saying I installed your package and now my system doesn't work and I said well I yeah it's probably not the best general approach anything else okay so we're going to do a little bit of a break do you want to come back up and do logistics talks stuff Brian I'm 15 minutes for everybody in this room to go to the bathroom