Good morning, everybody. Welcome to DTrace 24, the DTrace Olympiad. I feel like we need to have the procession, the torch procession to the Olympic theme.
We're very, very excited to have you here for an online conference. This is definitely a first for us. We are all online, so hopefully, if you aren't yet, hop into the Discord where you can heckle me from the Discord, or you can join me in heckling myself in the Discord. We've got a lobby in the Discord, and then we've got a talk chat, so if conversation goes on on a particular talk after that talk, we've got a channel to do that in. And then we want to still keep with kind of the unconference vibe, so if you've got something you'd like to present to us, it's not too late. Hop into the session topics, drop it in, and we'd love to hear from you.
So you are at dtrace.com. This is our accidentally quadrennial conference. So for those who haven't heard us, you described this history on the podcast on Monday, but the first dtrace.com was in 2008. Adam is featuring the shirt from exactly. Get Adam into the frame here. Adam has got the lovely shirt from our dtrace.com 2008. And I'm actually wearing, you know, everything's got meaning around here. I'm wearing the shirt from USENIX 2004.
This is a shirt that is older than all of my children. My oldest is 20. As I was leaving the house this morning, I'm like, you know, this is a shirt that's older than all of you. My 17-year-old says, why do you have a shirt that says Unisex on it? I'm like, it says USENIX, not unisex. So anyway, having to educate my own children about the history here of USENIX. But so in 2008, we had this dtrace conference. As I was trying to recall this, you know, this is kind of before social media. Like, how did we even advertise this? How did people sign up? How did people know anything about it? And as it turns out, Adam, we used this old thing called email a lot back in those days. There was like an email list of people who were interested in dtrace. So it's like exactly, and you would actually, so that was a bit of a surprise. But I kind of found my initial post to this dtrace list that was at Sun at the time, and you signed up for this conference by editing a wiki.
So very much of its time in 2008. It was an unconference, which was the style of the day. You wore an onion on your belt, and you would show up to a conference where the first thing you would do is people would say the things that they wanted to talk about, and we kind of went to the whiteboard. And so we had no agenda when people showed up. We had a couple of folks that we knew wanted to present, but really no agenda beyond that, and we put the agenda together on the fly. It was a lot of fun. It worked out great. It was a great conference, and in fact, it was so good that, you know, it was at the end of the day, and we were still having drinks and dinner, and Stephen O'Grady of RedMonk said, you know, I got to tell you, this was a really amazing conference, but next year is going to be worse. Like, you just can't recreate this next year, and the year after that will of course be worse still, and everyone's going to compare it to this year. And it was like, wow, can't even get through the conference without being bummed out about next year. So we decided we're never going to hold this thing again. We're going to be one and done. That's it. It's just going to have to live in our memory.
And then by like 2012, we realized we kind of forgotten about it, and like, you know what we should do is an Olympiad cadence. So I'm not sure if there are other, there are some other conferences that have this like very infrequent cadence. The History of Programming Languages is one of them. I think it's great. I think more conferences should do this, but it's definitely a little bit weird. Every year for dtrace.com has had something kind of like of the year. In 2012, we announced the new illumos, as was the style of the day. In 2016, dtrace.com 2016. dtrace.com 2012 was at the Children's Museum here in San Francisco, or across the bay in San Francisco. In 2016, again hosted by Joyent, and perhaps befitting the style of the day, all of the videos were on a Vimeo account that was accidentally downgraded to the free account, and the videos have apparently been lost despite my best efforts. I'm really gutted to have lost the videos from 2016, and I still haven't. I guess maybe I should say the videos haven't been found yet, but I think those hypervisors may be dead, unfortunately. They've been gone for a while. So that was in 2016. In 2020, befitting 2020, it was canceled. We just, like everything else in 2020, it was canceled. We didn't have anything. Unlike the Olympics, maybe in 2020, we didn't reschedule, and we were also obviously busy starting a new company here at Oxide. And then befitting the style of 2024, we have shamelessly cross-promoted it across all of our media channels. Like and subscribe, hit that like button. I guess we didn't do a TikTok. There are certain depths to which we will not stoop. So also, I don't know that we, I'm afraid to, with the TikTok phone. Yeah, it's not illegal now, anyway. So anyway, you're at dtrace.com 2024. Welcome, and befitting the style of 2024, we have gone strictly online. So we're really excited to have you here with us, and wanted to talk a little bit about dtrace since 2016.
So really, by any measure, dtrace is a mature technology. We, you know, I always hesitate. I never want to say dtrace is done because we don't want to think of it as done. Like the reality is that it's got like a plate tectonic-like rate of progression at this point. We just haven't had to do a lot to it. The most recent feature, I could be wrong on this, and maybe Patrick's on the stream can correct me, but Patrick did writable USDT probes in 2018. That's the last feature I can find. So we've not added a lot of features, but dtrace itself remains in heavy use. And you know, I should knock on wood here because by saying it works, I am inviting the gods to punish us with some sort of horrific dtrace bug. But it does broadly work, and there have been bugs in dtrace, not many really since 2016. Luke Ma at Oxide found this bug that I referenced here, illumos 1648. This was fun because this was a bug that has been in dtrace since the very beginning, but only a compiler change surfaced this bug. And it was fun when Luke Ma fixed that. So that was great to kind of have a bug that's been there since the beginning. We can't quite go Don Knuth and offering escalating bounties for bugs in dtrace. I would never go that far, but it does broadly work, and the rate of change has been very, very modest. Very few bugs, very few bugs found, honestly. There's also one of the things since 2016, we've done a little more reflection back on the origins. Adam, you had the idea to do a dtrace at 20 episode of Oxide and Friends, which was really great. And then perhaps riffing on that, P99 Conf asked me to do a dtrace at 21. I promise I'm not going to do dtrace at 22. We're not going to, I'm not going to do this every year. I think we're just going to treat like two decades and legal drinking age as separate milestones. And then I guess we'll do dtrace in its midlife crisis when it's 50. I don't know if I'm still ticking, but I don't want to necessarily kind of repeat some of the things we had said there. I thought those were kind of a great opportunity to reflect back. I really actually enjoyed the opportunity at P99 Conf, another online conference, to kind of look back on some of the things we had done right, some of the things that we kind of learned along the way, and then I think very, very importantly, the aspects in which we got lucky. I think people don't like to talk about when they got lucky. Everyone, you know, I'm fond of saying that success teaches you nothing, in part because people always like to factor out the luck factor. No one wants to say like, well, God, I was really goddamn lucky. We were really goddamn lucky with dtrace, just to be quick. We did a lot of things right, but we got very, very lucky. And we were kind of solving the right problem at the right time was one aspect of luck. And then honestly, another real aspect of luck, something that we really didn't control, but was very much in our favor, is the fact that dtrace was open sourced. It didn't have to be. dtrace was born in a very proprietary era of operating systems. And you know, by the time we were talking about open sourcing the operating system, I was personally very enthusiastic for it because I remember vividly thinking at the time, this technology is so useful to me that it would be really awful to not be able to carry it with me my entire career. And you know, now looking back on that, you know, two decades ago, and the fact that we have been able to carry it with us our entire career and our careers, and it's gone in so many different places. It's gone to different operating systems. It's inspired different work. And the fact that it's open source was a huge lucky break for us, an existential lucky break. So we feel very, very appreciative of that. And then of course, you said that we were in the right place at the right time with the right group of folks. So really feel fortunate about that. So given that this is kind of like a mature technology, given that it is, you know, we again, if not done, done with an asterisk, perhaps why have a conference on a mature technology? Well, one, because it's just fun to get together. It's cheap. We're doing it on the cheap, so why not? Also, I think we wanted an opportunity to make an awesome t-shirt. You saw Adam's shirt. Alan is kicking around here. Alan, do you want to come fashion model the dtrace.com shirt for 2024? So we were inspired by the shirt that we had in '08, but we took an, this is an actual giveaway. I want to actually like touch your body here. But we actually, Alan, very tempted because I want to point out that, sorry, so this is an actual gimlet sled that is, that's blown up here. And so Ben Leonard, our designer, did a terrific job. You got Oxide on the sleeve. So if you are, you're gonna, we need like a cat in here to go after the camera. Okay, now Alan's playing with the camera. So you've got your shirt. Hopefully, if you haven't already, you can go get that Oxide shirt. So I think we really want an excuse for a shirt, Adam. That was part of it. That's a big reason to have a conference on it. But I think we also just wanted to talk about how this thing is being used. And for that, I actually want to go back a little bit because one other, I wouldn't say a change since 2016, but something else that has happened since 2016. So in 2017, Bill Gurley, a venture capitalist who unfortunately passed away recently, but was very influential for us, for me personally, for my co-founder Steve Tuck, talking to me. Charles was running the Node Summit, and Charles asked me to give a talk on Node.js in 2017. Now, if you know anything about the history of Joyent with respect to Node.js, 2017 was after the relationship had become rather complicated, which is really a euphemism for saying that we basically divorced. We had decided that our future was really, we had this really kind of heart-wrenching breakup as the direction that the community wanted to take Node.js and the direction that Joyent wanted to take Node.js were really not the same direction. And Charles was like, I think it'd be great for you to give a keynote. And I'm like, no, that's a bad idea. I've got like, I got nothing nice to say. Like, that's, I just feel like that's not. He's like, well, I really actually, if you got nothing nice to say, that's an even better keynote. I'm like, okay, that's deranged. But it did actually let me actually think about what went wrong here. And I'm really grateful for the resulting talk, which is "Platform is a Reflection of Values," because it really caused me to think about like, wait, why did, because in Node.js we had well-meaning people that really did not see eye to eye. And you know, this kind of the split up was on the one hand heartbreaking, but not necessarily acrimonious. It's just that we were going in different directions. And it caused me to reflect on like, why do things go the directions they go? Like, what is, why do we pick things to work on that we work on? Why do we, why are, would we be in some communities where we agree on things and somewhere we don't? And you kind of look at these underlying values that will bind a body of software together, bind people to work together. And as part of that, I reflected back on the values of dtrace. I mean, this is, and you know, I looked at the values of a bunch of different things, but, and I kind of threw out a bunch of different values. And the thing about values is like, these are all like good things, but you're choosing among them. And you know, as I was thinking about some of the things that we made the deliberate choice on dtrace, like what did we choose? And some of the values that I highlighted for dtrace: safety, operability, expressiveness, and velocity. I kind of hate the term velocity because it makes people sound like a projectile, but velocity in terms of the ability to go from ideation to embodiment of that ideation very quickly and iterate. These are four values that I highlighted for dtrace, and these are, I think, essential, each one of these. And I think you could talk at length about each one of these values. I mean, you all know how important safety is to us and how much we've used safety as a constraint on the system. That it, you, it has to be true that you can go onto a production system and use dtrace and not be worrying about what dtrace is going to do to the system. And we've done a bunch of things for the safety of the system. Some of them were like very clear about what we need to do about safety. Some of them were more subtle. I mean, one of the things that when you, when you have a D program, you can have effectively arbitrary memory references in that program. And it's very important that if you reference memory that's not mapped, that we don't panic the system. Obviously we don't, but it's actually memory is more complicated than that because there is memory that has side effects. So you need to be sure that D is not permitted to issue a load to memory with side effects. And there are a bunch of things that we had to go do to assure that. There are things that we learned about safety that we, that we had to learn along the way. One of the things that we learned is, and you probably have heard us over the years say that the disabled probe effect of dtrace has to be zero. It's very important for dtrace to be present in a production system that, that, that the, that the disabled probe effect is zero. And the enabled probe effect, we were a little bit less concerned about. The, the enabled probe effect we knew would, uh, would scale with the kind of the scope of the question. If you enable every function entry and return in the kernel, that's going to have a much more invasive probe effect. But one of the things that we learned is that if you have once something that's really invasive and then you do something very intense on those enablings, the, the, the canonical one was a ustack. So with, you get a, a ustack allows you to get a user-level stack backtrace on kind of an arbitrary point, either in the kernel and userland. And we added the ability to have what we call ustack helpers that allow you in a dynamic system, Java was kind of the arch use case of this, but we, we did in other languages, including Node, Python, some others. Um, a ustack helper would allow, did kind of this, this, uh, not unbounded amount of work, not, it's not, uh, Turing complete, but it could be like just a lot of work. And if you do like a lot of work on something that is happening frequently, in particular, would happen out of, say, a profile provider. And if you want to do something, you know, a thousand times a second, and it's going to take you more than a millisecond of work, which was not inconceivable for some of these really invasive things, uh, you would have a system that was no longer responsive. So one of the things that dtrace does, and if you, if you've never seen this, um, then you've never known this sense of terror. Uh, if you have seen this, you know what I'm talking about. When you run dtrace and all of a sudden the system seems to have gotten very, very sad, sad, pingable maybe, but like your terminal is like, I'm not getting any response. Your terminal heart rate beginning to escalate. And after not too long, you will see a message: dtrace aborted due to systemic unresponsiveness. And dtrace imposes its own liveness criteria on the system. And if the, if the system fails to meet that liveness criteria, dtrace assumes I am to blame and I'm ripping down my enablings. Um, when that is correct, often is correct. Like dtrace, you were to blame. That it's great, that good intuition, you were to blame. Like you, like that is a bullet that just whizzed over your ear. Um, and something that we added to the system slightly later, I mean, but it was a, maybe a year or two after we initially integrated dtrace that we added that. Um, so for, and for each of these things, um, there are stories like that, um, where that, that do maybe set dtrace apart from some of the things that came certainly before it and to a certain degree after it. I mean, operability is super important to us. It's very important to us that dtrace is entirely present on the system that you need on. You don't have to install packages or download things or what have you. It's all there. Um, it's very important that you can express what we want to allow you to express your question concisely. We, you shouldn't have to ideally instrument the system in terms of its implementation details. You should be able to ask your question pretty directly. Um, and then we want to be able to do that really quickly. So it's important that you are able to ask a question, get an answer, ask a question, get an answer, ask a question, get an answer, and do that iteration very, very quickly. So each of these is essential, but there's a value that I didn't highlight there in that talk, and part because I was, you know, throwing together at the last minute, but a value that actually for dtrace is an overriding value, and it's, it's worth expanding on a little bit, and that's utility. So the, the single most important value of dtrace is its utility. So, and this is a bit of a contrast, certainly from the things that came before it. So historically, people that worked on debuggers didn't actually work on debugging things. In fact, the thing they would debug is often the debugger or the compiler. Debugger debugging groups often came out of compiler groups. And one of the frustrations with kind of the state of the art of debuggers in like the 90s is they were really designed for this kind of in situ debugging of kind of dead reproducible problems. Dead reproducible, a term that we, we agree that like I didn't invent that term, right? Right? Okay. Uh, very reproducible. Um, which like a problem you would have in a compiler, and it's not, it doesn't actually match what a complicated system that is misbehaving in situ. Breakpoint debugging is not that useful. It's, it's, it's, it's a technique that you kind of want to have, but it's like there are many other techniques that are way more important because you can't stop the system. You can't afford to stop the system. It's a production system. Um, we in dtrace, we developed the thing that we needed to debug the system. We used dtrace ourselves to actually debug the system, understand the system, and that, that is the utility of dtrace. And it's like really focused on that utility, understand the behavior of production systems. We're not actually trying to do more than that. I mean, one of the things that people looked at early on, it's like, oh, I could use dtrace, I could inst-, I could like implement hot patches in terms of dtrace. I could actually modify the operating system. And we're like, no, no, we don't want to do that because that's not, that doesn't serve our utility. Our utility is the comprehension of a production system, and everything that we do serves that, that goal. And we have then ourselves used dtrace and, and have augmented dtrace and fixed dtrace, um, but have extended it to serve that goal. And you, you've heard us use a couple mantras over the years. Uh, concise answers to arbitrary questions was kind of our tagline when we were developing it. Um, the, another thing you may have heard us say over the years is that the proof of the debugger is in the debugging. Um, it is kind of easy in a debugger to do things that seem they would be valuable, but actually aren't valuable. So the, um, you know, we've always believed that if you are extending the system because you're debugging a problem that is something that has utility, you know, you need it because you're debugging a problem that needs it. Um, in the same vein, we've always said dtrace is a workhorse, not a show horse. Um, so dtrace is, is designed to be put to work. And sometimes if you look at the arcana in dtrace, and there is some, there's some real arcana. In fact, the, the name of this very conference, dtrace.com, the conference named after a driver configuration file, and all drivers have in the operating system have configuration files. The dtrace.com file is somewhat special because we use that configuration file to specify properties of dtrace. Why would you want to do that for this somewhat arcane feature called anonymous tracing, where we can actually instrument the system during boot? Uh, and actually one, the, the things that at Oxide, one of the things actually I love that we've added to the system, Andy Fitten, who unfortunately couldn't join us today, but added to the system is the ability for on the, in the Oxide rack, we actually can pull, uh, that, that file, um, from actually over call IPCC from our service processor. So from our control plane, we can, without modifying the operating system, change what we're tracing boot, which is really pretty neat. Um, but, but arcane, and it's like, you don't frequently need to instrument during boot. That is not something that we do frequently. When you need it, you really, really need it. So if you've used dtrace a lot, you've been around people use dtrace a lot, one of the things you'll hear people say is like, there are these features that when you don't need, you don't care about, but when you need it, you really, really need it. Only the, and that's true of anonymous tracing. It's true, I mean, but true of aggregations and thread-local variables and clause-local variables. And, but it's, it's true things like speculative tracing. You may look at spec, me like, I like, why is that important? It's like, well, we've used it to solve like real problems. That's why it's there. It's there because it has been used to solve real problems. Uh, so the, and when you see that arcana, you know, for effectively all of it, there is a story behind it, a bug behind it, a time that, that we, we needed to extend the system in this way. And because utility is so important to dtrace, part of the reason to get together, even when the system is so mature, is to talk about some of that utility, talk about the way people are actually using dtrace. So a bunch of the talks you're going to hear today are from people that are just using dtrace and showing you how they're using dtrace, maybe some of the patterns they're using. But we are using it a lot here at Oxide. Um, and, uh, so we got a couple different folks who, and I, we were talking to the com-, like, God, I don't know if I had much to say. Like, I, I use dtrace all the time, but like, I'm not sure got much to say. It's like, actually, that use is something to say, and that's, that's interesting. And we, you're going to see a couple of folks today. Um, and, uh, we're, so we're excited to show you that. So that's like a reason to get together is kind of talk about that utility. Um, and we love seeing that. We love when people are actually using dtrace. If you're new to dtrace, you're kind of looking from the outside in, we really encourage you to get in and actually use it. Um, we're going to have a talk today, um, from Dave Chico on how do I learn how to use dtrace because one of the things that Dave has done here at Oxide is introduced a bunch of new people to dtrace. So, um, you're going to hear, um, I would really encourage you to jump in and go do that. Um, and Adam has got some great resources he's put together that, that Dave's going to point you to from a long time ago now, but that, that things are, that are still very much, uh, germane. Um, another reason to get together is because the dtrace values are not actually unique to dtrace, of course. Uh, they, we, we share values with other systems, with technologists in other domains. Uh, and just as frankly, dtrace was very inspired by systems that preceded it. I mean, the, um, and you, you see that inspiration a bunch of different places. I mean, we've been very upfront about the inspiration we drew from AWK. Um, if you look at that kind of that velocity and expressiveness, um, it looks and feels like AWK for, for a reason. We did learn when we were, uh, going in front of the architectural review board that at least at the time, I'm not sure this is still true, that compiler people, um, actually are not so into AWK. Um, and the, I, I remember we were in front of our ARB review and one of the reviewers said, I don't know, this just looks a lot like AWK to me. And we're like, highest praise. Thank you very much. Highest praise. Unworthy of that praise. Thank you. And then you realize like, oh, that was actually, sorry, that was an insult. I didn't really s- an insult. Um, we love AWK. Um, we love AWK. Um, we, we drew inspiration from AWK. We drew inspiration from other systems as well, certainly, um, from LISP, a bunch of other systems that, that, that we drew inspiration from. And then dtrace has inspired other systems. Um, and you're going to see some talks today on some other systems that dtrace has inspired. Um, you're going to see some talks about, and we've got, we've got Chris here with us in the studio. We're going to get a, a, a great talk from some of the work that Chris has done on the dtrace support to Linux and in particular using eBPF as a backend for it. I think I'm really excited to see that. Um, and you know, how, how can we kind of take these values to another system? How can we take these values upstack? And whom's got a talk that is, that I don't think is actually directly dtrace involved at all, but is a lot of those same, same values about the comprehensibility of a production system. Very, um, excited to see that. So with you're going to see some talks from some other bits of the stack that have those shared dtrace values. So we're, we're excited for those as well. Um, I think might get a talk from Ryan on some of the stuff that, that, that, that, that he's done, is doing. Um, really, really exciting stuff there. Um, so in terms of dtrace in 2024, um, you know, I said dtrace hasn't itself hasn't changed much, and that's true. I mean, the, the core of dtrace has not changed much. The world, however, has changed a lot. I mean, one thing is interesting is to think about, you know, from 2008, 2008, 2012, 2016, it's like the world has changed a lot, a lot. And you know, you think about 2008, going back to the, the video in 2008 where I was thanking Ben Rockwood for videotaping us. And I'm like, God, when is the last time that I said videotaping in earnest? Like, I hope it was that. I hope it was that day. Um, I got a hunch I was, you know, using that all the way through 19 dickity two. But the, uh, you know, you think it's, it's so long ago that we didn't, you know, this is like the, we, we barely have photos of dtrace.com in 2008. But it so, and even since 2016, so 2016, eight years ago now, um, you know, not much has changed in dtrace. A lot has changed in the world, and in particular the rise of Rust. Um, and you know, those of you who follow Oxide and indeed in the name, you know, it's a, it's a tip of the hat to Rust. Rust has been essential for us, and, uh, Rust and dtrace share many values. Uh, there's really a, a shared worldview around first principles thinking and rigor, comprehensibility, understanding the system. And it's really unsurprising that these two demographics have kind of found one another. Um, and you know, I think if you look at folks, certainly at Oxide, there are people certainly at Oxide to whom Rust and dtrace were equally new. There were also plenty of people who were kind of coming from that C world, knew dtrace well, and were learning Rust. And then there are plenty of people who were coming from the Rust world and learning dtrace. And it's been fun to kind of bring these two demographics together because really they are turning into kind of, there is a single demographic around shared values. Um, and the, the most, I think the most important work dtrace-related work that has happened since 2016 is the work on the USDT probes in Rust. So, um, the Ben, and, and Adam did, uh, absolutely terrific work on, on the, the, the USDT crate. We're going to get a, a talk from, from you two early on in the day, and I think it's important because that foundation has been, I mean, absolutely essential. I mean, this, this happens, has happened a lot for dtrace with me in my career of like, oh my God, what would we have done without this feature, or, or what would we have done, and what would we have done without the USDT crate? I mean, I guess, I guess we would have, we would have done the USDT crate, which we did it. So it's like we actually know what that world was like because that's why we did it. But this is one of the very first things that, uh, we did that was done, I think, in like 2021. I mean, 20, they pretty early. And you know, I'm like, hey, I think we might need this. And it's like, of course, absolutely. Um, but like a lot of things, dtrace, holy, we needed. I mean, this thing has bailed us out so many times, and it's been very, very important for us to understand the system. Um, to the point, I mean, just like to keep it very current, uh, we had an issue just this week, um, where we're trying to understand like the, what, why we had a, a transaction that was much larger than we expected in the database. And one of the things we need, it's like we, we needed some additional probes to be able to understand it. It was like a top priority to get that done, um, because we desperately needed dtrace. Uh, we use it all the time to understand our system that we've built at Oxide. Um, so for us, dtrace and Rust, um, are, um, I would say two great tastes that taste great together. Um, the, but I understand that that is now an extremely dated reference, but that is a, that, that is about, uh, chocolate and peanut butter coming together. For you millennials, this is chocolate falling in the peanut butter or peanut butter falling in the chocolate and vice versa. I think I need to stop making this reference because I explain it every time and I make it sound like even weirder. Like Gen X's are like, why are you bothering to explain? And millennials are like, what are you talking about? Um, just like so deep, so deep, so I just can't get it out. I'm so sorry. So anyway, here you are. It's the, it's the Gen X story time hour once again. Um, so this is kind of where with the, the state of where we are in, in, in 2024. And of course you're at dtrace.com. Uh, we've gone completely online this year. Um, we, one of the things we definitely appreciated is that with people much more dispersed, with the cost of travel high, with our conference admittedly like little, you, a little off, a little off piece, I would say boutique, boutique. Um, the, uh, we, we recognize this is a, this is a small conference. This is a one might say weird, but, um, a little offbeat, a little boutique, little off piece. Um, that it's not something it's necessarily easy to justify travel for. Um, so part of that zero cost, and really we wanted to make it accessible to everybody. Um, so we completely online. So really excited to experiment with this. Um, we are, I know a bunch of you already in the Discord. If you're not, hop in there. Um, where you can, uh, the, the Discord has been, um, you know, as, uh, presumably many folks know, but Adam and I co-host a weekly podcast, which used to be a Twitter space. And the, uh, there's certain things that we, some things, very few things, maybe a thing or two. I'm not taking, I may miss something. I reserve the right to miss something about Twitter spaces, even though I can't think of anything right now. But the, uh, the thing that we actually love about having Discord for the podcast is that that Discord discussion ends up being a really important part of the podcast and feeding back into the, and, and, you know, we monitor that discussion and, uh, and, and feed that back into the content. So, um, really excited to see your contributions today in the Discord. Hop in there. Um, the, I, I would also say that kind of keeping with that unconference vibe, it's not too late. We haven't, we, you know, we, if you got like, God, I got something I would love to present on today. Um, hop into the session topics. Um, so we can know who you are and kind of get you, get you slated in. Um, we don't have an actual schedule, but we do have an ordering. So we want to, and we, because we have some people with time constraints, we want to make sure we get them, um, and able to either go to bed or to reasonable hour for those who have joined us from Europe, um, or be able to go on to the next commitment. So, um, we, we do have an ordering. Um, so hop into session topics if you've got something that you'd like to present on or say. We want to, we want to keep it light. Um, you know, we, as I was telling folks inside of Oxide, we want to keep a, we call a demo Friday vibe. So every Friday we do demos, very unstructured. People can demo something very small. That's okay. You got some, like, it's only be like five minutes. Hop on there. Um, we, we would love to, love to hear from you. So hop into session top there. Um, again, hop into the, the particular, um, topic, uh, that for, for each talk we will be taking breaks. The day when we take a break, we'll, we'll hop up with a, um, the, a placard to let you know when we'll be back. So, um, so you know, so you can also be able to step away. Obviously, everything is recorded. All the recordings will be made available. We're not going to lose in this time. I'm not going to lose in this time. We're not definitely not putting him up on Vimeo. So we like to forget Vimeo. Uh, they'll be up on, uh, on YouTube. Um, and we are, uh, on the one hand, it's been lighter lift to do an online conference. On the other hand, it's definitely not zero lift, and there's a lot that needed to be done. Um, we are hugely grateful for, to Kevin Spring on the Oxide team, um, for all the logistics that, that Kevin has done in terms of, of set this thing up. Um, please, nobody tell Google that they have accidentally created a terrific integration across their properties between Google Meet and YouTube. I know if you're the Google employee who did this, I know you did this against like your management told like this is not your OKR. Why are you doing this? Like, I think this might be valuable for, for like people who use our products. Like, what? No, no, that's not what we optimize for around here. Uh, anyway, if you are the Google engineer who did this, thank you very much. Um, if you are at Google and were previously unaware of it, please don't make it your OKR to kill it. Um, because we are loving the fact that we can have a Google Meet here that we can really easily stream out over YouTube. So, um, really, uh, terrific stuff. Thank you very much for Google and thank you for Kevin for getting all that, that set up. Um, we're very grateful. Ben Leonard, um, our designer, uh, who has designed everything Oxide. When you're looking at anything Oxide, you're like, damn, that looks good. That looks good because of Ben Leonard. So we are very, very grateful for Ben, um, and his T-shirt design. Um, and I've been using the first person plural because this has been such a, a team effort between Adam and me. But you, Adam, thank you very much for being my partner along the way. Our, uh, we've had, um, and being on the podcast and, uh, Adam has been really, uh, encouraging of us to like, let's do this. Let's, let's get everyone back together. And we, we made our pledge at the D of 20 episode that we, we pledged it would be a D24. Here we are. Um, and, uh, and then for those of you who do happen to be in the neighborhood, we'll have a social, a social event, um, after the conference tonight. Um, um, and so if you can join us for that, that's great. You go get the tickets online for that. Um, so, and then finally, thanks to all of you. Thanks for joining us. I'm really, really excited to have you here. Um, really excited for the, uh, the ongoing use of dtrace for kind of expanding the community. Um, and excited for all the, the, the new folks that are, that, that are using it. Uh, and with that, uh, let's kick off dtrace.com 2024.
