Vittorio Romeo 0:00

It's my first time on the podcast, I want to say. I was expecting a C podcast, but it's so much more than that.

Conor 0:06

So it's when I think of C shops, I think of Google, I think of Microsoft, I think of Adobe, I think of Bloomberg.

Vittorio Romeo 0:12

Some companies have been divesting, they've been trying to maybe move to Rust or have their own competitor like Carbon.

JF Bastien 0:17

Yeah, Transmeta, it was it was the hot startup in the 90s. It was a company that made an X86 chip that competed with uh Intel and AMD and Via. Mike's keynote was eye-opening for the people, me included.

Bryce 0:30

That is up there with Sean Parents, C now, keynote from like 2012 or 2013.

Vittorio Romeo 0:36

A normal struct and automatically converting it to an SOA layout using reflection even in Cells Plus 17. The voice in my head is like Yes, the message of course is be angry. That's that's what we're saying. And also, you said you're Canadian knives, or you've been very mean to me today. I I will not forget that. Please use a monospace font.

Conor 0:57

Hi, my name's Jeff.

JF Bastien 1:00

What do you want to know?

Conor 1:11

Welcome to ADSP the podcast, episode 259, recorded on September 24th, 2025. My name is Connor, and today with my co-host Bryce, we interview live from NDC Tech Town 2025 in Kongsburg, Norway, both Vittorio Romeo and JF Bastion. We talk about their respective talks, C training, and more. Alright, folks, we're here. It's technically day three for Bryce and I of NDC Tech Town, but it is day one for most folks because the conference has started today. We have finished our CUDA trainings, which you heard last episode a week ago, as we chatted about that in the lobby of the hotel. And now we are here at the tail end of day one. We've had our first day of talks. Bryce is gonna be interviewing Vittorio Romeo, folks. You've heard about him on the podcast, we've never interviewed him. He is the author and uh behind many projects. The one that I bring up the most, as I've told Vittorio, is Camomila, the uh compiler, the template error compiler reducer. Actually, that's a terrible explanation. Uh but listen, I've had a couple Carlsbergs. The It's actually true. Wait, we're in Norway now and we're drinking Carlsberg? That's the beer of Denmark. What's going on here? Anyways, we're here with Vittorio. I'm handing the mic over to Bryce. He's headed out tonight, so we only have Bryce for another two hours minus the or two and a half hours minus the time for the C quiz. Bryce is gonna interview Vittorio. Hopefully Vittorio introduces himself for the listener. Here we go.

Bryce 2:42

Yeah, Vittorio, you I mean, how would you introduce yourself? I feel like you've done so many varied things. You're a trainer, you've worked in fintech, and you also work in game, the game design space.

Conor 2:58

I mean, how would you how would you introduce yourself? Last time I checked, you work at Bloomberg. Do you no longer work? Oh, how long have you not so wait? I the last time I checked, JF showed JF Bastion just showed up. I this is the thing. JF Bryce said like literally three minutes ago, uh, oh, JF and Chandler left. Chandler from uh Google and JF from some startup that we don't know what it does.

Bryce 3:20

Uh I have to read you this text exchange from JF. I uh I text JF, I ask him where he is, he says he's gone to the room to drop my bag in the room, coming back now, and I said, Have you seen Chandler? And he said, he's not in my room. Where clearly JF understood. That's not what I meant.

Conor 3:43

So wait, wait, wait, wait. Alright. We'll get to intro we'll get to introducing Victoria in a moment.

JF Bastien 3:48

This is gonna take some time, folks. So I I know you're young, but have you seen the Pink Panther, the movie? Yes, yes.

Conor 3:54

Yeah, we haven't seen the movie. We know who he is, though.

JF Bastien 3:56

With uh Peter Sellers, right?

Conor 3:58

No, we don't know the actors. We said we know what the Pink Panther is, that's it. We're in the 90s, okay?

JF Bastien 4:02

So there's one of the Pink Panthers, because there's multiple of them. I have him on VHS if you want to borrow them any day. Uh Peter Sellers is in a hotel and and he goes in and he he sees someone with the dog. I have the microphone. So he sees someone with a dog and he's like, Does your dog bite? And the person answers, No. And so Peter Sellers, the the inspector Cluseau, comes in and he goes to pet the dog, and then he gets bit by the dog. And he goes to the person and says, I thought you said your dog didn't bite. And then the person replies, That's not my dog. And so fairly often you and I have exchanges where you ask a question and I answer the question you asked.

Bryce 4:45

Did you bring your VHS tapes on your international move to Japan? I did.

Conor 4:54

Yes, but first, uh last time I checked, Vittorio was working at Bloomberg, and at one point, I mean, you can fill in the details, but this is what I know. You were an engineer at Bloomberg, and then you transitioned to an internal kind of trainer at Bloomberg, and you were in charge of kind of C education. I know that there's a guild system, I'm not super familiar because I've never worked there, but I remember having a conversation where you had kind of like transitioned to one of your dream roles of like being in charge of education at Bloomberg. But you no longer are at Bloomberg, and I saw it was Romeo.training, which hopefully is a URL, maybe it's not. But anyways, we're gonna hand the mic to Vittorio, and he can, or Bryce is checking if it actually is a uh uh a domain that you can go to. Introduce yourself and tell us about uh your work history and what you're up to now, because clearly you've moved on from Bloomberg.

Vittorio Romeo 5:42

Thank you, thank you. So it's my first time on the podcast, I want to say. I was expecting a C podcast, but it's so much more than that. So it's so it is very interesting, and I thank you for having me. And yeah, you're correct. I used to be a Bloomberg for the past 10 years, and the first half I kind of did software engineering on low latency, high throughput, high performance backends, including our microservice infrastructure, market order data analytics, and then I switched gradually organically to training. I felt like there was a gap in the way that we were teaching engineers C. There was nothing on Mono C, nothing on you know the newer standards, so I kind of started with something small. I realized it was something I was good at and I really liked. So over the course of multiple years it became my main focus and my main job, and I was in charge of the C education, especially the modern part. I stopped working at Bloomberg this September, so very, very it is very recent, yeah. So so yeah, right now I'm trying to start this venture where I'm gonna do independent consulting, contracting, and training. Um, so far it's new territory for me. So I have a few clients here and there, mostly individuals that are mentoring, and it's really rewarding to be able to focus on one person at a time and you know seeing them transfer transform their C Lite code into something a bit more you know modern and safe. So it's really rewarding to see that, but it's very new territory for me. So I'm exploring around seeing how it goes. But to answer your question, like yes, I've been doing a lot of very different things. I feel like the thing that ties them all together is my undying passion and love for C and good software engineering. So dabbled in very different domains, game development, finance training, but it's all always been C and nice code, but is probably the tiny.

Conor 7:26

Alright, so I've got two questions, and then I'll let Bryce ask the questions from here. But I am very curious. So if we had to rank corporations by their C usage, and also kind of when I say usage, I also mean like attendance at committee meetings. Bloomberg definitely makes the top ten and maybe makes the top five. I don't really know, but the point being is Bloomberg is one of the kind of major companies. I also think about it number three right now. Number three? Bryce says number three. Well, we're not talking about JavaScript right now. Wait, do you actually have like the top five or ten?

Bryce 8:05

Yeah, I know I I I know because it's easy to find it. You just go to the US uh national body and you look at the top five words. The C committee, yeah.

Conor 8:13

Yeah, you're not talking about like millions of lines of code. Anyways, the point being, Bloomberg is up there as like when I think of C shops, I think of Google, I think of Microsoft, I think of Adobe, I think of Bloomberg. Anyways, the point being, my first question is, how long were you in charge of education at Bloomberg? And my second question is, what is it like training one of these kind of frontier companies in terms of that, you know, you've been using C not just for years, but for decades. What is it like training and like getting people, I don't know if you're going from 98 to 11 or 11 to 20, what is that like? Because I imagine you have very interesting, I don't want to say war stories, but just stories to tell about that kind of education process.

Vittorio Romeo 8:51

Yeah, so I feel like training a Bloomberg was really interesting because you get an extremely diverse set of people, skill set, seniority, and you might have the super skilled engineer that knows everything about finance but hasn't moved from C in 10 years, or you might have the person that has been doing a lot of functional programming but never done that in C. So you get a very difficult, varied audience. So it does make it interesting because everybody has a different background, so the the interactivity and the the fun of the training is certainly higher as you get, you know, to get people to get to know better and they will kind of share their background and contribute in some way by you know having their train the training seen from their perspective, which is something that is interesting but definitely challenging. I've taught from other C 11, 14, 17, and 20. So I've done pretty much all the spectrum, and we offered them in uh you know separate courses so that people had this learning path that they could gradually learn newer and newer standards. For the first part of the question, I feel like Bloomberg is heavily investing in the future of C while some other companies are backing down a little bit. You mentioned Google, for example.

Conor 9:59

Google has been divesting from it, yeah.

Vittorio Romeo 10:01

Yes, exactly. Some companies have been divesting, they've been trying to maybe move to Rust or have their own competitor like Carbon. Bloomberg still heavily believes in C future. They are actively working in the standard to try and solve or mitigate, I don't know what the right word is, the safety issues and whatever makes C a bit less attractive nowadays compared to other languages. And they do have a huge code base which they don't intend to migrate to something else, they intend to keep maintaining and uh keep upgrading as the standard grows. So I feel like very big player in this area, and uh it explains why they put a lot of resources into the standard commit standard committee and keeping their code base in C ⁇ .

Conor 10:42

Well, I mean, we were just mentioning, was it I mean it might have been a week ago now, but whenever we had the Roku meetup, uh they were presenting uh we can't talk about the specific specifics of their presentation, but they showed some roasted code, which was row for Roku and STD for the standard namespace. And I went through off the top of my memory of all the different kind of insert letter SL standard libraries. And Bloomberg is one of the most well-known kind of open sourcing a lot of their tooling in their standard libraries, and there's a lot of talks, whether it's at uh, you know, just a plethora of C conference showing different aspects of that. So yeah, Bloomberg is very well known for kind of open sourcing their, you know, internal corporate standard library, and uh yes, anyways, very interesting uh to hear about the transitioning. What is the last question, then I'm gonna because I can tell Bryce got some questions. What is the uh what is the you said up to C20, is that the the latest that Bloomberg's at?

Vittorio Romeo 11:40

Officially, yes. Like we we used to have like a vetted compiler suite with a certain standard, and I do believe that we ended up at C20 while I was still in the company, but there were teams also dabbling with C23 before other teams. It's a very let's say teams have a lot of decision power, so if they had a good reason to, they could move to standards even before the other teams. But 20 was officially the the latest one. Good to know. All right, I'm handing the mic to Bryce.

Bryce 12:09

Well, it's actually it's uh very fortuitous timing that we're talking to uh Vittorio, a trainer, because we just did two workshops, and this whole year at uh at work at NVIDIA, there's been a big push on uh outreach and educational material, and I've been doing a lot more training and talks and developing training material. And uh Connor, this was your first time teaching a course. You did the first half of our C course. I guess I'm kind of I got a question for you, which is uh what was it like? As a first-time trainer, what was your experience?

Conor 12:44

It was exhausting. I love giving talks because they're an hour or 90 minutes max. An eight-hour training day. I don't know if it's the introvert in me, if it's the guy that likes to be in front of multiple monitors with an RGP keyboard, but just like eight hours is I mean, I didn't even you know, I wasn't speaking for the full eight hours, I only did half the day. Uh but it was exhausting. And I think also too, it's partly because I did not create the source material entirely myself. I feel like if I had designed a course specifically the way I wanted to teach it, and specifically geared towards the things that I wanted to teach, it would have been much different. But taking curated content from you know a bunch of folks and then having to teach content that's not, you know, you didn't you don't know it inside and out in terms of like, I don't know, like sometimes I was looking at the preview screen, looking at, is the solution on the next slide, or is there another slide showing like what to do? And uh you never have that when you're giving your own talk, like, or at least the way that I give talks, I've rehearsed it a couple times. It's almost like off the back of whatever, like, so you it's it's almost I don't know. Rehearse sounds bad because it sounds like you know, not the way, but you you practice and so you know what's coming, and so at no point are you ever like, oh, am I saying the right thing, or do I need to spend it longer on this point? Anyways, the point being, I think I could come to love training, but it if it was gonna be me loving a training course, I would have to basically do all the material from scratch. Which yeah, Bryce is like, oh, you just volunteer. No, I'm not volunteering to do that. I do not uh necessarily that's not the direction I want to take my career in. But anyways, I have a much greater appreciation for trainers. I assume you're gonna be creating all your content, so hopefully that will help with the training process. Uh but anyways, it's very exhausting. I think 90 minutes, in my opinion, it's the perfect talk length. 30 minutes is insanely too short. 45 minutes is okay, an hour's alright, 90's perfect, and eight hours is just insane. But yeah, this is my first time, so.

Bryce 14:49

I think at least we have longer 90. JF, do you have you ever done training? I mean, you you I assume when you were at Woven that you were you probably were responsible for some training, like figuring out like training and education. That was probably part of your.

Conor 15:02

We jumped the gun twice. JF has never actually formally been interviewed. The last time you were on Oh, not an interview. Yeah, yeah, not interviewed. The last time you were on was one of the highlights. Highlights of the podcast. We reference it often, folks. We're gonna see Sean Parent at C under the sea in T minus three weeks, two weeks. And JF was there. Part one, I think, of that three-part cut, uh, where the part two and part three were just with Sean was what's the word? A plethora, a schmorgisborg. There's no word to describe what happened. JF was there popping in and out and in and out. And I think we had we had Phil Nash, we had you, we had all um almost representation from every single C plus H podcast at the time. Anyways, folks may or may not remember you. You gotta introduce yourself, JF. I mean, you move, you move to Woven, aka Toyota. We talked about the Prius earlier, he said it wasn't his fault. He actually didn't say that. I'm just being nice. It might have been his fault, folks. But before that, he was at Apple. Now we don't know where he's at, and he's not telling.

Bryce 16:03

He actually does, he does have a because it sounds, without context, it sounds like you have some it sounds like you have some issue fundamentally with the entire Prius and that the Prius is JF's fault.

Conor 16:19

I mean we don't know. That might be the case. He he hasn't told us.

Bryce 16:22

You didn't clarify that it was that you have UI issues with the Prius software, not with the whole idea of the Prius A car.

Conor 16:30

I'm an owner of a Prius. Great car, bad software. It overheats my phone when charging, so we don't charge it. Anyways, on LinkedIn it says stealth startup. On his badge it says stealth. Am I allowed to say this? Physical systems. Actually, does it say on LinkedIn? It might say that on LinkedIn as well. Anyways, it says stealth, physical systems startup. So we're getting two more words. It's probably the extent. I mean, we can't ask, and he's allowed to say. It says physical systems startup on LinkedIn. I am rec I am corrected. And we can always ask, and he is always allowed to say at this moment in time, I can't say more. And potentially he'll say, actually, I can tell you exactly what we're doing. Anyways, introduce yourself from Apple to Toyota to who knows.

Bryce 17:13

And you know where JF used to work back in the day.

Conor 17:16

Nvidia. And I actually didn't know this when we first met. I think I was just falling in love with array languages, and you told me a little story where you had created, like, I don't know if it was an array DSL at NVIDIA, and I just thought that was crazy, and I was like, why didn't that take over? Like, array languages, GPUs, they're a perfect fit. And it sounds like you were you were tapping on like the gold mine and you just didn't tap hard enough because I think you really, you know, anyways, we're handing the mic to JF. I feel like we just asked you 17 questions, so feel free to answer as many as you would like. Uh here you go. Hi, my name's JF. What do you want to know? Yeah, I mean, uh So what what uh what's your AI startup on? It's not an AI startup. I know I was just I was hoping that you would be like answer the question accidentally because odds are it is an AI startup, but uh clearly you didn't fall for the bait.

JF Bastien 18:09

No, we we use GPUs. Yeah, GPUs are good, they go burr. It's like, yeah. Yeah, they do a lot of stuff.

Conor 18:15

Mostly just going burr.

JF Bastien 18:17

Yeah, yeah. Um, yeah, what do you want to know? So uh do you want to know about dynamic binary translation?

Conor 18:22

Yeah, well no, he asked you about uh the first question that was asked was about the training.

JF Bastien 18:27

And then we uh so I hired Robert Secord to do a lot of the trainings, and then we just at Woven?

Bryce 18:35

Yeah. What was the mindset? Like as as somebody as somebody was putting together internal training, what was like what were you trying to train?

JF Bastien 18:42

Well, so I'm not a company representative anymore, so I can't officially talk about the company. Uh but yeah, I mean every big company that has a uh variety of type of developers does training, so it makes sense to train your developers, especially when you develop safety systems. So, like one interesting aspect, and this is like part of international standards, like when you do uh safety software development, is you have to have a safety mindset when you develop stuff. That's part of like ISO 2662, but it's part of other standards as well. Right? And so part of having a safety mindset is actually having training towards safety and what it is and how to apply it.

Bryce 19:18

Um I don't really have training at NBDS. I guess we don't have a safety mindset.

JF Bastien 19:22

I you're you're your part of the world that does ISO 2662 does trainings aimed towards that. Um yeah, do you want to talk about binary translation? What do you want to do? Yeah? Wait, actually, a question popped into my head. TLB Hit. Seven episodes. What's going on, buddy? Yeah, so TLB Hit is a weekly podcast that my friend Chris and I start started uh during the pandemic. No, no, it inspired your podcast and and Matt Godbowt's podcast. You should go back, you should go back in the internet and look. Nothing vanishes.

Bryce 20:01

It's weekly in that like they release one podcast in the same week every year.

JF Bastien 20:05

You you guys work on concurrency and parallelism weekly in the sense of like weak memory systems. There's gonna be eventual Yeah, there's gonna be eventual consistency in the podcast. Uh but yeah, so Chris and I have this weekly podcast called TLB Hit, TLBH dot IT. Uh and it's a wonderful podcast. We have we have a plethora of episodes. There's even some unreleased episodes. What's going on? Yeah, we haven't released them. Uh he does have kids.

Bryce 20:34

Both him and his co-hosts have kids. Doesn't your cohost have kids?

JF Bastien 20:37

Not together, I just want to clarify. Um wonderful guy, but we uh we tend not to have twins, right? Yeah, he has twins and then know the kid. Yeah, yeah, yeah. Uh but one of the issues we have is he lives in California and I live in Japan, and so it makes it difficult to. But you were also not really It makes it difficult to have kids in that circumstance.

Bryce 20:57

You were not releasing You were not releasing episodes when you both lived in California.

JF Bastien 21:03

Well, so you need to acquire before you release when we were talking about weak memory systems, and uh I remember I forgot how painful it is kind of. I know, I know. So uh yeah, so but it's a great podcast, uh really good times when we record it. I love being on podcasts, it's amazing. You remember? Oh well it's uh we count from zero, but it's on the website. You should go and check it out. I mean I have checked it out. It's gonna be in the show notes. It's gonna be in the show notes.

Conor 21:29

Wait, wait. I mean, we will have uh at some point we actually do have to align time zones and we'll interview JF. But I I discovered I was doing a deep dive into Jax and Jax's back end XLA, now open XLA. And I look for some talks on it. I find two online, both by the same person. A guy named Chris, and what's his last name? Leary. A guy named Chris Leary. At the time, I was like, I don't know who this is. And then at some point, like a week later. I was like, wait a second. Because I think I I got to watching the second talk. The first talk was amazing. Both talks were amazing. And I was like, wait, this is JF's co-host? JF's co-host is basically, I don't know if he was the solo author or the co-author of XLA, which is the live you know this? The back end to like the premiere XLS also. XLS anyway, I was like, we gotta have him on the pod too. Will we have JF or will we have Chris Leary? And I had no idea, because he's at OpenAI now, so we know this guy's probably a babillionaire.

JF Bastien 22:31

Is this still order in XLS? Where where did you work again? Remind me? NVIDIA?

Bryce 22:36

Oh right, okay, cool. Right.

JF Bastien 22:39

You guys just gave a hundred billion to open AI, right?

Bryce 22:41

Yeah, I think we own credit of OpenAI. 10% are simple.

Conor 22:45

I haven't checked any. I've been here in Europe for a week. All I heard was Kimmel was cancelled. It's uncancelled?

Bryce 22:53

Let's told you.

Conor 22:54

It's uncancelled. It's uncancelled? Yeah, it's uncanceled. Well that's fantastic. I thought, well, actually, we're gonna just veer away from this topic because But uh we support democracy here in Canada, except we're in Norway, but Norway also supports democracy. Anyways, Chris, the author or like the founder of XLA, I found that uh very uh exciting. Oh, well I I have a fun fact for you.

JF Bastien 23:17

We have a fun fact. Or Teddy. Chris and I used to work together at NVIDIA. What? Yeah, Chris was in NVIDIA with me.

Conor 23:23

So we have now we have to our podcast.

JF Bastien 23:26

And he he sat right across the the the cubicle. The cubicle right in front of me was Chris. And so I start on Monday and I show up at work and people like introduce me to everyone, and then it's like 11 or something, and they're like, so this is Chris's office, he's not in yet. And so I met Chris like the next day or something. It was pretty good, yeah.

Conor 23:44

So what did you do at Nvidia? Wait, wait, wait, yeah. What did you do and why'd you leave? I mean, uh, you know, seems like a pretty great place to work since I've been there. What happened?

JF Bastien 23:55

Well, so uh you know NVIDIA builds things other than GPUs, right?

Conor 23:59

I don't know, honestly. I did an ivory tower research.

JF Bastien 24:02

Yeah, so I worked on the the first like CPU that NVIDIA designed. Uh called called Project Denver. It's uh it's on the Wikipedia as you can look it up. Uh and so it has a dynamic binary translator that translates from ARM to uh ARM64 to uh uh internal VLIW architecture. It's very exciting.

Bryce 24:21

Uh the idea was that it could translate in theory from other things.

JF Bastien 24:25

Right, so uh if if if you're you're familiar with history, it's I'm not, I'm not. So do you know TransMeta? TransMeta? Yeah, TransMeta was the hot startup. It was it was the hot startup in the 90s. Uh that's the 1900s, 1990s. Uh and um it it was uh uh Linus used to work there, Linus Torvalds. Yeah, and so uh it was a company that made an X86 chip uh that competed with uh Intel and AMD and Via at the time, and uh they they were they were first to market with really low-powered chips, it was really cool stuff. Uh they ch they shipped two CPUs that they designed, and then their third one kind of didn't really make it because reasons. Um but some of the IP ended up in other places, including the type of stuff that we were doing at uh was the reason that Intel sued them? Uh you're the NVIDIA representative. You can tell me why Intel sued NVIDIA. I it would not be my place to say this. Uh but there is there was a lawsuit at the time uh I worked at NVIDIA uh and there was a settlement for around a billion dollars, which is public. You can go and read the settlement and it has interesting clauses in it. And Intel at the time paid NVIDIA around a billion dollars uh for the settlement. It was around the time they did MCP. No, so they they had they had chipsets, right? Nvidia used to make chipsets, they were like the best chipsets for gaming and stuff, it was really good, and you could plug your GPU into it and stuff. Uh and then there was a lawsuit around it. You you could just go look it up online. It was a it was a good uh chipset thing that they did.

Conor 25:57

Okay.

JF Bastien 25:57

Yeah. And then you left the streets. Well, I left to go to Google.

Conor 26:01

Wait, you left to go to Google and then you tap.

JF Bastien 26:03

And then I would tap it, yeah.

Conor 26:05

I see. So you were moving up in the because at the time, Nvidia nobody knew who they were.

JF Bastien 26:11

I knew who they were, yeah.

Bryce 26:13

No, nobody knew who we were. I was I was told, I was told by somebody who's at this conference that I should like in 2017 that I should not go work at NVIDIA because they did not have competitive pay and and uh that it was not gonna be as good as going to Google or Amazon or Microsoft or somewhere. But I believed, folks. I believed, I thought it was the coolest company, the only company that made true parallel hardware. Okay, where were we?

Conor 26:43

Cool story. I don't know where to go from there. Are you allowed to tell us anything about your startup?

JF Bastien 26:51

Yeah, it's a physical systems startup, and so we build physical things, and you gotta program those things and have a platform to be able to program them. It's very exciting. Yeah.

Conor 27:01

They build physical things. Yeah, it's nice. Are you allowed to say anything about the tech stack? I mean, we're talking Rust, we're talking C, we're talking carbon. Are you allowed to say anything like that? That doesn't seem like proprietary, you know.

JF Bastien 27:14

I mean, you gotta use the right tool for the job. So the Python's pretty decent for some AI stuff. There's a good amount of C Rust, there's great TypeScript for web-related stuff. Are you just listening off everything or everything you use? We don't use APL. But if you come to my talk tomorrow, I have an APL joke. You'll like it. You'll like it. Yeah, it's a good talk.

Bryce 27:34

Apparently, I'm in the talk, but as I No, I showed you the slide, you're in the talk, yeah. As I told them, I'll be on a plane, so I can't be in this talk.

JF Bastien 27:40

No, you are in the talk. It's done. Yeah.

Conor 27:43

Very rude that you would leave before JF stock. Should have checked the schedule, buddy. I know, I know. I would never miss because JF famously. I think actually this could be wrong. I'll cut it if it is. You only give your talks once, right? Yeah. Yeah. So I mean, the last time I saw a talk of yours live was at C under the C. I imagine you've given a talk since then in between, because that was like a couple years ago.

JF Bastien 28:06

I think it was like a year ago.

Conor 28:08

Nah, 2023? Oh, so I guess that's a year and a half from 2025 now. Uh have you given talks in between? Can we link them in the show notes?

JF Bastien 28:18

Or is this your return from a year and a half uh I think I gave uh C now talk, the safety and security one? The C CPP on C one was my most popular though. It was a good talk. Yeah.

Conor 28:31

That was uh that was like bookended with like it was there were three safety and security talks. Sean gave one, you gave one. I can't remember the third one. I apologize to that person. Uh but I definitely recalled a panel.

JF Bastien 28:44

I did a panel also.

Conor 28:46

Was that recorded?

JF Bastien 28:47

Yeah, I think so.

Conor 28:48

I might was that C now?

JF Bastien 28:50

Yeah, I think so.

Conor 28:51

I might have watched that as well. But I definitely remember enjoying both yours and Sean's. Uh Sean, I mean, Sean's fantastic. He's always got great takes. Anyway, so this is your first talk since was it C now this year? Honestly. Two, three years ago. Because I was gonna say that uh it feels like ages ago uh when I watched that talk. Um anyways, tell us a little bit about your talk. I mean, Vittoria, you're speaking tomorrow as well, right? Or we spoke today, so wait, we'll go to Vittorio first. Give us a recap, how did your talk go? And then we'll get a preview of JF's talk tomorrow. Uh I know. Hold back and wait. No, uh Jason's talk is called implicit conversions considered harmful.

JF Bastien 29:34

Your talk is called Hoping Noping, what about noping or something like that? Much ado about noping. So it's a talk about nothing. It's really good. I've been working on it for eight years. Eight years, alright.

Conor 29:47

That's a bit that's a bit much, but I I do recall I got the nope or the noping correct. We got 15 minutes before this quiz. Uh Victorio's talk, I don't think uh I recall, I definitely didn't see it today. But uh tell us, how did your talk go? What was it about?

Vittorio Romeo 30:04

So not only you didn't come to my talk.

Conor 30:06

Wait, what time is it at? What time is it at?

Vittorio Romeo 30:08

You you don't even know if you came to my talk.

Conor 30:10

No, no, no. I definitely know I wasn't in it.

Vittorio Romeo 30:12

I was only in two and a half talks today. I'm pretty offended. I I think I want Bryce to ask me questions from now on, is that okay? What time is it at though? What time is it at? Was it 10 30?

Conor 30:22

That's why I wasn't in it. I was running during that.

JF Bastien 30:25

I'll just say, he said he would definitely be at my talk, which is at the same time.

Vittorio Romeo 30:30

So yeah, uh, I mean I I used to thought we were friends, I used to thought we had, you know, something, but there's no way to backpedal to this one.

Conor 30:38

I know, I know.

Vittorio Romeo 30:39

I mean, the audience heard everything, so they can make their own mind, right?

Conor 30:42

They they they'll understand this GitHub discussion, and uh you be the judge. You be the comic.

Vittorio Romeo 30:48

Anyway, my talk went great. I gave this talk also CPPCon as a keynote, which was my first ever keynote. So uh it is online right now. It was shared like immediately afterwards. And please go see it. Let me know if you enjoyed it. It's about data-oriented design and um sorry, wait, what design? Uh data-oriented design.

Conor 31:08

That sounds like the Mike Acton talk of like the CPPCon 2014 or something. Exactly, that's the right year.

Vittorio Romeo 31:14

And I actually started with it in my in my keynote because I was there in person and it was like the personal start of my journey into DOD.

Conor 31:21

I think to this day that is like one of the top five viewed. It is the it's number one. JF's correcting me. I mean, I'm not wrong in it wasn't in the top five, but JF is uh qualifying. It is the number one viewed talk on all of CVP Con videos released in history.

Vittorio Romeo 31:36

Yeah, I feel like Mike's keynote was eye-opening for the people, me included. I do feel like he's very dogmatic onto some stuff, especially regarding C feature usage and templates and abstraction. So I try to have a little bit of a more balanced outlook and kind of like balance out DOD with OOP, which are supposed to be mortal enemies, but I think they can coexist. And also I believe that there is quite a lot of room for C features and abstractions to actually make your life easier. One of the things I show in the talk, for example, is taking a normal struct and automatically converting it to an SOA layout using reflection even in C 17. And it's it's actually better.

Bryce 32:17

Yeah, it's one of the that that that that example is one of the ones that we actually contributed to the whole reflection effort, and it's one of my favorite things.

Vittorio Romeo 32:25

But but this I'm doing this in C 17. You can do it, you can do it before 2076.

Conor 32:31

I did see uh, was it on LinkedIn? It was some social media platform where you posted everyone's excited about C26 reflection, but don't forget, we can do a lot of this stuff.

Vittorio Romeo 32:40

Is that yeah, that that's it, that's it. So it's like everybody is super excited for reflection, but many of the use cases that are very useful with reflection you can do today. You have Boost BFR, you have this weird arcane magical trick that you're gonna do with templates to even extract names of fields from structs in a portable way and then reflect on those in 17. Or maybe it's 20 for the field names and 17 for aggregates. But still, you have a lot of power today that people maybe don't even know you have, which is a bit sad because you can do automatic serialization, automatic data layout changing. You can do quite a lot of cool stuff nowadays with uh what you have at your disposal.

Conor 33:17

That's amazing. So, and the great news is is I don't know exactly all right, JF's is gone. We won't ever hear about his talk. I mean, I guess he said it was about nothing, he's been working on it for eight years, seems odd. But uh the great news is that even if NDC Tech Town doesn't release the video at the time this is released, it's sometime probably late October, we will have a link to the keynote that was given at CPPCon. And this does sound like a fantastic talk. It's not a repeat or a different edition of Mike Acton's talk, it is an extension. It's an ext less angry version of Mike's Was Mike's talk angry?

Bryce 33:50

It Victoria gave a less angry version of Mike's talk. Mike's Mike was angry. That's I don't remember much about that. He was angry. He was angry?

Vittorio Romeo 34:00

I I feel like I didn't I didn't want to do the same thing on Mice Acton. I I took a different approach where it's like a walkthrough. I show people, hey, let's start with an OP sort of thing, then we benchmark it. We see why it's low, we talk about memory, then we actually gradually migrate to something more data-oriented, we benchmark it again. So it's a very kind of like uh step-by-step approach where you see pedagogically how you can shift your mindset and how that actually affects the final runtime of the of the project. So I feel like Mike Mike Acton is very uh angry talk, if you want to call it that way, still has a lot of value, and I would recommend people to see it. It's very different thing. So uh don't wanna, you know, overlap on the same thing.

Bryce 34:37

I do just want to say uh don't get me wrong, well I do think it was an angry talk. It was one of my the best talks I've ever seen. That that is up there with Sean Parents, C now, keynote from like 2012 or 2013, the one we always talk about that we love, where he shows uh where he basically talks about GPU programming in the early days.

Vittorio Romeo 34:57

I feel more than angry, the right word is frustrated, and and you can be frustrated for very good reasons, and I feel like that's what motivated Mike.

Conor 35:04

Well, I was gonna say, I don't think there's anything wrong with having a little emotion, a little anger, a little frustration. Uh to be honest, I think I'm way too Canadian nice, and uh, there's times on these podcasts where the voice in my head is like but then and and like literally my uh my wife, she my or I guess fiance soon to be wife, she'll be like, I will be she'll be listening to it in the car and then she'll give me feedback, and then she'll be like, why don't you push back harder? And she'll be like, oh, you know, I I completely disagree with the point being made. And she's like, Well, why don't you push back? And I was like, Well, you know, I don't want to be disagreeable. But the point being is emotion, and it's like you can still respectfully have like a an emotional kind of reaction to someone's opinion and say, listen, you know, I appreciate that's your opinion, but I wholeheartedly comp think you're completely wrong. And it's actually good for it's good for ratings, you know? I think in the year of 2025, in the last decade and a half, people are too worried about upsetting listeners and guests and stuff. Like we've lost the ability to have like honest conversations. Anyways, so a goal going forward, or at least she's trying to convince me, is like whenever I disagree with whether it's a guest or a host, you just say, listen, I appreciate your point of view, but you're wrong. And then you know, and then you go but respectfully. And anyways, Victoria wants to add to that.

Vittorio Romeo 36:21

Yes, the message of course is be angry. That's that's what we're saying. And also, you said you're Canadian nice, or you've been very mean to me today. I I will not forget that.

Conor 36:31

It is true. I didn't know the name uh I knew the name of JF's talk. Well, actually, I didn't know the name of JF's talk, but I did say I would be at it, and I knew the name of Jason's talk, and Jason, he's not around, I think he's gone. We might talk to him tomorrow, and by we I mean me, because Bryson's gonna be gone. But I did not know. I mean, I do recall reviewing the schedule, but I was trying to fit in a 20K run, and so I had to choose a two-talk slot, and the morning slot was the one. And if it makes you feel better, uh I missed lunch today because I miscast I thought if I showed up before it doesn't. Alright, Victoria says it does not make him feel better. Uh we will rectify this situation by going and watching the CPP keynote either on the train to Oslo or sometime before then.

Bryce 37:14

Wait, wait, wait. A keynote?

Conor 37:16

A keynote, so it's already online. That's what when he said keynote, I was like, a lot of the times now, C now, CPP con here, here's a here's a hot take and a disagreement. Stop doing this thing, or you wait like four months and then you roll out talks like one I get you think it's good for the algorithm, but you're upsetting your loyal like watcher fans.

Bryce 37:34

It's not about the algorithm. It's not a good thing. It's not about the algorithm, it's about selling in-person tickets. If they put the videos up immediately, the thinking is people are less likely to come in person. If you want to see the keynote without a form of delay, you gotta come in person.

Conor 37:50

That's a broken mindset though, because people, I mean, at least people that have been to a con you don't go to the conference for the talks. You go to the conference for the conversations, the hallway track. I mean, honestly, the most valuable thing about this trip was our conversation that we had in Copenhagen over that dinner, and I explained to you the iterator and validation thing, and you were like, that doesn't sound like an iterator and validation thing. Since then, I've gone and fixed that issue. It's it's as JF says, the GPU goes brrr, and we've got our Row Wise Softmax working in two kernels. It's beautiful, folks, and I am more confident about this library that we're launching than ever. We haven't even talked about it yet, but the point being, that has nothing to do about it's the in-person conversations that you have with people. It's not about the talks that you watch. I mean, admittedly, you do learn stuff from the talks you watch, but like the most valuable stuff is these moments right here. Alright, we're handed to Vittorio because he's the guest, right?

Vittorio Romeo 38:43

So I'm I'm gonna organize a conference with no talks. There's gonna be it's gonna be 3k per entry, and you're gonna be there, right?

Conor 38:50

Honestly? That is like you're joking? That's a pretty decent idea. Like they have had these kind of co-located, like what do they call unconferences where there are no scheduled talks, but they just get together a space. Yeah, there's a space, and so you'll have this kind of either one or two rooms, and then at some point they'll do like a hey, does someone have something they want to show or whatever, show and tell, or maybe someone does have a presentation they want to show. Like I think that those are a lot harder to like fund and to get people to show up to, but that like it's a joke, but I think there's actually a ton of value in that kind of stuff because if you can make it work right, it is the best form of like unstructured conversations, blah blah blah, breakout groups. Anyways, so I think you could be massively on I mean it's not my point. These things exist, like unconferences are a thing.

Bryce 39:41

I went to one for I don't remember what, but it was some some Google thing. So I uh I don't know if we've mentioned this in the podcast before, but uh me and uh Ramona and I are in the nascent stages of planning a New York-based conference, and we keep having this conversation about maybe we just won't have keynotes. Because I feel like a Vittorio keynote's gonna be good, but I feel like on average, I am disappointed with keynotes at a lot of conferences. No, legitimately, I am disappointed with a lot of keynotes because I think that uh uh oftentimes they give people keynotes and they got an open check about what they're gonna what they're gonna present, and there's no review or feedback process on the keynote, and so sometimes you get keynotes where they're just not they're just not so good.

Conor 40:24

I will say there was that was it a blog post or conference trip report? At some point I posted somewhere for the love of God and I'll call out the conference, it was CPP con. Please use a monospace font for your code examples. Just cause it's in a slide deck doesn't mean and everybody I mean calling out a very specific person. No no, I w three of the five keynotes did this, and one of them everybody's gonna think you're calling out a- I mean that's fine. I honestly, I honestly don't care. We just finished saying that it's okay to have disagreements with people, and honestly, using a monostace spot, it's not a lot of work. It's not a lot of work, folks. You can do it, I can do it, we can all do it. And I don't unless if your talk is about like uh what's the guy's name, Brett Victor, and he has this whole thing is that like we're stuck in this like page of text, everything is blah blah blah. It should be a newspaper. Some things should be bigger, like the function names. Unless if you have a talk about the way that we're formatting and styling code, I want to see a monospace font. It's easier to pattern recognize, like it's just it's just the the coding standard. There's no IDE or like editor or any online tool that offers you a non-monospace font. And when I see it, I'm just like, what are we doing here? And it is just laziness. I'm gonna be honest, it's just laziness. You're copying and passing and you're not choosing the font. I guarantee you, Victoria's talk, it had a monospace font. I can say that confidently because well actually I'm 99% confident.

Vittorio Romeo 41:56

Victoria can let us know. I just want to add that, like, you mentioned there's no editor, but my university professor's my favorite editor, Microsoft Word, comes with uh non-monospace font, and all the code was written there. But he didn't use one, but yeah, yeah, I agree with you, like in general, I feel like is the bare minimum you can do to make your slide more readable and making code monospace, and I'm still surprised that sometimes even very technically good talks, you know, completely do not care about the aspect of visual presentation and making things more readable, which is a bit of a shame.

Conor 42:33

Yeah. I mean I do think, and then this is uh we're in hot take territory, folks. I do really think it is just like a lack of effort. It's not that it's not a good talk, or it's just literally like it's extra effort to go and change the font, and they just copy pasta and they're done with it. Right now I'm on Linux, folks, and I make my talks in Microsoft 365 PowerPoint, the web version, and I lose my indentation. I get the for I get the syntax highlighting, I've learned a trick from Ashwin here. He's not saying much, but he's offering a lot, and he taught me the trick on how to and actually I'll tell you the trick. You gotta copy to Google Slides first, and then to Microsoft Online. I haven't tested it yet, but I trust that it works. But for the talk that I gave at the meetup at Roku and the meetup in Copenhagen, I went in painstakingly. Every code example, I went and I typed space by space for every single indentation. And I tried to choose code examples that had the minimum level of like nestedness, because if you get three levels, that's 12 spaces I have to type, but that's how much I care. Like it is my fault for not suffering through Windows, because Windows has the copy pasta working correctly. But the point being, folks, is you should care. You should care. Changing the font is a single click. We got multiple people with comments.

Bryce 43:44

I'm gonna make a sp certain specific coworker of ours listen to this. Because there's a certain slide deck which has code, which not only is in a non-monospace font, it's also got no indentation. It's all on the left line.

Conor 44:02

We're handing it to Vittorio, but I'm gonna beep this out. Uh who is it? You are not I'll beep it out, I'll beep it out, I'll beep it out.

Bryce 44:08

You're not getting it from me.

Conor 44:09

You're not gonna Yeah, that's unsurprising.

Vittorio Romeo 44:14

It's beeped out though, don't worry. Because we're talking about venting our frustrations, PowerPoint. It lacks one of the most basic features that every program has, which is resizing text proportionally. Like if you drag from the corner of the text box, it's just gonna resize the boundaries and it's gonna reflow the text. And I like morphing. I like when I have slider move around and code move around. So at the beginning I used to do this, I used to convert every single text box into an image so that I can resize it proportionally. And whenever I had to make a change in the keynote, I had to do it from scratch, change the code, and then reconvert to an image and morph it again. And then I realized I'm a programmer, I can fix things. So I made an add-in for PowerPoint that allows you to actually proportionally resize text. And you know, you have no excuse. If you want to make it look good, you can even go beyond and just fix PowerPoint. Which Microsoft should do honestly. It's been 20 years, 40 years since you cannot resize things proportionally. That makes me frustrated.

Bryce 45:10

I'll pay you like 20 euros for the plugin.

Conor 45:13

I was gonna say, is it public? I mean, that is uh that's an amazing point. I mean, I think that the the text, if you resize it, yeah, yeah. The quiz is starting, so we are we are ending this conversation. It's good timing because it's been 45 minutes. But I'm I'm pretty sure if you resize it, it just goes outside the bounds of the box. And so you actually only have to, but it is a good point. It it should fix that.

Bryce 45:34

We didn't even we didn't even talk about Vittorio's background and like game design, uh, but uh that's such a now that now that you've said that, like I definitely want that, and I de I haven't realized until now that that's something that I needed.

Conor 45:48

We have to have Vittorio clearly on the pod because Bryce and I care deeply about PowerPoint. We spend an inordinate amount of time. There's a couple episodes that I've I've titled like C X or Y. It's a fifth 40-minute podcast, 30 minutes are talking about PowerPoint, and really it should have PowerPoint in the title. Anyways, maybe we'll chat with you afterwards, Vittorio. Thank you for coming on. We'll get you on a formal interview. Uh thanks for coming on. I would love that. Thank you so much.

Bryce 46:14

If people want to hire you to do training, where do they find you?

Vittorio Romeo 46:17

Yeah, so I have a link called Romeo.training, which is my, let's say, personal webpage where I describe my services and you have a contact form there. But you can also find me on Twitter, LinkedIn, anywhere. I'd be super happy to chat with you.

Bryce 46:28

And that is Romeo like Romeo and Juliet. Romeo like the Shakespeare play dot training.

Conor 46:33

No one's confused about Romeo, my guy. Anyways, have a good one, buddy. Be sure to check these show notes either in your podcast app or at adspthepodcast.com for links to anything we mentioned in today's episode, as well as a link to a GitHub discussion where you can leave thoughts, comments, and questions. Thanks for listening. We hope you enjoyed and have a great day.

Bryce 46:49

Low quality, high quality, that is the tagline of our podcast.

Conor 46:55

It's not the tagline. Our tagline is chaos with sprinkles of information.