Podcasts about svgs

Join Kyle, Nader, Vibhu, and swyx live at NVIDIA GTC next week!Now that AIE Europe tix are ~sold out, our attention turns to Miami and World's Fair!The definitive AI Accelerator chip company has more than 10xed this AI Summer:And is now a $4.4 trillion megacorp… that is somehow still moving like a startup. We are blessed to have a unique relationship with our first ever NVIDIA guests: Kyle Kranen who gave a great inference keynote at the first World's Fair and is one of the leading architects of NVIDIA Dynamo (a Datacenter scale inference framework supporting SGLang, TRT-LLM, vLLM), and Nader Khalil, a friend of swyx from our days in Celo in The Arena, who has been drawing developers at GTC since before they were even a glimmer in the eye of NVIDIA:Nader discusses how NVIDIA Brev has drastically reduced the barriers to entry for developers to get a top of the line GPU up and running, and Kyle explains NVIDIA Dynamo as a data center scale inference engine that optimizes serving by scaling out, leveraging techniques like prefill/decode disaggregation, scheduling, and Kubernetes-based orchestration, framed around cost, latency, and quality tradeoffs. We also dive into Jensen's “SOL” (Speed of Light) first-principles urgency concept, long-context limits and model/hardware co-design, internal model APIs (https://build.nvidia.com), and upcoming Dynamo and agent sessions at GTC.Full Video pod on YouTubeTimestamps00:00 Agent Security Basics00:39 Podcast Welcome and Guests07:19 Acquisition and DevEx Shift13:48 SOL Culture and Dynamo Setup27:38 Why Scale Out Wins29:02 Scale Up Limits Explained30:24 From Laptop to Multi Node33:07 Cost Quality Latency Tradeoffs38:42 Disaggregation Prefill vs Decode41:05 Kubernetes Scaling with Grove43:20 Context Length and Co Design57:34 Security Meets Agents58:01 Agent Permissions Model59:10 Build Nvidia Inference Gateway01:01:52 Hackathons And Autonomy Dreams01:10:26 Local GPUs And Scaling Inference01:15:31 Long Running Agents And SF ReflectionsTranscriptAgent Security BasicsNader: Agents can do three things. They can access your files, they can access the internet, and then now they can write custom code and execute it. You literally only let an agent do two of those three things. If you can access your files and you can write custom code, you don't want internet access because that's one to see full vulnerability, right?If you have access to internet and your file system, you should know the full scope of what that agent's capable of doing. Otherwise, now we can get injected or something that can happen. And so that's a lot of what we've been thinking about is like, you know, how do we both enable this because it's clearly the future.But then also, you know, what, what are these enforcement points that we can start to like protect?swyx: All right.Podcast Welcome and Guestsswyx: Welcome to the Lean Space podcast in the Chromo studio. Welcome to all the guests here. Uh, we are back with our guest host Viu. Welcome. Good to have you back. And our friends, uh, Netter and Kyle from Nvidia. Welcome.Kyle: Yeah, thanks for having us.swyx: Yeah, thank you. Actually, I don't even know your titles.Uh, I know you're like architect something of Dynamo.Kyle: Yeah. I, I'm one of the engineering leaders [00:01:00] and a architects of Dynamo.swyx: And you're director of something and developers, developer tech.Nader: Yeah.swyx: You're the developers, developers, developers guy at nvidia,Nader: open source agent marketing, brev,swyx: and likeNader: Devrel tools and stuff.swyx: Yeah. BeenNader: the focus.swyx: And we're, we're kind of recording this ahead of Nvidia, GTC, which is coming to town, uh, again, uh, or taking over town, uh, which, uh, which we'll all be at. Um, and we'll talk a little bit about your sessions and stuff. Yeah.Nader: We're super excited for it.GTC Booth Stunt Storiesswyx: One of my favorite memories for Nader, like you always do like marketing stunts and like while you were at Rev, you like had this surfboard that you like, went down to GTC with and like, NA Nvidia apparently, like did so much that they bought you.Like what, what was that like? What was that?Nader: Yeah. Yeah, we, we, um. Our logo was a chaka. We, we, uh, we were always just kind of like trying to keep true to who we were. I think, you know, some stuff, startups, you're like trying to pretend that you're a bigger, more mature company than you are. And it was actually Evan Conrad from SF Compute who was just like, you guys are like previousswyx: guest.Yeah.Nader: Amazing. Oh, really? Amazing. Yeah. He was just like, guys, you're two dudes in the room. Why are you [00:02:00] pretending that you're not? Uh, and so then we were like, okay, let's make the logo a shaka. We brought surfboards to our booth to GTC and the energy was great. Yeah. Some palm trees too. They,Kyle: they actually poked out over like the, the walls so you could, you could see the bread booth.Oh, that's so funny. AndNader: no one else,Kyle: just from very far away.Nader: Oh, so you remember it backKyle: then? Yeah I remember it pre-acquisition. I was like, oh, those guys look cool,Nader: dude. That makes sense. ‘cause uh, we, so we signed up really last minute, and so we had the last booth. It was all the way in the corner. And so I was, I was worried that no one was gonna come.So that's why we had like the palm trees. We really came in with the surfboards. We even had one of our investors bring her dog and then she was just like walking the dog around to try to like, bring energy towards our booth. Yeah.swyx: Steph.Kyle: Yeah. Yeah, she's the best,swyx: you know, as a conference organizer, I love that.Right? Like, it's like everyone who sponsors a conference comes, does their booth. They're like, we are changing the future of ai or something, some generic b******t and like, no, like actually try to stand out, make it fun, right? And people still remember it after three years.Nader: Yeah. Yeah. You know what's so funny?I'll, I'll send, I'll give you this clip if you wanna, if you wanna add it [00:03:00] in, but, uh, my wife was at the time fiance, she was in medical school and she came to help us. ‘cause it was like a big moment for us. And so we, we bought this cricket, it's like a vinyl, like a vinyl, uh, printer. ‘cause like, how else are we gonna label the surfboard?So, we got a surfboard, luckily was able to purchase that on the company card. We got a cricket and it was just like fine tuning for enterprises or something like that, that we put on the. On the surfboard and it's 1:00 AM the day before we go to GTC. She's helping me put these like vinyl stickers on.And she goes, you son of, she's like, if you pull this off, you son of a b***h. And so, uh, right. Pretty much after the acquisition, I stitched that with the mag music acquisition. I sent it to our family group chat. Ohswyx: Yeah. No, well, she, she made a good choice there. Was that like basically the origin story for Launchable is that we, it was, and maybe we should explain what Brev is andNader: Yeah.Yeah. Uh, I mean, brev is just, it's a developer tool that makes it really easy to get a GPU. So we connect a bunch of different GPU sources. So the basics of it is like, how quickly can we SSH you into a G, into a GPU and whenever we would talk to users, they wanted A GPU. They wanted an A 100. And if you go to like any cloud [00:04:00] provisioning page, usually it's like three pages of forms or in the forms somewhere there's a dropdown.And in the dropdown there's some weird code that you know to translate to an A 100. And I remember just thinking like. Every time someone says they want an A 100, like the piece of text that they're telling me that they want is like, stuffed away in the corner. Yeah. And so we were like, what if the biggest piece of text was what the user's asking for?And so when you go to Brev, it's just big GPU chips with the type that you want withswyx: beautiful animations that you worked on pre, like pre you can, like, now you can just prompt it. But back in the day. Yeah. Yeah. Those were handcraft, handcrafted artisanal code.Nader: Yeah. I was actually really proud of that because, uh, it was an, i I made it in Figma.Yeah. And then I found, I was like really struggling to figure out how to turn it from like Figma to react. So what it actually is, is just an SVG and I, I have all the styles and so when you change the chip, whether it's like active or not it changes the SVG code and that somehow like renders like, looks like it's animating, but it, we just had the transition slow, but it's just like the, a JavaScript function to change the like underlying SVG.Yeah. And that was how I ended up like figuring out how to move it from from Figma. But yeah, that's Art Artisan. [00:05:00]Kyle: Speaking of marketing stunts though, he actually used those SVGs. Or kind of use those SVGs to make these cards.Nader: Oh yeah. LikeKyle: a GPU gift card Yes. That he handed out everywhere. That was actually my first impression of thatNader: one.Yeah,swyx: yeah, yeah.Nader: Yeah.swyx: I think I still have one of them.Nader: They look great.Kyle: Yeah.Nader: I have a ton of them still actually in our garage, which just, they don't have labels. We should honestly like bring, bring them back. But, um, I found this old printing press here, actually just around the corner on Ven ness. And it's a third generation San Francisco shop.And so I come in an excited startup founder trying to like, and they just have this crazy old machinery and I'm in awe. ‘cause the the whole building is so physical. Like you're seeing these machines, they have like pedals to like move these saws and whatever. I don't know what this machinery is, but I saw all three generations.Like there's like the grandpa, the father and the son, and the son was like, around my age. Well,swyx: it's like a holy, holy trinity.Nader: It's funny because we, so I just took the same SVG and we just like printed it and it's foil printing, so they make a a, a mold. That's like an inverse of like the A 100 and then they put the foil on it [00:06:00] and then they press it into the paper.And I remember once we got them, he was like, Hey, don't forget about us. You know, I guess like early Apple and Cisco's first business cards were all made there. And so he was like, yeah, we, we get like the startup businesses but then as they mature, they kind of go somewhere else. And so I actually, I think we were talking with marketing about like using them for some, we should go back and make some cards.swyx: Yeah, yeah, yeah. You know, I remember, you know, as a very, very small breadth investor, I was like, why are we spending time like, doing these like stunts for GPUs? Like, you know, I think like as a, you know, typical like cloud hard hardware person, you go into an AWS you pick like T five X xl, whatever, and it's just like from a list and you look at the specs like, why animate this GP?And, and I, I do think like it just shows the level of care that goes throughout birth and Yeah. And now, and also the, and,Nader: and Nvidia. I think that's what the, the thing that struck me most when we first came in was like the amount of passion that everyone has. Like, I think, um, you know, you talk to, you talk to Kyle, you talk to, like, every VP that I've met at Nvidia goes so close to the metal.Like, I remember it was almost a year ago, and like my VP asked me, he's like, Hey, [00:07:00] what's cursor? And like, are you using it? And if so, why? Surprised at this, and he downloaded Cursor and he was asking me to help him like, use it. And I thought that was, uh, or like, just show him what he, you know, why we were using it.And so, the amount of care that I think everyone has and the passion, appreciate, passion and appreciation for the moment. Right. This is a very unique time. So it's really cool to see everyone really like, uh, appreciate that.swyx: Yeah.Acquisition and DevEx Shiftswyx: One thing I wanted to do before we move over to sort of like research topics and, uh, the, the stuff that Kyle's working on is just tell the story of the acquisition, right?Like, not many people have been, been through an acquisition with Nvidia. What's it like? Uh, what, yeah, just anything you'd like to say.Nader: It's a crazy experience. I think, uh, you know, we were the thing that was the most exciting for us was. Our goal was just to make it easier for developers.We wanted to find access to GPUs, make it easier to do that. And then all, oh, actually your question about launchable. So launchable was just make one click exper, like one click deploys for any software on top of the GPU. Mm-hmm. And so what we really liked about Nvidia was that it felt like we just got a lot more resources to do all of that.I think, uh, you [00:08:00] know, NVIDIA's goal is to make things as easy for developers as possible. So there was a really nice like synergy there. I think that, you know, when it comes to like an acquisition, I think the amount that the soul of the products align, I think is gonna be. Is going speak to the success of the acquisition.Yeah. And so it in many ways feels like we're home. This is a really great outcome for us. Like we you know, I love brev.nvidia.com. Like you should, you should use it's, it's theKyle: front page for GPUs.Nader: Yeah. Yeah. If you want GP views,Kyle: you go there, getswyx: it there, and it's like internally is growing very quickly.I, I don't remember You said some stats there.Nader: Yeah, yeah, yeah. It's, uh, I, I wish I had the exact numbers, but like internally, externally, it's been growing really quickly. We've been working with a bunch of partners with a bunch of different customers and ISVs, if you have a solution that you want someone that runs on the GPU and you want people to use it quickly, we can bundle it up, uh, in a launchable and make it a one click run.If you're doing things and you want just like a sandbox or something to run on, right. Like open claw. Huge moment. Super exciting. Our, uh, and we'll talk into it more, but. You know, internally, people wanna run this, and you, we know we have to be really careful from the security implications. Do we let this run on the corporate network?Security's guidance was, Hey, [00:09:00] run this on breath, it's in, you know, it's, it's, it's a vm, it's sitting in the cloud, it's off the corporate network. It's isolated. And so that's been our stance internally and externally about how to even run something like open call while we figure out how to run these things securely.But yeah,swyx: I think there's also like, you almost like we're the right team at the right time when Nvidia is starting to invest a lot more in developer experience or whatever you call it. Yeah. Uh, UX or I don't know what you call it, like software. Like obviously NVIDIA is always invested in software, but like, there's like, this is like a different audience.Yeah. It's aNader: widerKyle: developer base.swyx: Yeah. Right.Nader: Yeah. Yeah. You know, it's funny, it's like, it's not, uh,swyx: so like, what, what is it called internally? What, what is this that people should be aware that is going on there?Nader: Uh, what, like developer experienceswyx: or, yeah, yeah. Is it's called just developer experience or is there like a broader strategy hereNader: in Nvidia?Um, Nvidia always wants to make a good developer experience. The thing is and a lot of the technology is just really complicated. Like, it's not, it's uh, you know, I think, um. The thing that's been really growing or the AI's growing is having a huge moment, not [00:10:00] because like, let's say data scientists in 2018, were quiet then and are much louder now.The pie is com, right? There's a whole bunch of new audiences. My mom's wondering what she's doing. My sister's learned, like taught herself how to code. Like the, um, you know, I, I actually think just generally AI's a big equalizer and you're seeing a more like technologically literate society, I guess.Like everyone's, everyone's learning how to code. Uh, there isn't really an excuse for that. And so building a good UX means that you really understand who your end user is. And when your end user becomes such a wide, uh, variety of people, then you have to almost like reinvent the practice, right? Yeah. You haveKyle: to, and actually build more developer ux, right?Because the, there are tiers of developer base that were added. You know, the, the hackers that are building on top of open claw, right? For example, have never used gpu. They don't know what kuda is. They, they, they just want to run something.Nader: Yeah.Kyle: You need new UX that is not just. Hey, you know, how do you program something in Cuda and run it?And then, and then we built, you know, like when Deep Learning was getting big, we built, we built Torch and, and, but so recently the amount of like [00:11:00] layers that are added to that developer stack has just exploded because AI has become ubiquitous. Everyone's using it in different ways. Yeah. It'sNader: moving fast in every direction.Vertical, horizontal.Vibhu: Yeah. You guys, you even take it down to hardware, like the DGX Spark, you know, it's, it's basically the same system as just throwing it up on big GPU cluster.Nader: Yeah, yeah, yeah. It's amazing. Blackwell.swyx: Yeah. Uh, we saw the preview at the last year's GTC and that was one of the better performing, uh, videos so far, and video coverage so far.Awesome. This will beat it. Um,Nader: that wasswyx: actually, we have fingersNader: crossed. Yeah.DGX Spark and Remote AccessNader: Even when Grace Blackwell or when, um, uh, DGX Spark was first coming out getting to be involved in that from the beginning of the developer experience. And it just comes back to what youswyx: were involved.Nader: Yeah. St. St.swyx: Mars.Nader: Yeah. Yeah. I mean from, it was just like, I, I got an email, we just got thrown into the loop and suddenly yeah, I, it was actually really funny ‘cause I'm still pretty fresh from the acquisition and I'm, I'm getting an email from a bunch of the engineering VPs about like, the new hardware, GPU chip, like we're, or not chip, but just GPU system that we're putting out.And I'm like, okay, cool. Matters. Now involved with this for the ux, I'm like. What am I gonna do [00:12:00] here? So, I remember the first meeting, I was just like kind of quiet as I was hearing engineering VPs talk about what this box could be, what it could do, how we should use it. And I remember, uh, one of the first ideas that people were idea was like, oh, the first thing that it was like, I think a quote was like, the first thing someone's gonna wanna do with this is get two of them and run a Kubernetes cluster on top of them.And I was like, oh, I think I know why I'm here. I was like, the first thing we're doing is easy. SSH into the machine. And then, and you know, just kind of like scoping it down of like, once you can do that every, you, like the person who wants to run a Kubernetes cluster onto Sparks has a higher propensity for pain, then, then you know someone who buys it and wants to run open Claw right now, right?If you can make sure that that's as effortless as possible, then the rest becomes easy. So there's a tool called Nvidia Sync. It just makes the SSH connection really simple. So, you know, if you think about it like. If you have a Mac, uh, or a PC or whatever, if you have a laptop and you buy this GPU and you want to use it, you should be able to use it like it's A-A-G-P-U in the cloud, right?Um, but there's all this friction of like, how do you actually get into that? That's part of [00:13:00] Revs value proposition is just, you know, there's a CLI that wraps SSH and makes it simple. And so our goal is just get you into that machine really easily. And one thing we just launched at CES, it's in, it's still in like early access.We're ironing out some kinks, but it should be ready by GTC. You can register your spark on Brev. And so now if youswyx: like remote managed yeah, local hardware. Single pane of glass. Yeah. Yeah. Because Brev can already manage other clouds anyway, right?Vibhu: Yeah, yeah. And you use the spark on Brev as well, right?Nader: Yeah. But yeah, exactly. So, so you, you, so you, you set it up at home you can run the command on it, and then it gets it's essentially it'll appear in your Brev account, and then you can take your laptop to a Starbucks or to a cafe, and you'll continue to use your, you can continue use your spark just like any other cloud node on Brev.Yeah. Yeah. And it's just like a pre-provisioned centerswyx: in yourNader: home. Yeah, exactly.swyx: Yeah. Yeah.Vibhu: Tiny little data center.Nader: Tiny little, the size ofVibhu: your phone.SOL Culture and Dynamo Setupswyx: One more thing before we move on to Kyle. Just have so many Jensen stories and I just love, love mining Jensen stories. Uh, my favorite so far is SOL. Uh, what is, yeah, what is S-O-L-S-O-LNader: is actually, i, I think [00:14:00] of all the lessons I've learned, that one's definitely my favorite.Kyle: It'll always stick with you.Nader: Yeah. Yeah. I, you know, in your startup, everything's existential, right? Like we've, we've run out of money. We were like, on the risk of, of losing payroll, we've had to contract our team because we l ran outta money. And so like, um, because of that you're really always forcing yourself to I to like understand the root cause of everything.If you get a date, if you get a timeline, you know exactly why that date or timeline is there. You're, you're pushing every boundary and like, you're not just say, you're not just accepting like a, a no. Just because. And so as you start to introduce more layers, as you start to become a much larger organization, SOL is is essentially like what is the physics, right?The speed of light moves at a certain speed. So if flight's moving some slower, then you know something's in the way. So before trying to like layer reality back in of like, why can't this be delivered at some date? Let's just understand the physics. What is the theoretical limit to like, uh, how fast this can go?And then start to tell me why. ‘cause otherwise people will start telling you why something can't be done. But actually I think any great leader's goal is just to create urgency. Yeah. [00:15:00] There's an infiniteKyle: create compelling events, right?Nader: Yeah.Kyle: Yeah. So l is a term video is used to instigate a compelling event.You say this is done. How do we get there? What is the minimum? As much as necessary, as little as possible thing that it takes for us to get exactly here and. It helps you just break through a bunch of noise.swyx: Yeah.Kyle: Instantly.swyx: One thing I'm unclear about is, can only Jensen use the SOL card? Like, oh, no, no, no.Not everyone get the b******t out because obviously it's Jensen, but like, can someone else be like, no, likeKyle: frontline engineers use it.Nader: Yeah. Every, I think it's not so much about like, get the b******t out. It's like, it's like, give me the root understanding, right? Like, if you tell me something takes three weeks, it like, well, what's the first principles?Yeah, the first principles. It's like, what's the, what? Like why is it three weeks? What is the actual yeah. What's the actual limit of why this is gonna take three weeks? If you're gonna, if you, if let's say you wanted to buy a new computer and someone told you it's gonna be here in five days, what's the SOL?Well, like the SOL is like, I could walk into a Best Buy and pick it up for you. Right? So then anything that's like beyond that is, and is that practical? Is that how we're gonna, you know, let's say give everyone in the [00:16:00] company a laptop, like obviously not. So then like that's the SOL and then it's like, okay, well if we have to get more than 10, suddenly there might be some, right?And so now we can kind of piece the reality back.swyx: So, so this is the. Paul Graham do things that don't scale. Yeah. And this is also the, what people would now call behi agency. Yeah.Kyle: It's actually really interesting because there's a, there's a second hardware angle to SOL that like doesn't come up for all the org sol is used like culturally at aswyx: media for everything.I'm also mining for like, I think that can be annoying sometimes. And like someone keeps going IOO you and you're like, guys, like we have to be stable. We have to, we to f*****g plan. Yeah.Kyle: It's an interesting balance.Nader: Yeah. I encounter that with like, actually just with, with Alec, right? ‘cause we, we have a new conference so we need to launch, we have, we have goals of what we wanna launch by, uh, by the conference and like, yeah.At the end of the day, where isswyx: this GTC?Nader: Um, well this is like, so we, I mean we did it for CES, we did for GT CDC before that we're doing it for GTC San Jose. So I mean, like every, you know, we have a new moment. Um, and we want to launch something. Yeah. And we want to do so at SOL and that does mean that some, there's some level of prioritization that needs [00:17:00] to happen.And so it, it is difficult, right? I think, um, you have to be careful with what you're pushing. You know, stability is important and that should be factored into S-O-L-S-O-L isn't just like, build everything and let it break, you know, that, that's part of the conversation. So as you're laying, layering in all the details, one of them might be, Hey, we could build this, but then it's not gonna be stable for X, y, z reasons.And so that was like, one of our conversations for CES was, you know, hey, like we, we can get this into early access registering your spark with brev. But there are a lot of things that we need to do in order to feel really comfortable from a security perspective, right? There's a lot of networking involved before we deliver that to users.So it's like, okay. Let's get this to a point where we can at least let people experiment with it. We had it in a booth, we had it in Jensen's keynote, and then let's go iron out all the networking kinks. And that's not easy. And so, uh, that can come later. And so that was the way that we layered that back in.Yeah. ButKyle: It's not really about saying like, you don't have to do the, the maintenance or operational work. It's more about saying, you know, it's kind of like [00:18:00] highlights how progress is incremental, right? Like, what is the minimum thing that we can get to. And then there's SOL for like every component after that.But there's the SOL to get you, get you to the, the starting line. And that, that's usually how it's asked. Yeah. On the other side, you know, like SOL came out of like hardware at Nvidia. Right. So SOL is like literally if we ran the accelerator or the GPU with like at basically full speed with like no other constraints, like how FAST would be able to make a program go.swyx: Yeah. Yeah. Right.Kyle: Soswyx: in, in training that like, you know, then you work back to like some percentage of like MFU for example.Kyle: Yeah, that's a, that's a great example. So like, there's an, there's an S-O-L-M-F-U, and then there's like, you know, what's practically achievable.swyx: Cool. Should we move on to sort of, uh, Kyle's side?Uh, Kyle, you're coming more from the data science world. And, uh, I, I mean I always, whenever, whenever I meet someone who's done working in tabular stuff, graph neural networks, time series, these are basically when I go to new reps, I go to ICML, I walk the back halls. There's always like a small group of graph people.Yes. Absolute small group of tabular people. [00:19:00] And like, there's no one there. And like, it's very like, you know what I mean? Like, yeah, no, like it's, it's important interesting work if you care about solving the problems that they solve.Kyle: Yeah.swyx: But everyone else is just LMS all the time.Kyle: Yeah. I mean it's like, it's like the black hole, right?Has the event horizon reached this yet in nerves? Um,swyx: but like, you know, those are, those are transformers too. Yeah. And, and those are also like interesting things. Anyway, uh, I just wanted to spend a little bit of time on, on those, that background before we go into Dynamo, uh, proper.Kyle: Yeah, sure. I took a different path to Nvidia than that, or I joined six years ago, seven, if you count, when I was an intern.So I joined Nvidia, like right outta college. And the first thing I jumped into was not what I'd done in, during internship, which was like, you know, like some stuff for autonomous vehicles, like heavyweight object detection. I jumped into like, you know, something, I'm like, recommenders, this is popular. Andswyx: yeah, he did RexiKyle: as well.Yeah, Rexi. Yeah. I mean that, that was the taboo data at the time, right? You have tables of like, audience qualities and item qualities, and you're trying to figure out like which member of [00:20:00] the audience matches which item or, or more practically which item matches which member of the audience. And at the time, really it was like we were trying to enable.Uh, recommender, which had historically been like a little bit of a CP based workflow into something that like, ran really well in GPUs. And it's since been done. Like there are a bunch of libraries for Axis that run on GPUs. Uh, the common models like Deeplearning recommendation model, which came outta meta and the wide and deep model, which was used or was released by Google were very accelerated by GPUs using, you know, the fast HBM on the chips, especially to do, you know, vector lookups.But it was very interesting at the time and super, super relevant because like we were starting to get like. This explosion of feeds and things that required rec recommenders to just actively be on all the time. And sort of transitioned that a little bit towards graph neural networks when I discovered them because I was like, okay, you can actually use graphical neural networks to represent like, relationships between people, items, concepts, and that, that interested me.So I jumped into that at [00:21:00] Nvidia and, and got really involved for like two-ish years.swyx: Yeah. Uh, and something I learned from Brian Zaro Yeah. Is that you can just kind of choose your own path in Nvidia.Kyle: Oh my God. Yeah.swyx: Which is not a normal big Corp thing. Yeah. Like you, you have a lane, you stay in your lane.Nader: I think probably the reason why I enjoy being in a, a big company, the mission is the boss probably from a startup guy. Yeah. The missionswyx: is the boss.Nader: Yeah. Uh, it feels like a big game of pickup basketball. Like, you know, if you play one, if you wanna play basketball, you just go up to the court and you're like, Hey look, we're gonna play this game and we need three.Yeah. And you just like find your three. That's honestly for every new initiative that's what it feels like. Yeah.Vibhu: It also like shows, right? Like Nvidia. Just releasing state-of-the-art stuff in every domain. Yeah. Like, okay, you expect foundation models with Nemo tron voice just randomly parakeet.Call parakeet just comes out another one, uh, voice. TheKyle: video voice team has always been producing.Vibhu: Yeah. There's always just every other domain of paper that comes out, dataset that comes out. It's like, I mean, it also stems back to what Nvidia has to do, right? You have to make chips years before they're actually produced.Right? So you need to know, you need to really [00:22:00] focus. TheKyle: design process starts likeVibhu: exactlyKyle: three to five years before the chip gets to the market.Vibhu: Yeah. I, I'm curious more about what that's like, right? So like, you have specialist teams. Is it just like, you know, people find an interest, you go in, you go deep on whatever, and that kind of feeds back into, you know, okay, we, we expect predictions.Like the internals at Nvidia must be crazy. Right? You know? Yeah. Yeah. You know, you, you must. Not even without selling to people, you have your own predictions of where things are going. Yeah. And they're very based, very grounded. Right?Kyle: Yeah. It, it, it's really interesting. So there's like two things that I think that Amed does, which are quite interesting.Uh, one is like, we really index into passion. There's a big. Sort of organizational top sound push to like ensure that people are working on the things that they're passionate about. So if someone proposes something that's interesting, many times they can just email someone like way up the chain that they would find this relevant and say like, Hey, can I go work on this?Nader: It's actually like I worked at a, a big company for a couple years before, uh, starting on my startup journey and like, it felt very weird if you were to like email out of chain, if that makes [00:23:00] sense. Yeah. The emails at Nvidia are like mosh pitsswyx: shoot,Nader: and it's just like 60 people, just whatever. And like they're, there's this,swyx: they got messy like, reply all you,Nader: oh, it's in, it's insane.It's insane. They justKyle: help. You know, Maxim,Nader: the context. But, but that's actually like, I've actually, so this is a weird thing where I used to be like, why would we send emails? We have Slack. I am the entire, I'm the exact opposite. I feel so bad for anyone who's like messaging me on Slack ‘cause I'm so unresponsive.swyx: Your emailNader: Maxi, email Maxim. I'm email maxing Now email is a different, email is perfect because man, we can't work together. I'm email is great, right? Because important threads get bumped back up, right? Yeah, yeah. Um, and so Slack doesn't do that. So I just have like this casino going off on the right or on the left and like, I don't know which thread was from where or what, but like the threads get And then also just like the subject, so you can have like working threads.I think what's difficult is like when you're small, if you're just not 40,000 people I think Slack will work fine, but there's, I don't know what the inflection point is. There is gonna be a point where that becomes really messy and you'll actually prefer having email. ‘cause you can have working threads.You can cc more than nine people in a thread.Kyle: You can fork stuff.Nader: You can [00:24:00] fork stuff, which is super nice and just like y Yeah. And so, but that is part of where you can propose a plan. You can also just. Start, honestly, momentum's the only authority, right? So like, if you can just start, start to make a little bit of progress and show someone something, and then they can try it.That's, I think what's been, you know, I think the most effective way to push anything for forward. And that's both at Nvidia and I think just generally.Kyle: Yeah, there's, there's the other concept that like is explored a lot at Nvidia, which is this idea of a zero billion dollar business. Like market creation is a big thing at Nvidia.Like,swyx: oh, you want to go and start a zero billion dollar business?Kyle: Jensen says, we are completely happy investing in zero billion dollar markets. We don't care if this creates revenue. It's important for us to know about this market. We think it will be important in the future. It can be zero billion dollars for a while.I'm probably minging as words here for, but like, you know, like, I'll give an example. NVIDIA's been working on autonomous driving for a a long time,swyx: like an Nvidia car.Kyle: No, they, they'veVibhu: used the Mercedes, right? They're around the HQ and I think it finally just got licensed out. Now they're starting to be used quite a [00:25:00] bit.For 10 years you've been seeing Mercedes with Nvidia logos driving.Kyle: If you're in like the South San Santa Clara, it's, it's actually from South. Yeah. So, um. Zero billion dollar markets are, are a thing like, you know, Jensen,swyx: I mean, okay, look, cars are not a zero billion dollar market. But yeah, that's a bad example.Nader: I think, I think he's, he's messaging, uh, zero today, but, or even like internally, right? Like, like it's like, uh, an org doesn't have to ruthlessly find revenue very quickly to justify their existence. Right. Like a lot of the important research, a lot of the important technology being developed that, that's kind ofKyle: where research, research is very ide ideologically free at Nvidia.Yeah. Like they can pursue things that they wereswyx: Were you research officially?Kyle: I was never in research. Officially. I was always in engineering. Yeah. We in, I'm in an org called Deep Warning Algorithms, which is basically just how do we make things that are relevant to deep warning go fast.swyx: That sounds freaking cool.Vibhu: And I think a lot of that is underappreciated, right? Like time series. This week Google put out time. FF paper. Yeah. A new time series, paper res. Uh, Symantec, ID [00:26:00] started applying Transformers LMS to Yes. Rec system. Yes. And when you think the scale of companies deploying these right. Amazon recommendations, Google web search, it's like, it's huge scale andKyle: Yeah.Vibhu: You want fast?Kyle: Yeah. Yeah. Yeah. Actually it's, it, I, there's a fun moment that brought me like full circle. Like, uh, Amazon Ads recently gave a talk where they talked about using Dynamo for generative recommendation, which was like super, like weirdly cathartic for me. I'm like, oh my God. I've, I've supplanted what I was working on.Like, I, you're using LMS now to do what I was doing five years ago.swyx: Yeah. Amazing. And let's go right into Dynamo. Uh, maybe introduce Yeah, sure. To the top down and Yeah.Kyle: I think at this point a lot of people are familiar with the term of inference. Like funnily enough, like I went from, you know, inference being like a really niche topic to being something that's like discussed on like normal people's Twitter feeds.It's,Nader: it's on billboardsKyle: here now. Yeah. Very, very strange. Driving, driving, seeing just an inference ad on 1 0 1 inference at scale is becoming a lot more important. Uh, we have these moments like, you know, open claw where you have these [00:27:00] agents that take lots and lots of tokens, but produce, incredible results.There are many different aspects of test time scaling so that, you know, you can use more inference to generate a better result than if you were to use like a short amount of inference. There's reasoning, there's quiring, there's, adding agency to the model, allowing it to call tools and use skills.Dyno sort came about at Nvidia. Because myself and a couple others were, were sort of talking about the, these concepts that like, you know, you have inference engines like VLMS, shelan, tenor, TLM and they have like one single copy. They, they, they sort of think about like things as like one single copy, like one replica, right?Why Scale Out WinsKyle: Like one version of the model. But when you're actually serving things at scale, you can't just scale up that replica because you end up with like performance problems. There's a scaling limit to scaling up replicas. So you actually have to scale out to use a, maybe some Kubernetes type terminology.We kind of realized that there was like. A lot of potential optimization that we could do in scaling out and building systems for data [00:28:00] center scale inference. So Dynamo is this data center scale inference engine that sits on top of the frameworks like VLM Shilling and 10 T lm and just makes things go faster because you can leverage the economy of scale.The fact that you have KV cash, which we can define a little bit later, uh, in all these machines that is like unique and you wanna figure out like the ways to maximize your cash hits or you want to employ new techniques in inference like disaggregation, which Dynamo had introduced to the world in, in, in March, not introduced, it was a academic talk, but beforehand.But we are, you know, one of the first frameworks to start, supporting it. And we wanna like, sort of combine all these techniques into sort of a modular framework that allows you to. Accelerate your inference at scale.Nader: By the way, Kyle and I became friends on my first date, Nvidia, and I always loved, ‘cause like he always teaches meswyx: new things.Yeah. By the way, this is why I wanted to put two of you together. I was like, yeah, this is, this is gonna beKyle: good. It's very, it's very different, you know, like we've, we, we've, we've talked to each other a bunch [00:29:00] actually, you asked like, why, why can't we scale up?Nader: Yeah.Scale Up Limits ExplainedNader: model, you said model replicas.Kyle: Yeah. So you, so scale up means assigning moreswyx: heavier?Kyle: Yeah, heavier. Like making things heavier. Yeah, adding more GPUs. Adding more CPUs. Scale out is just like having a barrier saying, I'm gonna duplicate my representation of the model or a representation of this microservice or something, and I'm gonna like, replicate it Many times.Handle, load. And the reason that you can't scale, scale up, uh, past some points is like, you know, there, there, there are sort of hardware bounds and algorithmic bounds on, on that type of scaling. So I'll give you a good example that's like very trivial. Let's say you're on an H 100. The Maxim ENV link domain for H 100, for most Ds H one hundreds is heus, right?So if you scaled up past that, you're gonna have to figure out ways to handle the fact that now for the GPUs to communicate, you have to do it over Infin band, which is still very fast, but is not as fast as ENV link.swyx: Is it like one order of magnitude, like hundreds or,Kyle: it's about an order of magnitude?Yeah. Okay. Um, soswyx: not terrible.Kyle: [00:30:00] Yeah. I, I need to, I need to remember the, the data sheet here, like, I think it's like about 500 gigabytes. Uh, a second unidirectional for ENV link, and about 50 gigabytes a second unidirectional for Infin Band. I, it, it depends on the, the generation.swyx: I just wanna set this up for people who are not familiar with these kinds of like layers and the trash speedVibhu: and all that.Of course.From Laptop to Multi NodeVibhu: Also, maybe even just going like a few steps back before that, like most people are very familiar with. You see a, you know, you can use on your laptop, whatever these steel viol, lm you can just run inference there. All, there's all, you can, youcan run it on thatVibhu: laptop. You can run on laptop.Then you get to, okay, uh, models got pretty big, right? JLM five, they doubled the size, so mm-hmm. Uh, what do you do when you have to go from, okay, I can get 128 gigs of memory. I can run it on a spark. Then you have to go multi GPU. Yeah. Okay. Multi GPU, there's some support there. Now, if I'm a company and I don't have like.I'm not hiring the best researchers for this. Right. But I need to go [00:31:00] multi-node, right? I have a lot of servers. Okay, now there's efficiency problems, right? You can have multiple eight H 100 nodes, but, you know, is that as a, like, how do you do that efficiently?Kyle: Yeah. How do you like represent them? How do you choose how to represent the model?Yeah, exactly right. That's a, that's like a hard question. Everyone asks, how do you size oh, I wanna run GLM five, which just came out new model. There have been like four of them in the past week, by the way, like a bunch of new models.swyx: You know why? Right? Deep seek.Kyle: No comment. Oh. Yeah, but Ggl, LM five, right?We, we have this, new model. It's, it's like a large size, and you have to figure out how to both scale up and scale out, right? Because you have to find the right representation that you care about. Everyone does this differently. Let's be very clear. Everyone figures this out in their own path.Nader: I feel like a lot of AI or ML even is like, is like this. I think people think, you know, I, I was, there was some tweet a few months ago that was like, why hasn't fine tuning as a service taken off? You know, that might be me. It might have been you. Yeah. But people want it to be such an easy recipe to follow.But even like if you look at an ML model and specificKyle: to you Yeah,Nader: yeah.Kyle: And the [00:32:00] model,Nader: the situation, and there's just so much tinkering, right? Like when you see a model that has however many experts in the ME model, it's like, why that many experts? I don't, they, you know, they tried a bunch of things and that one seemed to do better.I think when it comes to how you're serving inference, you know, you have a bunch of decisions to make and there you can always argue that you can take something and make it more optimal. But I think it's this internal calibration and appetite for continued calibration.Vibhu: Yeah. And that doesn't mean like, you know, people aren't taking a shot at this, like tinker from thinking machines, you know?Yeah. RL as a service. Yeah, totally. It's, it also gets even harder when you try to do big model training, right? We're not the best at training Moes, uh, when they're pre-trained. Like we saw this with LAMA three, right? They're trained in such a sparse way that meta knows there's gonna be a bunch of inference done on these, right?They'll open source it, but it's very trained for what meta infrastructure wants, right? They wanna, they wanna inference it a lot. Now the question to basically think about is, okay, say you wanna serve a chat application, a coding copilot, right? You're doing a layer of rl, you're serving a model for X amount of people.Is it a chat model, a coding model? Dynamo, you know, back to that,Kyle: it's [00:33:00] like, yeah, sorry. So you we, we sort of like jumped off of, you know, jumped, uh, on that topic. Everyone has like, their own, own journey.Cost Quality Latency TradeoffsKyle: And I, I like to think of it as defined by like, what is the model you need? What is the accuracy you need?Actually I talked to NA about this earlier. There's three axes you care about. What is the quality that you're able to produce? So like, are you accurate enough or can you complete the task with enough, performance, high enough performance. Yeah, yeah. Uh, there's cost. Can you serve the model or serve your workflow?Because it's not just the model anymore, it's the workflow. It's the multi turn with an agent cheaply enough. And then can you serve it fast enough? And we're seeing all three of these, like, play out, like we saw, we saw new models from OpenAI that you know, are faster. You have like these new fast versions of models.You can change the amount of thinking to change the amount of quality, right? Produce more tokens, but at a higher cost in a, in a higher latency. And really like when you start this journey of like trying to figure out how you wanna host a model, you, you, you think about three things. What is the model I need to serve?How many times do I need to call it? What is the input sequence link was [00:34:00] the, what does the workflow look like on top of it? What is the SLA, what is the latency SLA that I need to achieve? Because there's usually some, this is usually like a constant, you, you know, the SLA that you need to hit and then like you try and find the lowest cost version that hits all of these constraints.Usually, you know, you, you start with those things and you say you, you kind of do like a bit of experimentation across some common configurations. You change the tensor parallel size, which is a form of parallelismVibhu: I take, it goes even deeper first. Gotta think what model.Kyle: Yes, course,ofKyle: course. It's like, it's like a multi-step design process because as you said, you can, you can choose a smaller model and then do more test time scaling and it'll equate the quality of a larger model because you're doing the test time scaling or you're adding a harness or something.So yes, it, it goes way deeper than that. But from the performance perspective, like once you get to the model you need, you need to host, you look at that and you say, Hey. I have this model, I need to serve it at the speed. What is the right configuration for that?Nader: You guys see the recent, uh, there was a paper I just saw like a few days ago that, uh, if you run [00:35:00] the same prompt twice, you're getting like double Just try itagain.Nader: Yeah, exactly.Vibhu: And you get a lot. Yeah. But the, the key thing there is you give the context of the failed try, right? Yeah. So it takes a shot. And this has been like, you know, basic guidance for quite a while. Just try again. ‘cause you know, trying, just try again. Did you try again? All adviceNader: in life.Vibhu: Just, it's a paper from Google, if I'm not mistaken, right?Yeah,Vibhu: yeah. I think it, it's like a seven bas little short paper. Yeah. Yeah. The title's very cute. And it's just like, yeah, just try again. Give it ask context,Kyle: multi-shot. You just like, say like, hey, like, you know, like take, take a little bit more, take a little bit more information, try and fail. Fail.Vibhu: And that basic concept has gone pretty deep.There's like, um, self distillation, rl where you, you do self distillation, you do rl and you have past failure and you know, that gives some signal so people take, try it again. Not strong enough.swyx: Uh, for, for listeners, uh, who listen to here, uh, vivo actually, and I, and we run a second YouTube channel for our paper club where, oh, that's awesome.Vivo just covered this. Yeah. Awesome. Self desolation and all that's, that's why he, to speed [00:36:00] on it.Nader: I'll to check it out.swyx: Yeah. It, it's just a good practice, like everyone needs, like a paper club where like you just read papers together and the social pressure just kind of forces you to just,Nader: we, we,there'sNader: like a big inference.Kyle: ReadingNader: group at a video. I feel so bad every time. I I, he put it on like, on our, he shared it.swyx: One, one ofNader: your guys,swyx: uh, is, is big in that, I forget es han Yeah, yeah,Kyle: es Han's on my team. Actually. Funny. There's a, there's a, there's a employee transfer between us. Han worked for Nater at Brev, and now he, he's on my team.He wasNader: our head of ai. And then, yeah, once we got in, andswyx: because I'm always looking for like, okay, can, can I start at another podcast that only does that thing? Yeah. And, uh, Esan was like, I was trying to like nudge Esan into like, is there something here? I mean, I don't think there's, there's new infant techniques every day.So it's like, it's likeKyle: you would, you would actually be surprised, um, the amount of blog posts you see. And ifswyx: there's a period where it was like, Medusa hydra, what Eagle, like, youKyle: know, now we have new forms of decode, uh, we have new forms of specula, of decoding or new,swyx: what,Kyle: what are youVibhu: excited? And it's exciting when you guys put out something like Tron.‘cause I remember the paper on this Tron three, [00:37:00] uh, the amount of like post train, the on tokens that the GPU rich can just train on. And it, it was a hybrid state space model, right? Yeah.Kyle: It's co-designed for the hardware.Vibhu: Yeah, go design for the hardware. And one of the things was always, you know, the state space models don't scale as well when you do a conversion or whatever the performance.And you guys are like, no, just keep draining. And Nitron shows a lot of that. Yeah.Nader: Also, something cool about Nitron it was released in layers, if you will, very similar to Dynamo. It's, it's, it's essentially it was released as you can, the pre-training, post-training data sets are released. Yeah. The recipes on how to do it are released.The model itself is released. It's full model. You just benefit from us turning on the GPUs. But there are companies like, uh, ServiceNow took the dataset and they trained their own model and we were super excited and like, you know, celebrated that work.ZoomVibhu: different. Zoom is, zoom is CGI, I think, uh, you know, also just to add like a lot of models don't put out based models and if there's that, why is fine tuning not taken off?You know, you can do your own training. Yeah,Kyle: sure.Vibhu: You guys put out based model, I think you put out everything.Nader: I believe I know [00:38:00]swyx: about base. BasicallyVibhu: without baseswyx: basic can be cancelable.Vibhu: Yeah. Base can be cancelable.swyx: Yeah.Vibhu: Safety training.swyx: Did we get a full picture of dymo? I, I don't know if we, what,Nader: what I'd love is you, you mentioned the three axes like break it down of like, you know, what's prefilled decode and like what are the optimizations that we can get with Dynamo?Kyle: Yeah. That, that's, that's, that's a great point. So to summarize on that three axis problem, right, there are three things that determine whether or not something can be done with inference, cost, quality, latency, right? Dynamo is supposed to be there to provide you like the runtime that allows you to pull levers to, you know, mix it up and move around the parade of frontier or the preto surface that determines is this actually possible with inference And AI todayNader: gives you the knobs.Kyle: Yeah, exactly. It gives you the knobs.Disaggregation Prefill vs DecodeKyle: Uh, and one thing that like we, we use a lot in contemporary inference and is, you know, starting to like pick up from, you know, in, in general knowledge is this co concept of disaggregation. So historically. Models would be hosted with a single inference engine. And that inference engine [00:39:00] would ping pong between two phases.There's prefill where you're reading the sequence generating KV cache, which is basically just a set of vectors that represent the sequence. And then using that KV cache to generate new tokens, which is called Decode. And some brilliant researchers across multiple different papers essentially made the realization that if you separate these two phases, you actually gain some benefits.Those benefits are basically a you don't have to worry about step synchronous scheduling. So the way that an inference engine works is you do one step and then you finish it, and then you schedule, you start scheduling the next step there. It's not like fully asynchronous. And the problem with that is you would have, uh, essentially pre-fill and decode are, are actually very different in terms of both their resource requirements and their sometimes their runtime.So you would have like prefill that would like block decode steps because you, you'd still be pre-filing and you couldn't schedule because you know the step has to end. So you remove that scheduling issue and then you also allow you, or you yourself, to like [00:40:00] split the work into two different ki types of pools.So pre-fill typically, and, and this changes as, as model architecture changes. Pre-fill is, right now, compute bound most of the time with the sequence is sufficiently long. It's compute bound. On the decode side because you're doing a full Passover, all the weights and the entire sequence, every time you do a decode step and you're, you don't have the quadratic computation of KV cache, it's usually memory bound because you're retrieving a linear amount of memory and you're doing a linear amount of compute as opposed to prefill where you retrieve a linear amount of memory and then use a quadratic.You know,Nader: it's funny, someone exo Labs did a really cool demo where for the DGX Spark, which has a lot more compute, you can do the pre the compute hungry prefill on a DG X spark and then do the decode on a, on a Mac. Yeah. And soVibhu: that's faster.Nader: Yeah. Yeah.Kyle: So you could, you can do that. You can do machine strat stratification.Nader: Yeah.Kyle: And like with our future generation generations of hardware, we actually announced, like with Reuben, this [00:41:00] new accelerator that is prefilled specific. It's called Reuben, CPX. SoKubernetes Scaling with GroveNader: I have a question when you do the scale out. Yeah. Is scaling out easier with Dynamo? Because when you need a new node, you can dedicate it to either the Prefill or, uh, decode.Kyle: Yeah. So Dynamo actually has like a, a Kubernetes component in it called Grove that allows you to, to do this like crazy scaling specialization. It has like this hot, it's a representation that, I don't wanna go too deep into Kubernetes here, but there was a previous way that you would like launch multi-node work.Uh, it's called Leader Worker Set. It's in the Kubernetes standard, and Leader worker set is great. It served a lot of people super well for a long period of time. But one of the things that it's struggles with is representing a set of cases where you have a multi-node replica that has a pair, right?You know, prefill and decode, or it's not paired, but it has like a second stage that has a ratio that changes over time. And prefill and decode are like two different things as your workload changes, right? The amount of prefill you'll need to do may change. [00:42:00] The amount of decode that you, you'll need to do might change, right?Like, let's say you start getting like insanely long queries, right? That probably means that your prefill scales like harder because you're hitting these, this quadratic scaling growth.swyx: Yeah.And then for listeners, like prefill will be long input. Decode would be long output, for example, right?Kyle: Yeah. So like decode, decode scale. I mean, decode is funny because the amount of tokens that you produce scales with the output length, but the amount of work that you do per step scales with the amount of tokens in the context.swyx: Yes.Kyle: So both scales with the input and the output.swyx: That's true.Kyle: But on the pre-fold view code side, like if.Suddenly, like the amount of work you're doing on the decode side stays about the same or like scales a little bit, and then the prefilled side like jumps up a lot. You actually don't want that ratio to be the same. You want it to change over time. So Dynamo has a set of components that A, tell you how to scale.It tells you how many prefilled workers and decoded workers you, it thinks you should have, and also provides a scheduling API for Kubernetes that allows you to actually represent and affect this scheduling on, on, on your actual [00:43:00] hardware, on your compute infrastructure.Nader: Not gonna lie. I feel a little embarrassed for being proud of my SVG function earlier.swyx: No, itNader: wasreallyKyle: cute. I, Iswyx: likeNader: it's all,swyx: it's all engineering. It's all engineering. Um, that's where I'mKyle: technical.swyx: One thing I'm, I'm kind of just curious about with all with you see at a systems level, everything going on here. Mm-hmm. And we, you know, we're scaling it up in, in multi, in distributed systems.Context Length and Co Designswyx: Um, I think one thing that's like kind of, of the moment right now is people are asking, is there any SOL sort of upper bounds. In terms of like, let's call, just call it context length for one for of a better word, but you can break it down however you like.Nader: Yeah.swyx: I just think like, well, yeah, I mean, like clearly you can engage in hybrid architectures and throw in some state space models in there.All, all you want, but it looks, still looks very attention heavy.Kyle: Yes. Uh, yeah. Long context is attention heavy. I mean, we have these hybrid models, um,swyx: to take and most, most models like cap out at a million contexts and that's it. Yeah. Like for the last two years has been it.Kyle: Yeah. The model hardware context co-design thing that we're seeing these days is actually super [00:44:00] interesting.It's like my, my passion, like my secret side passion. We see models like Kimmy or G-P-T-O-S-S. I'm use these because I, I know specific things about these models. So Kimmy two comes out, right? And it's an interesting model. It's like, like a deep seek style architecture is MLA. It's basically deep seek, scaled like a little bit differently, um, and obviously trained differently as well.But they, they talked about, why they made the design choices for context. Kimmy has more experts, but fewer attention heads, and I believe a slightly smaller attention, uh, like dimension. But I need to remember, I need to check that. Uh, it doesn't matter. But they discussed this actually at length in a blog post on ji, which is like our pu which is like credit puswyx: Yeah.Kyle: Um, in, in China. Chinese red.swyx: Yeah.Kyle: It's, yeah. So it, it's, it's actually an incredible blog post. Uh, like all the mls people in, in, in that, I've seen that on GPU are like very brilliant, but they, they talk about like the creators of Kimi K two [00:45:00] actually like, talked about it on, on, on there in the blog post.And they say, we, we actually did an experiment, right? Attention scales with the number of heads, obviously. Like if you have 64 heads versus 32 heads, you do half the work of attention. You still scale quadratic, but you do half the work. And they made a, a very specific like. Sort of barter in their system, in their architecture, they basically said, Hey, what if we gave it more experts, so we're gonna use more memory capacity.But we keep the amount of activated experts the same. We increase the expert sparsity, so we have fewer experts act. The ratio to of experts activated to number of experts is smaller, and we decrease the number of attention heads.Vibhu: And kind of for context, what the, what we had been seeing was you make models sparser instead.So no one was really touching heads. You're just having, uh,Kyle: well, they, they did, they implicitly made it sparser.Vibhu: Yeah, yeah. For, for Kimmy. They did,Kyle: yes.Vibhu: They also made it sparser. But basically what we were seeing was people were at the level of, okay, there's a sparsity ratio. You want more total parameters, less active, and that's sparsity.[00:46:00]But what you see from papers, like, the labs like moonshot deep seek, they go to the level of, okay, outside of just number of experts, you can also change how many attention heads and less attention layers. More attention. Layers. Layers, yeah. Yes, yes. So, and that's all basically coming back to, just tied together is like hardware model, co-design, which isKyle: hardware model, co model, context, co-design.Vibhu: Yeah.Kyle: Right. Like if you were training a, a model that was like. Really, really short context, uh, or like really is good at super short context tasks. You may like design it in a way such that like you don't care about attention scaling because it hasn't hit that, like the turning point where like the quadratic curve takes over.Nader: How do you consider attention or context as a separate part of the co-design? Like I would imagine hardware or just how I would've thought of it is like hardware model. Co-design would be hardware model context co-designKyle: because the harness and the context that is produced by the harness is a part of the model.Once it's trained in,Vibhu: like even though towards the end you'll do long context, you're not changing architecture through I see. Training. Yeah.Kyle: I mean you can try.swyx: You're saying [00:47:00] everyone's training the harness into the model.Kyle: I would say to some degree, orswyx: there's co-design for harness. I know there's a small amount, but I feel like not everyone has like gone full send on this.Kyle: I think, I think I think it's important to internalize the harness that you think the model will be running. Running into the model.swyx: Yeah. Interesting. Okay. Bash is like the universal harness,Kyle: right? Like I'll, I'll give. An example here, right? I mean, or just like a, like a, it's easy proof, right? If you can train against a harness and you're using that harness for everything, wouldn't you just train with the harness to ensure that you get the best possible quality out of,swyx: Well, the, uh, I, I can provide a counter argument.Yeah, sure. Which is what you wanna provide a generally useful model for other people to plug into their harnesses, right? So if youKyle: Yeah. Harnesses can be open, open source, right?swyx: Yeah. So I mean, that's, that's effectively what's happening with Codex.Kyle: Yeah.swyx: And, but like you may want like a different search tool and then you may have to name it differently or,Nader: I don't know how much people have pushed on this, but can you.Train a model, would it be, have you have people compared training a model for the for the harness versus [00:48:00] like post training forswyx: I think it's the same thing. It's the same thing. It's okay. Just extra post training. INader: see.swyx: And so, I mean, cognition does this course, it does this where you, you just have to like, if your tool is slightly different, um, either force your tool to be like the tool that they train for.Hmm. Or undo their training for their tool and then Oh, that's re retrain. Yeah. It's, it's really annoying and like,Kyle: I would hope that eventually we hit like a certain level of generality with respect to training newswyx: tools. This is not a GI like, it's, this is a really stupid like. Learn my tool b***h.Like, I don't know if, I don't know if I can say that, but like, you know, um, I think what my point kind of is, is that there's, like, I look at slopes of the scaling laws and like, this slope is not working, man. We, we are at a million token con

Grow faster than 99% of Etsy shops

Ratboi, Clop, Oracle, svgs, Impact Solutions, The Pentagon, Open AI, Josh Marpet, and More on this edition of the Security Weekly News. Visit https://www.securityweekly.com/swn for all the latest episodes! Show Notes: https://securityweekly.com/swn-517

Ratboi, Clop, Oracle, svgs, Impact Solutions, The Pentagon, Open AI, Josh Marpet, and More on this edition of the Security Weekly News. Show Notes: https://securityweekly.com/swn-517

Ratboi, Clop, Oracle, svgs, Impact Solutions, The Pentagon, Open AI, Josh Marpet, and More on this edition of the Security Weekly News. Visit https://www.securityweekly.com/swn for all the latest episodes! Show Notes: https://securityweekly.com/swn-517

Ratboi, Clop, Oracle, svgs, Impact Solutions, The Pentagon, Open AI, Josh Marpet, and More on this edition of the Security Weekly News. Show Notes: https://securityweekly.com/swn-517

In this special episode of NFT Morning, John welcome Amanda Terry, co-founder of OnChain Monkey and MetaGood, for a deep and inspiring conversation about the future of art on Bitcoin.Amanda recounts the visionary journey of OnChain Monkey, which started on Ethereum in 2021 with a technically groundbreaking mint: 10,000 SVGs inscribed in a single transaction. Early on, the team showed a strong commitment to sustainability and decentralization. This spirit led them to become the first full 10K collection inscribed as Ordinals on Bitcoin, using parent-child provenance.The project's ambition goes beyond PFPs. Amanda shares their mission to advance the art on Bitcoin movement—with a clear cultural, technological, and even humanitarian vision. The conversation also highlights exciting artist collaborations (like Alexis André, Patrick Amadon, and Bryan Brinkman), the launch of their curated Ordinals marketplace Osura, and their ongoing outreach to museums and institutions. Their belief: Ordinals are the future of digital heritage—immutable, enduring, and deeply Bitcoin-native.With Amanda's contagious optimism and a sharp strategic vision, this episode is a must-listen for anyone curious about Bitcoin's artistic renaissance and the pioneers leading the charge.More details here: * OnChainMonkey official website* Osura marketplace official website* OnChainMonkey Twitter account* Amanda Terry's LinkedIn profile* video replay of this episode This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.nftmorning.com

Accessibility and sustainability collide in this eye-opening conversation with Heather Neff, CEO of Equivalent. She returns to AXSChat to share groundbreaking developments in their mission to transform static graphics into responsive, accessible code.What if every image online could instantly adapt to your device settings, work seamlessly with screen readers, and consume a fraction of the energy? That's exactly what Equivalent is achieving through their revolutionary approach to Scalable Vector Graphics (SVGs). Unlike traditional pixel-based images that remain fixed and deteriorate when enlarged, these "smart SVGs" behave more like webpages - responding to dark mode, custom color settings, and accessibility preferences without requiring additional effort from users or developers.The implications are profound. With AI systems generating 34 million images daily and users uploading 64,000 images every second, we're facing an unprecedented accessibility crisis. Equivalent's technology addresses this by dramatically reducing file sizes (from megabytes to kilobytes) while ensuring graphics can be navigated by screen readers with custom reading orders. The sustainability benefits are equally impressive - their dark mode implementation alone can reduce energy consumption by 28% and extend mobile battery life by up to 63% on OLED displays.The small startup is now preparing to launch an API that will allow enterprises to batch-convert thousands of images, ensuring both brand consistency and accessibility compliance without requiring specialized knowledge from design or engineering teams. Their vision extends to a future where all imagery becomes code-based, delivering a web experience that's simultaneously more inclusive, more performant, and more environmentally responsible.Are you interested in helping shape this revolution in web graphics? Equivalent is currently seeking beta testers for their API and connecting with angel investors who share their passion for accessibility and sustainability. Visit their website to try their current tools with a free two-week trial and witness firsthand how smart SVGs are poised to transform our digital experience.Support the showFollow axschat on social media.Bluesky:Antonio https://bsky.app/profile/akwyz.com Debra https://bsky.app/profile/debraruh.bsky.social Neil https://bsky.app/profile/neilmilliken.bsky.social axschat https://bsky.app/profile/axschat.bsky.social LinkedInhttps://www.linkedin.com/in/antoniovieirasantos/ https://www.linkedin.com/company/axschat/Vimeohttps://vimeo.com/akwyzhttps://twitter.com/axschathttps://twitter.com/AkwyZhttps://twitter.com/neilmillikenhttps://twitter.com/debraruh

Our choice of hacks included a fond look at embedded systems and the classic fashion sense of Cornell's Bruce Land, risky open CRT surgery, a very strange but very cool way to make music, and the ultimate backyard astronomer's observatory. We talked about Stamp collecting for SMD prototyping, crushing aluminum with a boatload of current, a PC that heats your seat, and bringing HDMI to the Commodore 64. We also took a look at flight tracking IRL, a Flipper-based POV, the ultimate internet toaster, and printing SVGs for fun and profit. Finally, we wrapped things up with a look at the tech behind real-time river flow tracking and a peek inside the surprisingly energetic world of fuel cells. Check out all the links over at Hackaday!

Our choice of hacks included a fond look at embedded systems and the classic fashion sense of Cornell's Bruce Land, risky open CRT surgery, a very strange but very cool way to make music, and the ultimate backyard astronomer's observatory. We talked about Stamp collecting for SMD prototyping, crushing aluminum with a boatload of current, a PC that heats your seat, and bringing HDMI to the Commodore 64. We also took a look at flight tracking IRL, a Flipper-based POV, the ultimate internet toaster, and printing SVGs for fun and profit. Finally, we wrapped things up with a look at the tech behind real-time river flow tracking and a peek inside the surprisingly energetic world of fuel cells. Check out all the links over at Hackaday!

Scott and Wes sit down with Cassie Evans of GSAP to talk all things animation—from SVGs and scroll timelines to GSAP's new rebrand and exciting features. They also dig into performance, plugins, and what makes GSAP such a powerful tool for web developers. Show Notes 00:00 Welcome to Syntax! 00:59 What is GSAP? Cassie Codes. 01:53 GSAP Rebranding. 04:44 GSAP under the hood. 05:29 The big announcement! 07:19 GSAP Showcase. 11:01 Brought to you by Sentry.io. 11:26 Why is GSAP easier for animations? GSAP Docs. 12:38 Animating with SVGs. 13:33 The love of SVG. 14:55 GSAP is performant. 16:06 Gotchas to watch out for. 18:12 Does GSAP work with canvas? 19:02 What GSAP projects are you most proud of? 20:30 Does it play nice with web frameworks? GSAP with React. 22:32 What are you excited about in CSS right now? Scroll Timeline. 24:27 Will any of these make their way into GSAP? 26:31 Timelines. 29:24 Building animations with timelines. 34:55 What are the best GSAP plugins? Split Text spanran-wrap. Physics 2D Plugin. 38:44 GSAP docs and philosophy. 39:50 Scrubbing animations by frame. 41:09 GSAP Video Exporter. 41:45 Animating with JavaScript. 45:19 JavaScript in unconventional applications. 47:56 Is there anything missing in web tech? 50:53 What about AI in GSAP? 52:40 Sick Picks + Shameless Plugs. Sick Picks Cassie: Eyesy Video Synthesis. Shameless Plugs Cassie: Smashing Conf. Hit us up on Socials! Syntax: X Instagram Tiktok LinkedIn Threads Wes: X Instagram Tiktok LinkedIn Threads Scott: X Instagram Tiktok LinkedIn Threads Randy: X Instagram YouTube Threads

In diesem Deep Dive zeigt Lorenz Grabosch, warum Rive-Animationen für Web- und App-Projekte oft das „Next-Level“-Upgrade zu Lottie sind. Du erfährst: • seinen Weg von After Effects zu Rive • die größten Performance- und Workflow-Vorteile • wie du SVGs aus Figma in Rive importierst, animierst und nahtlos in Webflow einbindest – inkl. Loop-Kontrolle, State Machines & GSAP-Triggern. Perfekt für alle, die interaktive Animationen schneller bauen und leichter warten wollen. Laurenz Grabosch online • LinkedIn: https://www.linkedin.com/in/laurenzgrabosch • Website: https://www.graboschmedia.de --------------------------------------------- Meine Onlinekurse:

Episode #220 - In the latest episode of The Presentation Podcast, Troy and Nolan delve into everything PowerPoint and vector. What does PowerPoint consider a vector graphic? What types of vector files does PowerPoint recognize? What are PowerPoint limitations with vector graphics? This episode is a treasure trove of insights, practical tips, and expert advice on leveraging vector graphics, particularly .SVGs, for great presentation design. Have a listen to see where the conversation goes!   Full Episode Show Notes https://thepresentationpodcast.com/2025/e220   Show Suggestions? Questions for your Hosts? Email us at: info@thepresentationpodcast.com   Listen and review on iTunes. Thanks! http://apple.co/1ROGCUq   New Episodes 1st and 3rd Tuesday Every Month  

Today I'm interviewing Bail from @JustOneMoreProject. This mom of 2 went from curious about Cricut to rocking SVGs with her Glowforge.  Bail left her corporate job to be with her girls. When she launched her Etsy shop to sell SVG files in the middle of a pandemic she discovered her true purpose as a creative entrepreneur. In this episode, she fills us in on Cricut vs. Glowforge from a designer's perspective.EPISODE NOTES: https://kaseyclin.com/blogs/podcast/a-designer-s-take-on-glowforge-with-bail-from-just-one-more-projectThanks for listening! Click one of the links below to. Learn how to design for crafting machines such as Cricut, Silhouette, Glowforge, sublimation or embroidery. Check out my designs on Etsy

Unless you've recently avoided the online world you would have heard all the buzz about laser cutting machines. Well, I have a treat for you today! I sat down with a trusted leader in the laser industry, Emily, otherwise known as “That Mom With A Laser”. If you are interested in getting a laser cutting machine or you want to design SVGs for them, you will definitely benefit from this episode. Click the link and listen in as we chat about the most desired designs for lasers and all the excitement such as fumes, fires, and how to charge top dollar for your designs.EPISODE NOTES: https://kaseyclin.com/blogs/podcast/what-you-need-to-know-about-laser-machines-with-emily-from-that-mom-with-a-laserThanks for listening! Click one of the links below to. Learn how to design for crafting machines such as Cricut, Silhouette, Glowforge, sublimation or embroidery. Check out my designs on Etsy

Adaptability and quality are critical online. Designers need graphics that maintain clarity on any screen size and load seamlessly across devices and platforms. Scalable Vector Graphics (SVGs) meet this demand perfectly, offering dynamic resizing, lightweight code, and native support in all modern browsers. As businesses and creatives increasingly focus on responsiveness, SVGs have become the …

This episode of Generation AI introduces Claude 3.5, Anthropic's latest AI model, and its new "artifacts" feature. Hosts JC Bonilla and Ardis Kadiu explore how Claude 3.5 can create interactive apps and visualizations, making complex topics like p-values easier to teach. They demonstrate Claude's ability to generate code, create web-based apps, and offer collaborative workspaces.Introduction to Claude 3.5Overview of Anthropic's new AI modelComparison to GPT-4 and other AI modelsDiscussion of Claude's improved writing and reasoning capabilitiesThe "Artifacts" FeatureExplanation of Claude's new ability to create interactive contentExamples of artifact types: code snippets, web apps, SVGs, and moreHow artifacts can be shared and collaborated onPractical Application: Teaching P-ValuesDemonstration of Claude creating an interactive app to explain p-valuesDiscussion of the "Surprise Party Planner" analogy for teaching statisticsHow AI-generated examples can make complex topics more relatableTechnical Capabilities of Claude's ArtifactsBreakdown of the code generation processExplanation of the React-based web app created by ClaudeDiscussion of Claude's ability to adapt and correct code based on feedbackImplications for Higher EducationHow AI tools like Claude can enhance teaching methodsPotential for creating personalized learning materialsDiscussion of the time-saving aspects for educatorsCollaboration and Version ControlOverview of Claude's team collaboration featuresExplanation of version control for artifactsHow these features can help in group projects and iterative learning - - - -Connect With Our Co-Hosts:Ardis Kadiuhttps://www.linkedin.com/in/ardis/https://twitter.com/ardisDr. JC Bonillahttps://www.linkedin.com/in/jcbonilla/https://twitter.com/jbonillxAbout The Enrollify Podcast Network:Generation AI is a part of the Enrollify Podcast Network. If you like this podcast, chances are you'll like other Enrollify shows too! Some of our favorites include The EduData Podcast and Visionary Voices: The College President's Playbook.Enrollify is made possible by Element451 — the next-generation AI student engagement platform helping institutions create meaningful and personalized interactions with students. Learn more at element451.com.

Want to start creating digital products to sell as a side hustle or full-time income? We are chatting with Abbi Kirsten this week about creating and selling digital files, e-books, and so much more. If you want to start creating SVG files or any other type of digital product, be sure to catch this discussion from three experts who have successful digital product shops.   Abbi, the creator of Abbi Kirsten Collections, is a self-made creative entrepreneur on a mission to nurture the creative magic inside your soul and bring that bursting imagination inside you to life with easy-to-learn tutorials. Abbi spends every day making messes and magic in her home studio, with a mission to make the world a more colorful place, one craft at a time, through sharing her Cricut tutorials, papercraft projects, and handmade gifts. Be sure to check out Abbi's website: https://www.abbikirstencollections.com/   Get the Launch and Grow Your Creative Business Workbook for only $7! A series of step-by-step questions that will guide you on what and where to sell any handmade or digital product. https://link.craftingcamps.com/biz   Join the FREE! Magic of SVGs workshop: https://link.craftingcamps.com/magic   Don't forget to shop our merch store to support the podcast! https://link.craftingcamps.com/merch   Let us help you craft your future by turning your passion into a paycheck. Angie Holden and Cori George are teaming up for a series of live events dedicated to helping you start and grow your craft business. Be sure to subscribe so you don't miss any of the future episodes! Sign up for our email newsletter here: https://crafting-camps.ck.page/4715c59751   Ask us questions here: https://forms.gle/ShKt64gKjeuneMLeA   Want more from Cori and Angie? Be sure to subscribe to our YouTube channels and follow on Instagram using the links below: https://instagram.com/angieholdenmakes https://instagram.com/heyletsmakestuff https://youtube.com/angieholden https://youtube.com/heyletsmakestuff

Today we discuss advancements in web image standards and new CSS features. Topics include object fit, aspect ratio, border images, filters, clip path, mask-image, and backdrop filters, which offer alternatives to traditional image editing tools like Photoshop. The episode also covers HTML and CSS considerations, such as img and picture elements, alt tags, and loading attributes. New image formats like WebP and AVIF are examined, along with the use of SVGs for accessibility and performance benefits. Go listen...

Are you uncertain about how to begin your journey into opening an SVG shop?

Brodie is a Supercars driver who won the 2023 Supercars Championship! He competed in the NASCAR Cup Series in 2023 and plans to compete more in 2024! We brought Brodie on to talk all things Supercars, NASCAR, and more!

Join me as I unravel the secrets to a thriving SVG business! This episode is packed with practical tips and strategies on creating and selling SVGs that mean serious business. Whether you're a seasoned creator or just starting your SVG shop, this is your guide to turning creativity into profit. How interested are you in elevating your crafting game? Tune in now!

Join me and my SVG Superstar student - Cheryl as we talk about her journey from a resin business to selling ASL SVG cut files with no prior design experience. Want to make passive income selling SVGs? Join me now in SVG Superstar. $500 off until Nov 30. https://cutnmakecrafts.com/svgs-on-desk-ipad-2/

Summary Exploration of the next generation web maps for the National Park Service (NPS) using Map Libra and transitioning from Mapbox GJS. Emphasis on data standards, community collaboration, and avoiding a centralized API approach. Highlights

Join me and @‌thepartycrafter - my student from the SVG Superstar Business Coaching Program as we chat about her successful journey on transitioning from a 6-figure handmade business to making passive income from SVGs.

Stephanie is consciously trying to make meetings better for herself by limiting distractions. A few episodes ago, Joël talked about a frustrating bug he was chasing down and couldn't get closure on, so he had to move on. This week, that bug popped up again and he chased it down! AND he got to use binary search to find its source–which was pretty cool! Together, Stephanie and Joël discuss dependency graphs as a mental model, and while they apply to code, they also help when it comes to planning tasks and systems. They talk about coupling, cycles, re-structuring, and visualizations. Ruby Graph Library (https://github.com/monora/rgl) Graphviz (https://graphviz.org/) Using a Dependency Graph to Visualize RSpec let (https://thoughtbot.com/blog/using-a-dependency-graph-to-visualize-rspec-let) Mermaid.js (https://mermaid.js.org/) Strangler Fig pattern (https://martinfowler.com/bliki/StranglerFigApplication.html) Transcript: JOËL: Hello and welcome to another episode of The Bike Shed, a weekly podcast from your friends at thoughtbot about developing great software. I'm Joël Quenneville. STEPHANIE: And I'm Stephanie Minn. And together, we're here to share a bit of what we've learned along the way. JOËL: So, Stephanie, what's new in your world? STEPHANIE: So, I'm always trying to make meetings better for me [chuckles], more tolerable or more enjoyable. And in meetings a lot, I find myself getting distracted when I don't necessarily want to be. You know, oftentimes, I really do want to try to pay attention to just what I'm doing in that meeting in the moment. In fact, just now, I was thinking about the little tidbit I had shared on a previous episode about priorities, where really, you know, you can only have one priority [laughs] at a time. And so, in that moment, hopefully, my priority is the meeting that I'm in. But, you know, I find myself, like, accidentally opening Slack or, like, oh, was I running the test suite just a few minutes before the meeting started? Let me just go check on that really quick. And, oh no, there's a failure, oh God, that red is really, you know, drawing my eye. And, like, could I just debug it really quick and get that satisfying green so then I can pay attention to the meeting? And so on and so forth. I'm sure I'm not alone in this [laughs]. And I end up not giving the meeting my full attention, even though I want to be, even though I should be. So, one thing that I started doing about a year ago is origami. [laughs] And that ended up being a thing that I would do with my hands during meetings so that I wasn't using my mouse, using my keyboard, and just, like, looking at other stuff in the remote meeting world that I live in. So, I started with paper stars, made many, many paper stars, [laughs] and then, I graduated to paper cranes. [laughs] And so, that's been my origami craft of choice lately. Then now, I have little cranes everywhere around the house. I've kind of created a little paper crane army. [laughs] And my partner has enjoyed putting them in random places around the house for me [laughs] to find. So, maybe I'll open a cabinet, and suddenly, [laughs] a paper crane is just there. And I think I realized that I've actually gotten quite good at doing these crafts. And it's been interesting to kind of be putting in the hours of doing this craft but also not be investing time, like, outside of meetings. And I'm finding that I'm getting better at this thing, so that seemed pretty cool. And it is mindless enough that I'm mentally just paying attention but, yeah, like, building that muscle memory to perfecting the craft of origami. JOËL: I'm curious, for your army of paper cranes, is there a standard size that you make, or do you have, like, a variety of sizes? STEPHANIE: I have this huge stack of, like, 500 sheets of origami paper that are all the same size. So, they're all about, let's say, two or three inches large. But I think the tiny ones I've seen, really small paper cranes, maybe that would be, like, the next level to tackle because working with smaller paper seems, you know, even more challenging. JOËL: I'd imagine the ratio of, like, paper thickness to the size of the thing that you're making is different. STEPHANIE: At this point, they say that if you make 1,000, then you bring good luck. I think I'm well on my way [laughs] to hopefully being blessed with good luck in this household of my little paper crane army. JOËL: It's interesting that you mentioned the power of having something tactile to do with your hands during a meeting, and I definitely relate to that. I feel like it's so easy, even, like, mindlessly, to just hit Command-Tab when I'm doing things on a screen. Like, my hands are on the keyboard. If I'm not doing something, I'm just going to mindlessly hit Command-Tab. It's kind of like on your phone sometimes. I don't know if you do this, like, just scrolling side to side. You're not actually doing anything. You just want motion with your fingers. STEPHANIE: Yes. I know exactly what you're talking about. And it's funny because it's a bit of a duality where, you know, when you are in your development workflow, you want things to be as quick and convenient as possible, so that Command-Tab, you know, is very easy. It's just built in, and that helps speed up your, you know, day-to-day work. But then it's also that little bit of mindlessness, I think, that can get you down the distraction path. When I was first looking for something to do with my hands, to have, like, a little tactile thing to keep me focused in meetings, I did explore getting one of those fidget cubes; I have to say. [laughs] It's just a little toy, you know, that comes with a bunch of different settings for you to fidget with. There's, like, a ball you can roll, you know, with your thumb, or maybe some buttons to click, and it gives you that really satisfying tactile experience. And I know they work really well for a lot of people, but I've really enjoyed the, I guess, the unexpected benefits [chuckles] of getting better at a hobby [laughs] while spending my time at my work. Joël, what is new with you? JOËL: So, a few episodes ago, I talked about a really kind of frustrating bug that I was chasing down that was due to some, like, non-determinism in the environment. And it kind of came, and then it went away. And I wasn't able to get sort of closure on that and had to move on. Well, this week, that bug popped up again, and this time, I was actually able to chase it down. So, that felt really exciting. And I got to use binary search to try to find the source of it, which made me feel really cool. STEPHANIE: Oooh, do tell. What ended up being the issue? JOËL: I'm connecting to an external Snowflake data warehouse, and ActiveRecord tries to fetch the schema and crashes as part of that with some cryptic error that originates from the C extension ODBC Ruby driver package. I figured out that it's probably something to do with, like, a particular table name or something in the table metadata when we're pulling this schema that we're not happy about. But I don't know which table is the one that it's not happy with. Well, this time, I was able to figure out, by reading through some of the documentation, that I can pull subsets of the schema. So, I can pull the first n values of that schema, and it won't crash. It only crashes if I try to fetch the entire set, which is what is happening under the hood. At that point, you know, I could fetch each row individually, but there's hundreds of these. So, you know, I try, okay, what happens if I try to fetch 1,000 of these? Is it going to crash? Because it's a massive system. So, yes, I get a crash. So, I know that a table less than a thousandth in the list of tables is what's causing the problems. So, okay, fetch 500 halfway in between there. It's still going to crash. Okay, 250, 125. I then kind of keep halving all the time until I find one that doesn't crash. And now I know that it is somewhere between the last crash and this one. So, I think it was between 125 and 250. And now I can say, okay, well, let's fetch the first, you know, maybe 200 tables, okay, that crashes. And I keep halving that space until you finally find it. And then, like, okay, so it's this one right here. Now, the problem is the bad table actually crashes. So, I think it ended up being, like, number 175 or something like that. So, I never get to see the actual table itself. But because the list of tables is in alphabetical order, and I can see because I can fetch the first 174 and it succeeds, so I can tell what the previous 5, 6, you know, previous 174 are. I can pretty easily go and look at the actual database and the list of tables and say, okay, well, it's in the same order. And the next one is this one, and hey, look, there is some metadata there that has some very long fields that are longer than one might expect, specifically going over a potentially implied 256-character limit. That seems somewhat suspicious. And, oh, if we remove this table, all of a sudden, everything works. STEPHANIE: Wow, binary search, an excellent debugging tool [laughs] when you have no idea, you know, what could possibly be causing your issue. JOËL: It's such a cool tool. Like, I'm always so happy when I get a chance to use it. The problem is, you need a way to be able to answer the question, like, have I found it? Yes or no? Or, generally, is it greater or less than this current position? STEPHANIE: Well, that's really exciting that you ended up figuring out how to solve the bug. I know last time we talked about it, you kind of had left off in a space of, hopefully, we won't run into this issue again because it's no longer happening. But it seems like you were also set up this time around to be able to debug once it cropped up again. JOËL: Yes. So, binary search is really cool. It's got this, like, very, like, fancy computer science name. But in reality, it's a fairly simple, straightforward technique that I use fairly frequently in my development. And there's another kind of computer sciency fancy-sounding concept that I use all the time. You've all heard me reference this multiple times on the show. You're right; we're finally doing it. This is the dependency graph episode. STEPHANIE: Woo. [laughter] It's time. I'm excited to really dig into it because, you know, as someone who has heard you talk about it a lot, you know, and is maybe a little less familiar with graph theory and how, you know, it can be applied to my day to day work, I'm really excited to dig into a little bit about, you know, what a regular developer needs to know about dependency graphs to add to their toolbox of skills. JOËL: So, I think at its core, the idea of a dependency graph is that you have a group of entities, some of which depend on each other. They can't do a task, or they can't be created unless some other subtasks or dependent actions take place. And so, we have a sort of formal structural way of describing these things. Visually, we often draw these things out where each of the pieces is like a little bubble or a circle, and then we draw arrows towards the things that it depends on. So, if A cannot be done without B being done first, we draw an arrow from A to B. That's kind of how it is in the abstract. More concretely, this kind of thing shows up constantly throughout the work that we do because a lot of what we do as developers is managing things that are connected to each other or that depend on each other. We build complex systems out of smaller components that all rely on each other. STEPHANIE: Yeah, I think it's interesting because I use the word dependency, you know, very frequently when talking about normal work that I'm doing, you know, dependencies as in libraries, right? That we've pulled into our application, or dependencies, like, talking about other classes that are referenced in this class that I'm working in. And I never really thought about what could be explored further or, like, what could be learned from really digging into those connections. JOËL: It's a really powerful mental model. And, like you said, dependencies exist all over our work, and we often use that word. So, you mentioned something like packages, where your application depends on Rails, which in turn depends on ActiveRecord, which in turn depends on a bunch of other things. And so, you've got this whole chain of maybe immediate dependencies, and then those dependencies have dependencies, and those dependencies have dependencies, and it kind of, like, grows outward from there. And in a very kind of simplistic model, you might think, oh, well, it's more, like, a kind of a tree structure. But oftentimes, you'll have things like branches on one side that connect back to branches on the other. And now you've got something that's no longer really tree-like. It's more of a sort of interconnected web, and that is a graph. STEPHANIE: I think understanding the dependencies of your system has also become more important to me as I learn about things that can go wrong when I don't know enough about what my system is, you know, relying on that I had kind of taken for granted previously. I'm especially thinking about packages like we were mentioning, and, you know, not realizing that your application is dependent on this other library, right? That's brought in by a gem that you're using. And there's maybe, like, a security issue, right? With that. And suddenly, you have this problem on your hands that you didn't realize before. And I know that that has been more of a common discussion now in terms of security practices, just being more aware of all the things that you are depending on as really our work becomes more and more interconnected with the things available to us with open source. JOËL: I think where understanding the graph-like nature of this becomes really important is when you're doing something like an upgrade. So, let's say you do have a gem that has a security problem, and you want to upgrade it to fix that security issue. But the upgrade that includes the security patch is also a breaking upgrade. And so, now everything else in your system that depends on that gem or on that package is going to break unless you have them in a version that is compatible with the new version of that gem. And so, you might have to then go downstream and upgrade those packages in a way that's compatible with your app before you can bring in the security patch. And a lot of that can be done automatically by Bundler. Bundler is software that is built around navigating dependency graphs like that and finding versions that are compatible with each other. But sometimes, your code will need to change in order to upgrade one of these downstream gems so that you can then pull in the upgrade from the gem that needs a security patch. And so, understanding a little bit of that graph is going to be important to safely upgrading that gem. STEPHANIE: So, I know another application of dependency graphs that you have thought about and written a blog post for is RSpec let declarations and how a lot of the time when we are using let, you know, we are likely calling other variables defined by let. And so, when you are encountering a test file, it can be really hard to grok what data is being set up in your test. JOËL: Yeah, so that is really interesting because you can define something that will get executed in a lazy fashion if it gets referenced. But then not only is the let lazy and will not trigger unless it's referenced, but a let can reference other lets, which are also lazy, and only get triggered if they get referenced. So, you might have a bunch of lets defined in any order you want throughout a file, and they're all kind of interconnected with these references to each other. But they only get triggered if something calls it directly or it's in this, like, chain of dependencies. And getting a grasp on what actually gets created, which lets will actually execute, which ones don't in a file can quickly get out of hand. And so, thinking of this in terms of a dependency graph has been a really helpful mental model for me to understand what's going on in a complex test file. STEPHANIE: Yeah, absolutely. Especially when sometimes the lets are coming from all over the place, you know, maybe a describe block hundreds of lines away, or even a completely different file if you are using a shared context that's being pulled in. So, I can see why this was a complex problem that could be made a little simpler with plotting out a dependency graph. And in preparation for this episode, I was doing a little bit of my own exploration on this because I certainly know, you know, the pain of trying to figure out what is being executed in my tests when there are a lot of lets that reference each other. And in the blog post, you kind of gave a little step-by-step of how you could start with creating a dependency graph for the test that you're working with. And I was really curious if this process could be automated because, you know, I do enjoy, you know, pulling out the pen and paper [chuckles] every now and then. But I'm not, like, a particularly visual person. God forbid I, like, draw a circle, but then, like, don't have enough space for the rest of the circles. [laughs] So, I was really hoping for a tool that could do this for me, especially if, you know, you do, you have a lot of tests that you have to try to understand in a relatively short amount of time. And so, I ended up doing something kind of hacky with RSpec and overriding let definitions to automate this process. JOËL: That's really cool. So, is the tool that you're trying to build something where you feed it in a spec file, and it gives you some kind of graphical representation like an SVG or something as output? STEPHANIE: Yeah. I did consider that approach first, where you feed in the file, but then I ended up going with something more dynamic where you are running the test, and then as it gets executed, tracing the let definitions and then registering them to build your dependency graph. JOËL: So, you've got some sort of internal modeling that describes a dependency graph. And then, somehow, you're going to turn that, you know, a series of Ruby objects into some kind of visual. STEPHANIE: Yeah, exactly. And the bulk of that work was actually done with a library called RGL, which stands for just Ruby Graph Library. [laughs] And what's nice is that it has a really easy interface for plugging in the vertices and edges of the dependency graph that you want to build. And then, it is already hooked up with Graphviz to, you know, write the SVG to a file. And so, I ended up really just having to build up an array of my dependencies and the connections to each other and then feed it into the constructor of the graph. JOËL: And for all of our listeners, you mentioned Graphviz. That is a third-party tool that can be installed on your machine that can generate these SVG diagrams from...I believe it has its own sort of syntax. So, you create, I believe it's dot, D-O-T, so dot dot file. And based off of that, it generates all sorts of things, but SVG being potentially one of them. STEPHANIE: Yeah. The nice thing was that I actually didn't end up having to use the DSL of Graphviz because the RGL gem was doing them for me. JOËL: Nice. So, it plugs in directly. STEPHANIE: Yeah, exactly. And I was really curious about using this gem because I, you know, just wanted to write Ruby, especially to plug into other things that are already in Ruby. And I found that surprisingly easy, thanks to all of the RSpec config options that they make available to you, including an option to extend the example group class, which is actually where let and let bang is defined. And so, I ended up overriding those classes and using, you know, the name of the let that you're defining and then the block to basically register the dependencies. And I also ended up exploring a little bit with using Ruby's built-in parser to figure out in the block that's being passed to the let, what parts of that block could potentially be a reference to another let. JOËL: That's really cool. Did you get any fun results from that? STEPHANIE: I did. It worked pretty well in being able to capture all of the let declarations, and other lets that it references. And so, I was able to successfully, you know, like, generate a visual dependency graph of all of the lets, so that was really neat. The part that I was really kind of excited about trying next, though I didn't end up having time to yet, was figuring out which of those let values are executed by way of the let bang, right? Which is eager or what is referenced in the test that then gets executed as well. And so, the RGL library is pretty neat and has some formatting options, too, with the Graphviz output. So, you can change the font color or styling options for different, you know, nodes and edges. And so, I was really curious to pursue this further, maybe, and use it to show exactly what gets evaluated now that I have successfully mapped my let graph. JOËL: Right. Because the whole point of this exercise is that not the entire graph is going to get evaluated. The underlying question is, what data actually gets created when my test runs? And so, you build out this whole dependency graph, and then you can follow a few simple rules to say, okay, this branch gets called, this branch gets called, this series of things gets called. And okay, this subset of let blocks trigger, and therefore this data has been created for my given test. STEPHANIE: Yeah. Though I will say that even where I got so far to, just seeing all of the let definitions in a spec file was really helpful to have a better understanding, you know, if I do have to add a test in here, and I'm thinking about reaching for a pre-existing let declaration, to be like, oh, like, it actually, you know, goes on to reference all of these other things that may be factories [chuckles] that are created might make me, you know, think twice, or just have a little better understanding of what I'm really dealing with. JOËL: Right. The idea that when you're calling out to a let, or a factory, or something else that's just a node in a large graph, you're not necessarily referencing just one thing. You might actually be referencing the head of a very long chain of things that maybe you don't intend to trigger the whole thing. STEPHANIE: Yeah, exactly. JOËL: So, in that sense, having a sort of visual or at least an idea of the graph can give you a much better sense of the cost of certain operations that you might have to do. STEPHANIE: The cost of the operations certainly, especially when, you know, you are working in a legacy codebase, and you, you know, like, maybe don't know how everything plays together or is connected. And it's very tempting to just reach for [chuckles] the things that have been, you know, created or built for you. And I'm certainly guilty of that sometimes on this client project, where the domain is so complex, and there are so many associated models. And I'm like, well, like, let me just, you know, use this let that already, you know, has a factory set up for what I think I need for this test. But then realizing, oh, actually, like, it is creating all these things, and do I really need them? I think it can be really challenging to unravel all of that in your head. And so, with this very scrappy tool that I [chuckles] built for my own purposes, you know, maybe it makes it, like, one step easier to try to fully understand what I'm working with and maybe do something different. JOËL: One aspect that I think is really powerful about dependency graphs is that it takes this kind of, like, abstract concept that we oftentimes have an intuitive sense around, the idea that we have different components that depend on each other, and it shows it to us visually on, like, a 2D plane. And that can be really helpful to get an understanding or an overview of a system. You mentioned that RGL uses Graphviz to generate some SVGs. A visual tool that I've been using to draw some of my dependency graphs has been mermaid.js. It has a syntax that's, like, a text-based syntax, but it's almost visual in that you have a piece of text and name of a node. And then, you'll draw a little ASCII arrow, you know, two dashes and a greater than sign to say this thing depends on, and then write another name, and just have a row, like, a bunch of entries to say; A depends on B. A also depends on C. C depends on D, and so on, and, like, build up that list. And then Mermaid will just generate that diagram for you. STEPHANIE: Yeah. I've used Mermaid a few times. One really helpful use that I had for it was diagramming out a bunch of React components that I had and wanting to understand the connections between them. And I think you can even paste the Mermaid syntax into your GitHub pull request description, and it'll render as the graph image. JOËL: Yeah, that's what's really cool is that Mermaid syntax has become embedded in a lot of other places in the past few years. So, it's really easy to embed graphs now into all sorts of things. You mentioned GitHub. It works in pull requests descriptions, comments, I think pretty much anywhere that Markdown is accepted. So, you could put one in your README if you wanted. Another place that I use a lot, Obsidian, my note-taking tool, allows me to embed graphs directly in there, which is really much nicer than previously; sometimes, when I wanted to express something as a visual, I would use some sort of drawing tool to do something and export an image, and then embed that in my note. But now I can just put in this text, and it will automatically render that as a diagram. And part of what's really nice about that is that then it's really easy for me to go and change that if I'm like, oh, but actually, I want to add one more connection in here. I don't have to re go back to, hopefully, a file that I've saved somewhere and, like, change an image file and re-export it. I just, you know, I add one line of text to my note, and it just works. STEPHANIE: That's awesome. Yeah, the ability to change it seems really useful. So, we've talked a little bit about tools for creating a visual aid for understanding our dependencies. And now that we have our graph, maybe we might have some concerning observations about what we see, especially when perhaps some of our dependencies are pointing back to each other. JOËL: Yes. So, I think you're referencing cycles, in particular. That would be the formal term for it. And those are really interesting. They happen in dependency graphs. And I would say, in many cases, they can be a bit of a smell. There's definitely situations where they're fine. But there are things that you look at, and you're like, okay, this is going to be a more complex kind of tricky bit of the graph to work with. Some cases, you just straight up can't have them. So, I want to say that the way RSpec lets are set up, you cannot write code that produces cycles. But you might have...I think Ruby allows classes to reference each other in such a way that it creates a cycle, and not all languages do that. So, Elm and F#, I believe, require that modules cannot reference each other. The fancy term for this is a dependent acyclic graph, or DAG, which basically just means that there are no cycles in that graph. STEPHANIE: Yeah. What you said about classes referencing each other is very interesting because I've definitely seen that. And then, if I have to go about changing something, maybe even it's just the class name, right? Now there's no way in which I can really make just one change. I have to kind of do it all in one go. JOËL: I think that's a common property of a cycle, and a graph is that changes that happen somewhere in that cycle often need to be all shipped together as one piece. You can't break it up into smaller chunks because everything depends on everything else. So, it has to be kind of boxed together and shipped as one thing. STEPHANIE: And you'd mentioned that cycles, you know, can be a bit of a code smell. And if the goal is to be able to break it up so that it is a little bit more manageable to work with, how would you go about breaking a cycle? JOËL: So, I think breaking a cycle is going to vary a little bit based on your problem domain. So, are you modeling a series of classes that are referencing each other? Is this a function call graph? Is this even, like, a series of tasks that you're trying to do? But typically, what you want to do is make sure that eventually, at some point, like, something doesn't loop back to referencing something higher up in your hierarchy. And so, oftentimes, it ends up being about what is allowed to know about what? Do you have higher-level concepts that can know and depend on lower-level concepts but not vice versa? And again, we are talking about this a little bit at the abstract level. But in terms of, let's say, different code modules, or classes, or something like that, commonly, you might say, well, we want some sort of layering where we have almost, like, more primitive types of classes at the bottom. And they don't get to know about anything above them. But the ones above that might be more complex that are composed of smaller pieces know about the ones below them. And you might have multiple layers kind of like that that all kind of point down, but nothing points up. STEPHANIE: That is a very common heuristic. [chuckles] I think you were basically just describing how I also understand creating React components, where you want to separate your presentational ones from your functional ones. And, yeah, it makes a lot of sense that as soon as you start adding that complexity of, you know, those primitive classes at the bottom, starting to, you know, point to things higher up or to know about things higher up, that is where a cycle may be accidentally introduced. JOËL: It's interesting just how many design principles that we have in software. If you dig into them a little bit, you find out that they're about decoupling things, and oftentimes, it's specifically breaking up cycles. So, one way that you might have something like this that actually has dependency in the name, the dependency inversion principle, where what you're effectively doing is you're taking one of those dependency arrows, and you're flipping it the other way. So, instead of A depending on B, you're flipping it. Now B depends on A, and that can be enough to break a cycle. STEPHANIE: So, one thing I've picked up from our conversations about dependency graphs is that oftentimes, you know, when you're trying to figure out where to start, you want to look for those areas or those nodes where there's nothing else that depends on it. JOËL: Yeah. I think you have those nodes that, if this were a tree, you would call them the leaf nodes. In the case of a graph, I'm not sure if that's technically correct, but they don't depend on anything. They're kind of your base case. And so, you can, you know, if it's a function, you can run it. If it's a file, you can load it; if it's a class, also you can load it up and not have to do anything else because it has no dependencies. And knowing that those are there, I think, can be really useful in terms of knowing an order you might want to execute something in. And this is really interesting for one of my favorite uses of a graph, which is breaking down a series of tasks that you need to do. So, commonly, you might say, okay, I have a large task I need to do. I break it down into a series of subtasks. And, you know, maybe I draw out, like, a bulleted list and, you know, task 1, 2, 3, 4, 5. The problem is that they're not necessarily just a flat list. They all have, like, orders, like dependencies between each other. So, maybe one has to happen before 2, but it also has to happen before 3, which needs to happen before two, and, like, there's all these interconnections. And then, you find out that you can't ship them independently the way you thought initially. So, by building up a graph, you end up with something that shows you exactly what depends on what. And then, like you said, the parts that are really interesting where you can start doing work are the ones that have no dependencies themselves. Other things might depend on them, but they have no dependencies. Therefore, they can be safely built, shipped, deployed to production, and they can be done independently of the other subtasks. STEPHANIE: Yeah. I was also thinking about things that could be done in parallel as well. So, if you do have multiple of those items with no dependencies, like, that is a really good way to be able to break up that work and, yeah, identify things that are not blocked. JOËL: For a complex set of tasks, it's great to see, okay, these two pieces have no dependencies. We can have them be done in parallel, shipped independently. And then you can just kind of keep repeating that process. Because once all of the tasks that have no dependencies have been done, well, you can almost, like, remove them from the graph and see, okay, what's the new set of things that have no dependencies? And then, keep doing that until you've eventually done the whole graph. And that may sound like, oh okay, we're just kind of using a little bit of intuition and working through the graph. It turns out that this is a, like, actual, like, formal thing. When it comes to graphs, it's a traversal algorithm called topological sort is the fancy name for it, and it basically, yeah, it goes through that. It gives you a list of nodes in order where each node that you're given has no dependencies that have not been evaluated yet. So, it works from effectively to use our tree terminology, from the leaf nodes to the root, potentially roots plural, of the graph, and each step is independent. So that's a lot of, like, fancy terminology, and getting a little bit of, like, computer science graph theory into here. So, my, like, general heuristic is that graphs should be evaluated from the bottom up when you're trying to evaluate each piece independently. So, when you do that, you get to do each piece independently, as opposed to if you're evaluating from the top down. So, starting from the one thing that depends on everything else, well, it can't be shipped until all of its dependencies have been shipped. And all the transitional dependencies can't be shipped until their dependencies have been shipped. And so, you end up being not able to ship anything until you've built the entire graph. And that's when you end up with, you know, a 2,000-line PR that took you multiple weeks and might be buggy. And it's going to take a long time to review. And it's just not what anybody wants. STEPHANIE: I'm glad you brought this up because I think this is where I am really curious to get better at because oftentimes, when I am breaking down a complex task, it's quite hard for me to see all of the steps that need to happen. And so, you know, you maybe start out with that, like, top-level node, like, the task that needs to be done as you understand it immediately. And it's really hard to actually identify the dependencies and, like, the smaller pieces along the way. And because you're not able to identify that, you think that you do have to just do it all in one go. JOËL: Yeah, that sort of root node is typically the overarching task, the goal of what you want to do. And a common, I think, scenario for something like this would be, let's say, you're doing a Rails upgrade. And so, that root node is upgrade Rails. And a common thing that you might want to do is say, okay, let's go to the gem file, upgrade Rails, see what breaks, and then just keep fixing those things. That's working from the top down. And you're going to be in a long-running branch, and you're going to keep fixing things, fixing things, fixing things until you have found all the things but done all the things. And then you do a big bang upgrade that may have taken you weeks. As opposed to if you're working from the bottom up, you try to figure out, okay, what are all the subtasks? And that might take some exploration. You might not know upfront. But then you might say, okay, here, I can upgrade RSpec versus a dependency, or I need to change the interface of this class and ship all these pieces one at a time. And then, the final step is flipping that upgrade in the gem file, saying, okay, now I've upgraded Rails from 4 to 5, or whatever the version is that you're trying to do. STEPHANIE: I think you've really hit the nail on the head when it comes to trying to do something but not knowing what subtasks may compose of it and getting into that problem of, you know, having not broken it down, like, enough to really see all the dependencies. And, you know, maybe this is a conversation [chuckles] for another episode, but the skill of breaking up those tasks and exploring what those dependencies are, and being able to figure them out upfront before you start to just do that upgrade and then see what happens, that's definitely an area that I want to keep investing in. And I'm sure other people would be really curious about, too, to help them make their jobs easier. JOËL: I think one tip that I've learned that's really fun and that connects into all of this is sometimes you do end up with a cycle in your dependencies of tasks. A technique for breaking that up is a pattern that I have pitched multiple times on the show: the strangler fig pattern. And part of why it's so powerful is that it allows you to work incrementally by breaking up some of these cycles in your dependency graph. And one of the lessons that I've learned from that is that just because you have sort of an initial set of subtasks and you have a graph of them doesn't mean that you can't change them. If you're following strangler fig, what you're actually doing is introducing one or more new subtasks to that graph. But the way you introduce them breaks up that cycle. So, you can always add new tasks or split up existing ones as you get a better understanding of the work you need to do. It's not something that is fixed or set in stone upfront. STEPHANIE: Yeah, that's a really great tip. I think next time, what I really want to explore, you know, your heuristic of going from bottom up, yeah, sure, it sounds all fine and dandy. But how to get to a point where you're able to see everything at the bottom, right? And, like, when you are tasked, or you do start with the thing at the top, like, the end goal. Yeah, I'm sure that's something we'll explore [chuckles] another day. JOËL: On that note, shall we wrap up? STEPHANIE: Let's wrap up. Show notes for this episode can be found at bikeshed.fm. JOËL: This show has been produced and edited by Mandy Moore. STEPHANIE: If you enjoyed listening, one really easy way to support the show is to leave us a quick rating or even a review in iTunes. It really helps other folks find the show. JOËL: If you have any feedback for this or any of our other episodes, you can reach us @_bikeshed, or you can reach me @joelquen on Twitter. STEPHANIE: Or reach both of us at hosts@bikeshed.fm via email. JOËL: Thanks so much for listening to The Bike Shed, and we'll see you next week. ALL: Byeeeeeee!!!!!! ANNOUNCER: This podcast is brought to you by thoughtbot, your expert strategy, design, development, and product management partner. We bring digital products from idea to success and teach you how because we care. Learn more at thoughtbot.com.

Host Sean Lennon talked to NWA Pro Wrestler and former MMA Fighter Blake “Bulletproof” Troop on PWE he's on half of the NWA SVGS with Jax Dane togeher they will take on NWA United States Tag Team Champions The Country Gentlemen at HP Cares for Cooper event July 8th in Highland Park and will be prefoming July 9th for NWA Tapings at Studio One Highland Park. Hes also wrestled for CCW, Control Your Narrative/GLCW, and PCW UltraBulletproof talked about Pro Wrestling about his prior career in MMA and how he was able to use it to adapt your current wrestling style, How is working Television different than Independent shows, how much has he learned from being in NWA, his match with Talos who's over 7ft tall, being in theCrocket Cup where him and Jax Dane fell short to Los Jinetes Del Aire ( Mysteziz Jr and Octagon jr) in his first lucha style match, the Veterans in NWA he looks up to for advice, competing n Submission Grappling, possibly working Josh Barnett Bloodsport event one day, NWA 312 and being apart of a landmark event where owner of NWA Billy Corgan was from, SVGS matches against Country Gentlemen Anthony Andrews and AJ Cazana and their rematch Saturday, the loudest reactions in his former hometown of Los Angelas and more!For the Official PWE Shirt shop at -https://www.prowrestlingtees.com/pro-wrestling-enforcer-logo.htmlFor Exclusive Pro Wrestling Articles and Updated Results for WWE, NXT, AEW, and Impact WrestlingLike the Facebook Page-https://m.facebook.com/PWEPodcast/And Support your favorite Independent Wrestlers by buying Shirts and More at https://www.prowrestlingtees.com/

Host Sean Lennon talked to NWA Pro Wrestler "Thoroughbred" Anthony Andrews on PWE he's one half of the NWA United States Tag Team Champions with AJ Cazana as The Country Gentlemen. They will be in Highland Park July 9th for NWA Tapings at Studio One Highland Park and HP Cares for Cooper July 8th.Anthony talked about Pro Wrestling how has it changed his life for the better, How is working Television different than Independent shows for him, Being In the Crocket Cup they fell short to The Immortals Kratos and Odinson, Their match was against SVGS when you're in there with Blake Troop and Jax Dane in Highland Park PPV pre-match and the rematch at The Crocket Cup Night 2 Pre Show Tag Team Gauntlet, What is the loudest reaction he's had to either their entrance or moves, and more!For the Official PWE Shirt shop at -https://www.prowrestlingtees.com/pro-wrestling-enforcer-logo.htmlFor Exclusive Pro Wrestling Articles and Updated Results for WWE, NXT, AEW, and Impact WrestlingLike the Facebook Page-https://m.facebook.com/PWEPodcast/And Support your favorite Independent Wrestlers by buying Shirts and More at https://www.prowrestlingtees.com/

Schepp, Hans, und Peter ließen mal wieder das Glücksrad kreisen! FRONTEERS KONFERENZ Schepp lädt Euch ein, die von ihm mit-organisierte und nicht gewinnorientierte Fronteers Konferenz am 21. und 2…

Today's Feature episode of The Two Man Power Trip of Wrestling is with NWA Superstars, the SVGS: Jax Dane, Bulletproof Blake Troop, and Chris Silvio. The SVGS join the show to talk about their entire professional wrestling careers. Host John Poz and the SVGS will talk about the NWA, The Crockett Cup, Billy Corgan, winning NWA Tag team titles, Country Gentlemen, Tyrus, Chris Adonis, Trevor Murdoch, Mike Knox, The Briscoe Brothers, plus so much more!The best and easy way to win money is playing fantasy! Join @underdogfantasy today (underdogfantasy.com) and enter the promo code POWERTRIP to double your deposit. Then all you have to do is pick a game, guess higher or lower on or draft your team; then sit back, relax & watch the money roll in! - Underdog Fantasy Promo Code: POWERTRIPStore - Teepublic.com/stores/TMPTFollow us @TwoManPowerTrip on Twitter and IG

Aurooba and Brian explore the new WP_HTML_Tag_Processor that landed in WordPress 6.2, and how handy it would be in scenarios where they've had to use regular expressions to modify HTML in their WordPress projects. They then walk through a function using the new processor to inline and safely modify SVGs, ways they could improve the function and take advantage of the processor's capabilities more, and discuss other ways the processor could enhance complicated filtering of rendered output in WordPress.A full transcript of the episode is available on the website. Watch the video podcast on YouTube and subscribe to our channel and newsletter to hear about episodes (and more) first!The code from this episode – https://github.com/viewSourcePodcast/viewSource-blocks/tree/episode/html-tag-processorAurooba's post with a walkthrough of the code – https://aurooba.com/inline-svgs-in-your-wordpress-code-with-this-helper-function/The Dev Note – https://github.com/WordPress/gutenberg/issues/44410#issuecomment-1420787816Brian's website – https://www.briancoords.comAurooba's website – https://aurooba.comPHP Inline SVG helper function from Ian – https://gist.github.com/ian-svoboda-prom/b7a436d5397e007b7d8d5427465ff161Gist to get the file contents of an SVG – https://gist.github.com/rugor/1decc15655891825d15a7024a06d898bWP SVG Helper functions – https://gist.github.com/kingkool68/6d72977fe8c82eeb9a85295ac3623cdeHow to Modify HTML in a PHP WordPress Plugin Using The New Tag Processor API – https://adamadam.blog/2023/02/16/how-to-modify-html-in-a-php-wordpress-plugin-using-the-new-tag-processor-api/

Gutenburg shipped the first working printing press around 1450 and typeface was born. Before then most books were hand written, often in blackletter calligraphy. And they were expensive.    The next few decades saw Nicolas Jensen develop the Roman typeface, Aldus Manutius and Francesco Griffo create the first italic typeface. This represented a period where people were experimenting with making type that would save space. The 1700s saw the start of a focus on readability. William Caslon created the Old Style typeface in 1734. John Baskerville developed Transitional typefaces in 1757. And Firmin Didot and Giambattista Bodoni created two typefaces that would become the modern family of Serif. Then slab Serif, which we now call Antique, came in 1815 ushering in an era of experimenting with using type for larger formats, suitable for advertisements in various printed materials. These were necessary as more presses were printing more books and made possible by new levels of precision in the metal-casting. People started experimenting with various forms of typewriters in the mid-1860s and by the 1920s we got Frederic Goudy, the first real full-time type designer. Before him, it was part of a job. After him, it was a job. And we still use some of the typefaces he crafted, like Copperplate Gothic. And we saw an explosion of new fonts like Times New Roman in 1931. At the time, most typewriters used typefaces on the end of a metal shaft. Hit a kit, the shaft hammers onto a strip of ink and leaves a letter on the page. Kerning, or the space between characters, and letter placement were often there to reduce the chance that those metal hammers jammed. And replacing a font would have meant replacing tons of precision parts. Then came the IBM Selectric typewriter in 1961. Here we saw precision parts that put all those letters on a ball. Hit a key, the ball rotates and presses the ink onto the paper. And the ball could be replaced. A single document could now have multiple fonts without a ton of work. Xerox exploded that same year with the Xerox 914, one of the most successful products of all time. Now, we could type amazing documents with multiple fonts in the same document quickly - and photocopy them. And some of the numbers on those fancy documents were being spat out by those fancy computers, with their tubes. But as computers became transistorized heading into the 60s, it was only a matter of time before we put fonts on computer screens. Here, we initially used bitmaps to render letters onto a screen. By bitmap we mean that a series, or an array of pixels on a screen is a map of bits and where each should be displayed on a screen. We used to call these raster fonts, but the drawback was that to make characters bigger, we needed a whole new map of bits. To go to a bigger screen, we probably needed a whole new map of bits. As people thought about things like bold, underline, italics, guess what - also a new file. But through the 50s, transistor counts weren't nearly high enough to do something different than bitmaps as they rendered very quickly and you know, displays weren't very high quality so who could tell the difference anyways.  Whirlwind was the first computer to project real-time graphics on the screen and the characters were simple blocky letters. But as the resolution of screens and the speed of interactivity increased, so did what was possible with drawing glyphs on screens.  Rudolf Hell was a German, experimenting with using cathode ray tubes to project a CRT image onto paper that was photosensitive and thus print using CRT. He designed a simple font called Digital Grotesk, in 1968. It looked good on the CRT and the paper. And so that font would not only be used to digitize typesetting, loosely based on Neuzeit Book. And we quickly realized bitmaps weren't efficient to draw fonts to screen and by 1974 moved to outline, or vector, fonts. Here a Bézier curve was drawn onto the screen using an algorithm that created the character, or glyph using an outline and then filling in the space between. These took up less memory and so drew on the screen faster. Those could be defined in an operating system, and were used not only to draw characters but also by some game designers to draw entire screens of information by defining a character as a block and so taking up less memory to do graphics.  These were scalable and by 1979 another German, Peter Karow, used spline algorithms wrote Ikarus, software that allowed a person to draw a shape on a screen and rasterize that. Now we could graphically create fonts that were scalable.  In the meantime, the team at Xerox PARC had been experimenting with different ways to send pages of content to the first laser printers. Bob Sproull and Bill Newman created the Press format for the Star. But this wasn't incredibly flexible like what Karow would create. John Gaffney who was working with Ivan Sutherland at Evans & Sutherland, had been working with John Warnock on an interpreter that could pull information from a database of graphics. When he went to Xerox, he teamed up with Martin Newell to create J&M, which harnessed the latest chips to process graphics and character type onto printers. As it progressed, they renamed it to Interpress. Chuck Geschke started the Imaging Sciences Laboratory at Xerox PARC and eventually left Xerox with Warnock to start a company called Adobe in Warnock's garage, which they named after a creek behind his house. Bill Paxton had worked on “The Mother of All Demos” with Doug Engelbart at Stanford, where he got his PhD and then moved to Xerox PARC. There he worked on bitmap displays, laser printers, and GUIs - and so he joined Adobe as a co-founder in 1983 and worked on the font algorithms and helped ship a page description language, along with Chuck Geschke, Doug Brotz, and Ed Taft.  Steve Jobs tried to buy Adobe in 1982 for $5 million. But instead they sold him just shy of 20% of the company and got a five-year license for PostScript. This allowed them to focus on making the PostScript language more extensible, and creating the Type 1 fonts. These had 2 parts. One that was a set of bit maps And another that was a font file that could be used to send the font to a device.  We see this time and time again. The simpler an interface and the more down-market the science gets, the faster we see innovative industries come out of the work done. There were lots of fonts by now. The original 1984 Mac saw Susan Kare work with Jobs and others to ship a bunch of fonts named after cities like Chicago and San Francisco. She would design the fonts on paper and then conjure up the hex (that's hexadecimal) for graphics and fonts. She would then manually type the hexadecimal notation for each letter of each font.  Previously, custom fonts were reserved for high end marketing and industrial designers. Apple considered licensing existing fonts but decided to go their own route. She painstakingly created new fonts and gave them the names of towns along train stops around Philadelphia where she grew up. Steve Jobs went for the city approach but insisted they be cool cities. And so the Chicago, Monaco, New York, Cairo, Toronto, Venice, Geneva, and Los Angeles fonts were born - with her personally developing Geneva, Chicago, and Cairo. And she did it in 9 x 7.  I can still remember the magic of sitting down at a computer with a graphical interface for the first time. I remember opening MacPaint and changing between the fonts, marveling at the typefaces. I'd certainly seen different fonts in books. But never had I made a document and been able to set my own typeface! Not only that they could be in italics, outline, and bold. Those were all her. And she inspired a whole generation of innovation. Here, we see a clean line from Ivan Sutherland and the pioneering work done at MIT to the University of Utah to Stanford through the oNLine System (or NLS) to Xerox PARC and then to Apple. But with the rise of Windows and other graphical operating systems. As Apple's 5 year license for PostScript came and went they started developing their own font standard as a competitor to Adobe, which they called TrueType. Here we saw Times Roman, Courier, and symbols that could replace the PostScript fonts and updating to Geneva, Monaco, and others. They may not have gotten along with Microsoft, but they licensed TrueType to them nonetheless to make sure it was more widely adopted. And in exchange they got a license for TrueImage, which was a page description language that was compatible with PostScript. Given how high resolution screens had gotten it was time for the birth of anti-aliasing. He we could clean up the blocky “jaggies” as the gamers call them. Vertical and horizontal lines in the 8-bit era looked fine but distorted at higher resolutions and so spatial anti-aliasing and then post-processing anti-aliasing was born. By the 90s, Adobe was looking for the answer to TrueImage. So 1993 brought us PDF, now an international standard in ISO 32000-1:2008. But PDF Reader and other tools were good to Adobe for many years, along with Illustrator and then Photoshop and then the other products in the Adobe portfolio. By this time, even though Steve Jobs was gone, Apple was hard at work on new font technology that resulted in Apple Advanced Typography, or AAT. AAT gave us ligature control, better kerning and the ability to write characters on different axes.  But even though Jobs was gone, negotiations between Apple and Microsoft broke down to license AAT to Microsoft. They were bitter competitors and Windows 95 wasn't even out yet. So Microsoft started work on OpenType, their own font standardized language in 1994 and Adobe joined the project to ship the next generation in 1997. And that would evolve into an open standard by the mid-2000s. And once an open standard, sometimes the de facto standard as opposed to those that need to be licensed. By then the web had become a thing. Early browsers and the wars between them to increment features meant developers had to build and test on potentially 4 or 5 different computers and often be frustrated by the results. So the WC3 began standardizing how a lot of elements worked  in Extensible Markup Language, or XML. Images, layouts, colors, even fonts. SVGs are XML-based vector image. In other words the browser interprets a language that displays the image. That became a way to render Web Open Format or WOFF 1 was published in 2009 with contributions by Dutch educator Erik van Blokland, Jonathan Kew, and Tal Leming. This built on the CSS font styling rules that had shipped in Internet Explorer 4 and would slowly be added to every browser shipped, including Firefox since 3.6, Chrome since 6.0, Internet Explorer since 9, and Apple's Safari since 5.1. Then WOFF 2 added Brotli compression to get sizes down and render faster. WOFF has been a part of the W3C open web standard since 2011.  Out of Apple's TrueType came TrueType GX, which added variable fonts. Here, a single font file could contain a number or range of variants to the initial font. So a family of fonts could be in a single file. OpenType added variable fonts in 2016, with Apple, Microsoft, and Google all announcing support. And of course the company that had been there since the beginning, Adobe, jumped on board as well. Fewer font files, faster page loads.  So here we've looked at the progression of fonts from the printing press, becoming more efficient to conserve paper, through the advent of the electronic typewriter to the early bitmap fonts for screens to the vectorization led by Adobe into the Mac then Windows. We also see rethinking the font entirely so multiple scripts and character sets and axes can be represented and rendered efficiently.  I am now converting all my user names into pig Latin for maximum security. Luckily those are character sets that are pretty widely supported. The ability to add color to pig Latin means that OpenType-SVG will allow me add spiffy color to my glyphs. It makes us wonder what's next for fonts. Maybe being able to design our own, or more to the point, customize those developed by others to make them our own. We didn't touch on emoji yet. But we'll just have to save the evolution of character sets and emoji for another day. In the meantime, let's think on the fact that fonts are such a big deal because Steve Jobs took a caligraphy class from a Trappist monk named Robert Palladino while enrolled at Reed College. Today we can painstakingly choose just the right font with just the right meaning because Palladino left the monastic life to marry and have a son. He taught jobs about serif and san serif and kerning and the art of typography.  That style and attention to detail was one aspect of the original Mac that taught the world that computers could have style and grace as well. It's not hard to imagine if entire computers still only supported one font or even one font per document. Palladino never owned or used a computer though. His influence can be felt through the influence his pupil Jobs had. And it's actually amazing how many people who had such dramatic impacts on computing never really used one. Because so many smaller evolutions came after them. What evolutions do we see on the horizon today? And how many who put a snippet of code on a service like GitHub may never know the impact they have on so many?

In this episode, the amazing Liz Wilcox (love her energy!) talks about why email marketing is SOOOOO important for digital businesses and WHY every single business owner NEEDS to attend her FREE workshop on email marketing at the Sell Digital Crafts - The Smart Way summit. This episode is sponsored by Creative Fabrica. Creative Fabrica has generously offered this free All Access subscription trial deal: https://www.creativefabrica.com/promo/6768/0P536-EFGHIJKLMNO/ref/1444514/ With this deal, you can try out the All Access Subscription by downloading 10 fonts & designs of your choice for free! After the 10th download or after 1 month, the subscription renews at $9. This is almost a 70% discount for the normal $29! But don't worry you can cancel anytime. Sign-up here https://www.creativefabrica.com/promo/6768/0P536-EFGHIJKLMNO/ref/1444514/ PS. I'm putting together the second Sell Digital Crafts summit for busy crafters who want to make passive income. In the summit you'll learn how to design and sell stickers, SVGs, Canva templates, Procreate brushes, and hand-lettering Etsy marketing workshops for digital businesses How to skyrocket your sales through Pinterest, Instagram and TikTok The event is completely FREE of cost. You'll have 24 hours to watch the videos for FREE. However, if you want on-going access to the presentations you can grab the All-Access Pass for just $67 for a limited time. The FREE summit runs from April 3-5, 2023! Registration opens on March 15th. Wohoo! Sign up here: https://selldigitalcrafts.com/

On this week's episode of the Alliance Guys Podcast– we will examine the landscape for the United Wrestling Network. Of course with the very popular Championship Wrestling Presented by Car Shield, Memphis Wrestling, and the addition of Derby City Wrestling, could David Marquez be conquering wrestling, the old-fashioned way? More matches have been announced for the NWA 312. The women's Tag Team Championship Match has been made. The Women's Tag Team Match is added to Tyrus defending against Adons, Cyon defending against EC3, Kamille defending against La Rosa Negra, and the Country Gentlemen defending against the SVGS. Let's talk about it. Taya Valkyrie has officially signed with AEW. Ashley D'amboise has recently signed with AEW. What will these most recent signings mean for the NWA? And the breakdown of Thom Latimer's 1st title defense and the rest of the results of NWA POWERRR!!!!

In this episode, I'm pulling back the curtains on some of the mindset challenges I faced in my business and how I almost threw everything in the trash until I started a daily 5-minute exercise. Listen to find out more! It's just 11 minutes ;). This episode is sponsored by Creative Fabrica. Creative Fabrica has generously offered this free All Access subscription trial deal: https://www.creativefabrica.com/promo/6768/0P536-EFGHIJKLMNO/ref/1444514/ With this deal, you can try out the All Access Subscription by downloading 10 fonts & designs of your choice for free! After the 10th download or after 1 month, the subscription renews at $9. This is almost a 70% discount for the normal $29! But don't worry you can cancel anytime. Sign-up here https://www.creativefabrica.com/promo/6768/0P536-EFGHIJKLMNO/ref/1444514/ PS. I'm putting together the second Sell Digital Crafts summit for busy crafters who want to make passive income. In the summit you'll learn how to design and sell stickers, SVGs, Canva templates, Procreate brushes, and handlettering Etsy marketing workshops for digital businesses How to skyrocket your sales through Pinterest, Instagram and TikTok The event is completely FREE of cost. You'll have 24 hours to watch the videos for FREE. However, if you want on-going access to the presentations you can grab the All-Access Pass for just $67 for a limited time. Follow me on Instagram to stay on top of the summit news. Registration opens on March 15th. Wohoo! Join the waitlist here: https://selldigitalcrafts.com/waitlist/

The SVG market on Etsy is an exciting (and growing!) place to be. Tune in today to hear from Jav Sid, who is a top SVG seller and coach, and learn how you can build a {mostly} passive income selling this digital product on Etsy. No design experience required!

John Inness is a guidance, navigation, and control engineer at Marshall Space Flight Center supporting various projects including SLS, HLS, Solar Cruiser, SVGS, and more. He is originally from North Carolina where he went to North Carolina State University for aerospace engineering and co-oped at Marshall Space Flight Center while a student. Upon graduation, he began full time work at MSFC. In his spare time, John enjoys playing music, cooking and trying various foods, hiking, and ultimate frisbee. As always, stay tuned after for our takeaways. --- Support this podcast: https://anchor.fm/shawna-christenson2/support

Chris and Elecia talk about house maintenance, blinking LEDs, paper engineering and more.  Cutting Mobius Strips Video: Tadashi Tokieda cuts various combinations of loops and Mobius loops - with surprising results. festi.info/boxes.py generates boxes for laser cutting (or other SVG consuming device). Boxes.py is a python module that lets you programmatically generate the SVGs. (Github repo) Amanda Ghassaei's Sugarcube is a MIDI instrument using this SparkFun button pad. We also talked about the Mikroe 8800 Retro Click. Elecia is taking Paper Engineering with Kelli Anderson. Chris is taking songwriting courses from School of Song. Transcript

Hackers target Japanese politicians with new MirrorStealer malware Crooks use HTML smuggling to spread QBot malware via SVG files FBI charges 6, seizes domains linked to DDoS-for-hire service platforms Thanks to this week's episode sponsor, Fortra The cybersecurity landscape is full of single-solution providers, making it easy for unexpected cyberthreats to sneak through the cracks. That's why Fortra is creating a stronger, simpler strategy for protection. One that increases your security maturity while decreasing the operational burden that comes with it. Fortra's integrated, scalable solutions help customers face their toughest challenges with confidence. Learn more at Fortra.com. For the stories behind the headlines, head to CISOseries.com.

This month I'm hosting a special event for busy crafters who want to make passive income. It's called Sell Digital Crafts - The Smart Way Summit. In the summit you'll learn how to design and sell stickers, SVGs, Canva templates, Procreate brushes, and handlettering Etsy marketing workshops for digital businesses How to skyrocket your sales through Pinterest, Instagram and TikTok The event is completely FREE of cost. You'll have 24 hours to watch the videos for FREE. However, if you want on-going access to the presentations you can grab the All-Access Pass for just $67 for a limited time. Follow me on Instagram to stay on top of the summit news. Registration opens on October 5th. Wohoo! Follow me here: https://www.instagram.com/cutnmakecrafts/

In today's episode I'm interviewing an SVG Superstar alumni, Sarah from Heart SVGs who started her SVG design journey feeling scared and overwhelmed but after 12 months in the program, she feels confident and a pro at designing SVGs. Listen to this podcast to find out the no.1 secret behind her tremendous success how she was able to design and sell SVGs without any prior graphic designing experience Check out Sarah's shop below: https://www.etsy.com/au/shop/heartSVGs https://www.instagram.com/heart.svgs/

Accessible SVGs (Scalable Vector Graphics) featuring Guest Heather Neff of Equivalent Design. If you have ever wanted to understand how smart SVG graphics can increase accessibility while maintaining brand style this is the podcast for you. Chad and Dax talk with Heather Neff about her goal to bring SVGs to the world "One graphic at a time." We talk about embedded alt-text and how WordPress now pulls description metadata straight from your images and how that can improve your search engine placement.

I know as crafters our no. 1 choice for learning something is to hop onto Youtube to learn literally anything and everything. Without giving it a second thought, we hop onto Youtube and get caught in the spiral of watching one video after another. So the question is, is this the right thing to do? Or is there a better alternative out there? Find out in today's podcast.

What to do when you are in a dire need of cash and your SVG shop sales are slow? When you're in desperate need of money, and you want to instantly make use of your new skill which is designing SVG cut files. Listen to this episode to find out 5 THINGS you can do to make money from SVGs besides selling them in your shop.

In this episode, I'm chatting with one of the talented SVG Superstar (SVG Business Coaching Program) students about her experience in running a successful SVG business side hustle besides her full-time day job.

This week I partnered up with Rafi from Ideastrider to talk about what it's like to grow in your comfort zone. In this episode, we discussed...

In this week's episode, Tif interviews Naa Ardua. Naa discusses how she got started creating clipart, how she transitioned to creating scalable vector graphics (SVG), and how her SVGs bring a voice to a unique community.Here are links to Naa Ardua's SVG courses: So Simple SVG Course and So Simple SVG Files course - 3 Month Payment PlanNaa Ardua's Free SVG Course -> So Quick SVG Course The Book PileWant to sound smart? Two comics discuss a book every Monday so you can pretend you read.Listen on: Apple Podcasts Spotify

This week, we make the case for designing product icons and discuss what to do if your team doesn't want to invest in great iconography. In the Sidebar, we talk about how to incorporate default typefaces into your design system.The Sidebar:The Sidebar is an exclusive weekly segment for our Patreon supporters. You can subscribe starting at $1 per month for access to bonus content going forward! Sign up at patreon.com/designdetails.Latest VIP Patrons:Christian RotzollZack MeredithJSZach GeistKates Hartmanmike kennyWilsonAndre DinizMattias KronbergFollow up:Uhl Albert pointed out that “ruinous empathy” comes from the Radical Candor Framework.Main Topic:This week, we make the case for designing product icons and discuss what to do if your team doesn't want to invest in great iconography.Priscilla Then asks on GitHub:The Lead Developer at my work requested to remove some of the icons, because they slow down development time when having to wait for one to be created, grappling with SVGs, etc. Unfortunately, he is right as they are time-consuming to make (especially good ones) and since I'm the only UX/UI designer in the company, it slows me down and I become the bottleneck.That said, I feel that icons give character, polish and professionalism to an otherwise extremely utilitarian UI. Icons can also be functional in that they help make certain elements stand out from an otherwise huge sea of text, but they are still seen as "fluff". I've asked for another designer to be hired, or request our graphic designer to create icons when he has time, but these suggestions didn't get any support. How would you make a case for icons when they are seen as mere decoration?Feather IconsIconic IconsBootstrap IconsAnt IconsMaterial Design IconsIcons ToolboxIcon SizesCool Things:Brian shared the Opal C1, a professional webcam that claims DSLR-quality images in a $300 chassis. So far: it's definitely a beta product, but depending on your needs it might be a worthwhile upgrade!Marshall shared the ohsnap! Snapmount 2.0, an accessory that helps you mount your phone to a wall (with MagSafe!), making it ideal for turning your phone into a mounted shower speaker.Design Details on the Web:

In this episode, Amy and James discuss all things SVGs: what is, why, and when to reach for it, and seven different ways to get an SVG on the page, and the pros and cons of each method.SponsorsVercelVercel combines the best developer experience with an obsessive focus on end-user performance. Their platform enables frontend teams to do their best work. It is the best place to deploy any frontend app. Start by deploying with zero configuration to their global edge network. Scale dynamically to millions of pages without breaking a sweat.For more information, visit Vercel.comZEAL is hiring!ZEAL is a computer software agency that delivers “the world's most zealous” and custom solutions. The company plans and develops web and mobile applications that consistently help clients draw in customers, foster engagement, scale technologies, and ensure delivery.ZEAL believes that a business is “only as strong as” its team and cares about culture, values, a transparent process, leveling up, giving back, and providing excellent equipment. The company has staffers distributed throughout the United States, and as it continues to grow, ZEAL looks for collaborative, object-oriented, and organized individuals to apply for open roles.For more information visit softwareresidency.com/careersDatoCMSDatoCMS is a complete and performant headless CMS built to offer the best developer experience and user-friendliness in the market. It features a rich, CDN-powered GraphQL API (with real-time updates!), a super-flexible way to handle dynamic layouts and structured content, and best-in-class image/video support, with progressive/LQIP image loading out-of-the-box."For more information, visit datocms.comShow Notes0:00 Introduction3:50 What is an SVG?Raster vs Vector6:21 Benefits to using an SVGChange the SizeSmall File SizeChange the color within your codeEasily Cached9:51 Seven Different Ways to get an SVG on the Page11:28 Sponsor - ZEAL12:59 Option 1 - Image Tag14:03 Option 2 - Inline SVG tag15:53 Option 3 - CSS as a  background Image16:18 Option 4 - CSS, as a Mask18:20 Sponsor - Vercel19:29 Option 5 - SVG directly within our Image tag21:20 Option 6 - Base64 or UTF8 with as a CSS Background Image21:47 Option 7 - An SVG Sprite22:34 Writing your own SVGs27:00 Going Deep on a Specific Topic, The Broken Comb28:34 ResourcesAmy's SVG Series on YouTubeSarah Drasner - Course on Frontend Masters, SVG Essentials & Animation, v2Sarah Drasner - SVG Animations: From Common UX Implementations to Complex Responsive AnimationChris Coyier - Practical SVG29:19 Sponsor - DatoCMS30:12 Grab Bag Questions30:56 Picks and Plugs31:07 Amy's Pick - Animal Cable Clips32:00 Amy's Plug - Advent of CSS32:32 James's Pick - Castle on Hulu33:34 James's Plug - Advent of JavaScript