Michael Labbé

@michaellabbe" target="_blank" rel="nofollow noopener">mastodon.gamedev.place/@michaellabbe

@neogoose_btw Check out FoundationDB. Ordered key/value store with ACID properties.

I never liked databases as an idea. You literally send a string query over TCP to postgres and it returns you data over tcp as strings. There is so much potential to make this whole thing better …. but everyone seems to be just fine with it.

@ScottApogee I used to listen to him as a teenager while I coded into the early am hours on weekends. The Prey reference landed immediately for me.

I remember many nights coding games with Art's radio show playing in the background like an old friend keeping me company. We had him record several "reports from Earth" for Prey (2006) and he nailed each one on the first take. Truly a legend. RIP

On June 6 we are hosting the first-ever Handmade Network Expo in Vancouver, BC, an in-person event celebrating software made from scratch. Among the lineup: a new game from @antovsky & @nicbarkeragain, a time travel debugger for macOS by @leddoo_dev, a custom metaprogramming language and IDE from @frogtoss, a custom game engine by @gdechichi, and more. More info and tickets below :)

@SebAaltonen @mrdoob @akirodic I am wondering out loud here, but you cheaply host a large range of compiled variants on a CDN to avoid burdening your user with building when selecting between options.

Yeah, but I do fear about the ergonomics. JS/TS ecosystem is so nice. Compiling WASM modules (download C++/emscripten tool chain) breaks that flow. Could create a JS interface for configuring the renderer (like Tensorflow does for configuring CUDA with Python), but it has architectural challenges. Unity SRP architecture failed badly. Practically nobody did a custom SRP, and even Unity failed to maintain two SRPs. HDRP is practically killed and URP has new extension mechanisms, instead of writing a custom SRP. Extendible render architecture between fast<->slow language is still an unsolved problem. Solving that architectural problem is crucial if I want wide adaptation. Otherwise it's going to be a WASM module for WASM-only projects. Which is a significantly smaller market.

@SebAaltonen @akirodic If there's anyone who can achieve this, it's you 💪

@KazeEmanuar I think the experience is more about not picking it up. If I have five minutes I'm not going to pop into a matchmaker after a splash screen and attract mode.

Have you ever put down a game specifically because of long load times?

Inspired by this, I ran /autoresearch on a wasm32 compiler program I already hand wrote and optimized in C. It achieved an 8% reduction in code size using a local Qwen3-Coder-Next:q8_0 model overnight. Like Tobi's, it was an overfit, but I was able to keep 3%. Compelling ROI.

@rfleury And for anyone who says "that's just good security" -- it makes no sense for locally executing programs. I can just intercept WM_KEYDOWN events from the editor's message loop from another local program. The isolation is only in-process.

@rfleury "native has nothing to offer" Every text editor written in Electron spawns a new project with a new window because each window has the isolation of a browser tab. You have to code an explicit backchannel to overcome this, and none of them have done this seamlessly.

“native has nothing to offer” No, you’ve just either never bothered to seriously try it, or you just suck.

@ZoidCTF Casey poured himself fully into this. It was an important early project for us, too. Building content to showcase hardware features ahead of commercial game support is still important 26 years later.

My codesigner on Threewave CTF for Quake 2 and Quake made this map. Casey made several amazing levels and he fully cranked it up on this map. He pushed QERadiant to its level design limits. Cool to see it again.

@ZoidCTF I do like catching stutters closer to when they got added. In the last engine I wrote, a long frame dumps the time spent in each subsystem to the console. Prevent stutter-generating code from being built on top of unknowingly.

This gives me PTSD for when I was optimizing games. I spent weeks staring at Telemetry graphs when working on Dota 2. Ari is jokingly here, but there's a basic truth in what he's saying. It's not this simple, but its honestly where you start.

@gdechichi A good exercise to build appropriate trust in the ai learning process is to do a memory-free chat, tell it you're just starting out and your goals are where you are currently. Then ask it to give you training steps to reach your current position.

@TylerGlaiel This is the way. Congratulations, by the way.

no mentions of technical problems in any reviews. Which means I did (one of) my jobs right. it is funny how doing that right means people don't notice it though

@cmuratori Unfortunately, in hindsight that makes sense - Valve would look at customer sentiment before locking things in.

I don't remember the exact exchange, but we talked about this on the Stand Up with Luke from LTT. We were worried that Steam machine pricing would be affected, and the hope was that they had already allocated their allotment at old pricing. I guess they did not :(

@KirkLubimov I read it this way as well, and it seemed like a classic opening salvo threat about future action, not revenge.

It is important to point out: Donald Trump's threat of 100% tariffs on Canada is based on *if* Canada makes a deal with China. Canada isn't making a (free) trade deal with China, they just solved some trade dispute. Any expansion of the trade deal could spark things. The welcoming of Chinese EV into Canada's market will hurt Canadian auto manufacturing industry more than anything. Mexico has Chinese EVs and have a greater tariff on them than what Canada has offered. Mexican residents can drive them into America temporarily. American citizens can't.

@Grady_Booch @OfficialLoganK Accurate

engineers are artists now thanks to AI

You Can't Promote a Programming Language By Its Features Alone "In reviewing why languages are successful there is often a retroactive feature-driven mythology wherein we over-attribute the success of a language to its unique syntax and functionality." 🔗

@TheGingerBill @rfleury Why call it AI when you can call it a Prebugger

@rfleury And when it corrects with something incorrect, you just ask it again to fix it. It's so simple! So revolutionary! So turn tabling!

If it were as revolutionary as they claim, AI programming promoters would spend a lot less time talking about it, and a lot more time stealing market share from existing software products. Has anyone generated a better debugger yet?

@ID_AA_Carmack What would a company full of people who seeded LLMs with job-prospect enhancing prompts look like?

In some important ways, a user’s LLM chat history is an extended interview. The social media algorithms learn what you like, but chats can learn how you think. You should be able to provide an LLM as a job reference, just like you would a coworker, manager, or professor. It can form an opinion and represent you without revealing any private data. Most resumes are culled by crude filters in HR long before they get to the checking-references stage, but this could greatly increase the fidelity. Our LLM will have an in-depth conversation with your LLM. For everyone. Most people probably shudder at the idea of an LLM rendering a judgement on them, but it is already happening in many interview processes today based on the tiny data in resumes. Better data helps everyone except the people trying to con their way into a position, and is it really worse than being judged by random HR people? Candidates with extensive public works, whether open source code, academic papers, long form writing, or even social media presence, already give a strong signal, but most talent is not publicly visible, and even the most rigorous (and resource consuming!) Big Tech interview track isn’t as predictive as you would like. A multi-year chat history is an excellent signal. Taken to the next level, you could imagine asking “What are the best candidates in the entire world that we should try to recruit for this task?” There is enormous economic value on the table in optimizing the fit between people and jobs, and it is completely two-sided, benefitting both employers and employees.

After nine years of development, meshoptimizer has reached its first major version, 1.0! This release focuses on improvements in clusterization and simplification as well as stabilization. Release announcement with more details on past, present and future is below; please RT!