Hacker News: hmry

New comment by hmry in "Dithered QR Codes"

hmry — Sun, 09 Aug 2026 04:47:25 +0000

That's absolutely hilarious

New comment by hmry in "Now is the time to give LLMs access to the ACM digital library"

hmry — Wed, 29 Jul 2026 01:30:08 +0000

This is about the ACM licensing it to for-profit AI companies, who will then make you pay a subscription to access the knowledge. Not about making it "freely available to the public."

New comment by hmry in "$100 AI Music Video: Claude Fable 5 vs. GPT-5.6 Sol"

hmry — Mon, 27 Jul 2026 09:07:08 +0000

FYI it's not possible to downvote replies to your own comment

New comment by hmry in "Memory safety absolutists"

hmry — Sun, 26 Jul 2026 07:12:59 +0000

Sorry for not being clear, I meant the creators of Fil-C and Zig (since those are the ones mentioned in the article)

New comment by hmry in "Stolen Buttons"

hmry — Sat, 25 Jul 2026 21:28:14 +0000

Awesome, the Haiku one is my favorite

New comment by hmry in "Memory safety absolutists"

hmry — Sat, 25 Jul 2026 21:16:14 +0000

It's annoying how many comment sections online are now just Rust vs Fil-C / Zig flamewars, and the creators of those languages are deliberately fanning the flames.

I also feel there's a second dimension to the politics, where Rust is the "woke" language and C / Zig / Odin are now the "anti-woke" languages. At least, going by their most vocal online communities.

I'm sure offline there's still engineering decisions being made (I hope).

New comment by hmry in "Learn OpenGL, extensive tutorial resource for learning Modern OpenGL"

hmry — Thu, 23 Jul 2026 19:22:45 +0000

Thanks, that makes sense

New comment by hmry in "Learn OpenGL, extensive tutorial resource for learning Modern OpenGL"

hmry — Thu, 23 Jul 2026 19:01:59 +0000

WebGPU only got started in 2021, so Im not sure what you mean. Could you explain? (Or are you thinking of WebGL?)

New comment by hmry in "Learn OpenGL, extensive tutorial resource for learning Modern OpenGL"

hmry — Thu, 23 Jul 2026 16:25:29 +0000

I do agree Vulkan is too cumbersome, but OpenGL is definitely obsolete.

The OpenGL API with its binding-based approach is also very confusing for beginners, but the replacement Direct State Access API came too late (after Apple already decided they would deprecate OpenGL, so it never got supported on macOS).

And the fact that the GLSL parser and compiler are part of the driver means there are tons of hardware-specific bugs and miscompilations. Intel integrated drivers on Windows are notorious for being especially buggy.

All of those make OpenGL a pretty poor target for learning. Personally, I would recommend WebGPU to people who want to get into graphics programming. It's very similar to DirectX/Metal, and like a more streamlined Vulkan. The WGPU implementation of WebGPU has a C API, which also has C++ headers, and a native Rust API. And you can also use WebGPU in your browser from JavaScript, if you don't know any native languages.

New comment by hmry in "Learn OpenGL, extensive tutorial resource for learning Modern OpenGL"

hmry — Thu, 23 Jul 2026 16:18:50 +0000

Mesa has Zink which translates OpenGL to Vulkan. AFAIK Asahi uses it because they only wrote a native Vulkan driver for the hardware, no native OpenGL driver.

Apple themselves also don't ship a native OpenGL driver anymore, only a layer that translates OpenGL to Metal.

New comment by hmry in "Advertise in ChatGPT"

hmry — Wed, 22 Jul 2026 07:15:28 +0000

"I will disclose, as a personal bias, that I hate ads."

"They fundamentally misalign a user's incentives with the company providing the service"

"Ads plus AI is sort of uniquely unsettling to me, I kind of think of ads as a last resort for us for a business model."

- Altman 2024

Soooo... No? Sort of the opposite, they confirmed they would do it. ;p

New comment by hmry in "Advertise in ChatGPT"

hmry — Wed, 22 Jul 2026 04:39:40 +0000

That's a shame, I love sarcastic posts. Although, when everyone is sarcastic all the time, it does severly degrade the discussion.

New comment by hmry in "SQLite should have (Rust-style) editions"

hmry — Thu, 16 Jul 2026 03:38:48 +0000

These proposed editions are per connection.

A system-wide config file in /etc that changes defaults for every program would break any program that assumes the old defaults.

It also wouldn't solve the problem of having to manually find out what the current recommended defaults are. With editions, you can simply enable the latest one and know you've got the right defaults.

New comment by hmry in "Duskers, the scary command line game, is getting a sequel"

hmry — Thu, 16 Jul 2026 03:28:01 +0000

It's only easy if you own a mac, otherwise it's somewhere between annoying and impossible.

Last time I tried it, you needed a mac if you wanted to:

- use AoT compilation (IL2CPP) instead of JIT

- use any native libraries (or use any assets that depend on native libraries)

- sign your executable

And that was before Apple Silicon, I would be surprised if it's gotten any easier

New comment by hmry in "RISC-V Is Inevitable: State of the Union Keynote Argues"

hmry — Wed, 15 Jul 2026 12:00:33 +0000

Non-US hedging will be big for RISC-V in the future, but the geopolitics can also cut the other way. E.g. US banning US companies from using Chinese RISC-V chips, to protect their domestic players. Especially now that intel is partially state-owned.

New comment by hmry in "A voxel Tokyo in real Japan time – ride the Yamanote line and study Japanese"

hmry — Mon, 13 Jul 2026 20:13:04 +0000

It's not running at 400 fps, the UI has the current global listener count next to a button that says "fps". The fps is only shown once you click that button. Weird design, I know.

New comment by hmry in "Can you build a recognizable World Map in under 500 bytes?"

hmry — Sun, 05 Jul 2026 01:44:41 +0000

Why make Claude play the guitar?

New comment by hmry in "Performance per dollar is getting faster and cheaper"

hmry — Sat, 04 Jul 2026 19:09:55 +0000

They're asking what precision the model parameters were quantized/rounded to, look up "LLM quantization" to learn more.

The old headline was something like "We served GLM5.2 on AMD MI355X at 2626 tok/s/node". They think it's bad to advertise performance numbers like that without specifying how much you quantized the model (Since you can 'cheat' more performance by quantizing more, and not mentioning the reduced output quality)

New comment by hmry in "Odin, Wikipedia and engagement farming"

hmry — Sat, 04 Jul 2026 03:41:07 +0000

Being fraudulent doesn't make something less notable. It might even make something more notable, provided enough sources report on it.

New comment by hmry in "Memorizing session transcripts isn't useful"

hmry — Fri, 03 Jul 2026 20:27:34 +0000

I have to imagine these types of features are for people who treat chatgpt like their friend/therapist/girlfriend/assistant/... instead of people who use it to answer questions