Hacker News: joerick

New comment by joerick in "Python's experimental JIT to be removed"

joerick — Wed, 28 Oct 2026 08:58:04 +0000

I only just saw it mentioned in this thread, but CinderX might suit you well.

New comment by joerick in "Using FireWire on a Raspberry Pi"

joerick — Thu, 26 Mar 2026 23:29:24 +0000

I have a Focusrite Saffire that has lots of nice, quiet preamps and ADC channels. I managed to get an old Mac to connect over FireWire and using the Focusrite Control app, I configured the routing to map all the analog inputs to ADAT. That works great but it's always at the wrong sample rate, and I can't change that without getting the old Mac out. Maybe I'll look into one of these rPi shields too. Anyone got any reverse engineering tips for the control protocol?!

New comment by joerick in "Python Steering Council unanimously accepts "PEP 810, Explicit lazy imports""

joerick — Mon, 03 Nov 2025 18:54:17 +0000

It should land in 3.15, so October next year. https://peps.python.org/pep-0790/

New comment by joerick in "Uncertain"

joerick — Fri, 29 Aug 2025 06:05:16 +0000

I've been wondering for a while if a program could "learn" covariance somehow. Through real-world usage.

Otherwise, it feels to me that it'd be consistently wrong to model the variables as independent. And any program of notable size is gonna be far too big to consider correlations between all the variables.

As for how one might do the learning, I don't know yet!

New comment by joerick in "Police identify seven as main suspects in Post Office Horizon scandal inquiry"

joerick — Fri, 27 Jun 2025 10:00:11 +0000

If we assume no conspiracy, what's the sentencing like for perjury due to carelessness?

New comment by joerick in "Introducing Gemma 3n"

joerick — Thu, 26 Jun 2025 21:23:49 +0000

Google do have an API for this. It has limits but perfectly good for personal use.

https://developers.google.com/custom-search/v1/overview

New comment by joerick in "Guess I'm a rationalist now"

joerick — Fri, 20 Jun 2025 22:42:51 +0000

Beautifully put.

New comment by joerick in "SeedLM: Compressing LLM Weights into Seeds of Pseudo-Random Generators"

joerick — Sun, 06 Apr 2025 10:27:54 +0000

Pretty fascinating from an information theory point of view. Surprising that it works at all. Is this, like, the JPEG of uniformly distributed, uncorrelated data?

New comment by joerick in "Blue noise – white noise alternative for graphics (2016)"

joerick — Fri, 14 Mar 2025 08:40:16 +0000

I came across this page in a discussion about dithering. It was posted by @rikroots.

Fascinating stuff. I love to see the application of the 2D Fourier transform for analysis. Added bonus - they tile brilliantly.

Blue noise – white noise alternative for graphics (2016)

joerick — Fri, 14 Mar 2025 08:40:16 +0000

Article URL: https://momentsingraphics.de/BlueNoise.html

Comments URL: https://news.ycombinator.com/item?id=43360664

Points: 3

# Comments: 1

New comment by joerick in "EA Open Sources Command and Conquer: Red Alert, along with other games"

joerick — Fri, 28 Feb 2025 10:07:01 +0000

Hahaha. We used to get desyncs on networked games of Generals pretty regularly. I remember if a game took more than 30-40 mins I'd start to get a spidey sense things were about to go wrong

New comment by joerick in "Embeddings are underrated"

joerick — Fri, 01 Nov 2024 12:50:20 +0000

Yeah, 'everything' as in 'everything that the model cares about' :)

New comment by joerick in "Embeddings are underrated"

joerick — Fri, 01 Nov 2024 12:49:36 +0000

Nice workaround. I just wish there was a less 'lossy' way to go about it!

New comment by joerick in "Embeddings are underrated"

joerick — Fri, 01 Nov 2024 12:48:03 +0000

Interesting, but your hypothesis assumes that 'tone' is one-dimensional, that there is a single axis you can remove. I think tone is very multidimensional, I'd expect to be removing multiple 'directions' from the embedding.

New comment by joerick in "Embeddings are underrated"

joerick — Fri, 01 Nov 2024 11:13:31 +0000

The thing that puzzles me about embeddings is that they're so untargeted, they represent everything about the input string.

Is there a method for dimensionality reduction of embeddings for different applications? Let's say I'm building a system to find similar tech support conversations and I am only interested in the content of the discussion, not the tone of it.

How could I derive an embedding that represents only content and not tone?

Pyinstrument 5 – Flamegraphs for Python

joerick — Fri, 11 Oct 2024 14:29:45 +0000

Article URL: https://joerick.me/posts/2024/10/3/pyinstrument-5/

Comments URL: https://news.ycombinator.com/item?id=41809781

Points: 2

# Comments: 0

New comment by joerick in "Orion, our first true augmented reality glasses"

joerick — Thu, 26 Sep 2024 06:39:27 +0000

The touchscreen might be better than you're giving it credit for. Direct interaction (vs the indirectness of mice, keyboards, gesture) is one benefit that I can't see done better any other way. I think people will always have some kind of handheld touchscreen as a result.

New comment by joerick in "Vue 3.5 released"

joerick — Wed, 04 Sep 2024 07:19:41 +0000

Biggest functional difference is that Vue's reactivity is built on Proxy objects, which lets you simply bind your view layer to your model layer and things just keep in sync, without any extra APIs to think about. React has useState which gives you a setter that you need to use with an entirely new object each time. In Vue you can just change a property deep inside the model layer and the view updates.

That's what's keeping me in Vue primarily. I think MobX does something similar in React but the API is clunkier than Vue.

On top of that, the API is just much nicer. None of the confusing functional language and concepts e.g. "what exactly did the programmer intend by this 'useEffect' call??"

New comment by joerick in "No One's Ready for This"

joerick — Fri, 23 Aug 2024 07:50:46 +0000

I've been thinking for a while that digital photos should include a signature derived from a hardware key on each camera. That at least would prevent people lying in metadata, and give us some means to verify. If I trust Canon's signature, I trust that these pixels were taken by a camera.

New comment by joerick in "Rye and Uv: August Is Harvest Season for Python Packaging"

joerick — Wed, 21 Aug 2024 23:28:59 +0000

...and a lot of solo developers and academics besides. (I say that as a corporately unaffiliated PyPA member)