Top 29 stories from Hacker News. Top 10 include comment highlights. Compiled at 20:10 UTC.
364 points by predkambrij · 199 comments
What HN said:
aaddrick: Hey! I manage the unofficial build at https://github.com/aaddrick/claude-desktop-debian Debian is in the name, but scope has grown to all backends, compositors, etc. The main reason must companies don't publish Linux electron apps is fragmentation.
Retr0id: If only Anthropic had some kind of automated tool that was good at porting software
splwjs: What's the market for linux users who want an electron app so they can vibecode in a visual studio fork but wont just build it themselves nor do they want to clone and build someone else's repo
shanewei: What do you miss from the Desktop app that the CLI doesn’t cover? I’m mostly on Linux too and have just been using the CLI, so I’m curious.
334 points by matt_d · 77 comments
What HN said:
haunter: The GameBoy emulator's code also looks like the GameBoy. Slow clap this is insane, definitely my favorite entry. https://github.com/ioccc-src/winner/blob/master/2025/ncw1/pr... The author, Nick Craig-Wood, is the creator of rclone!
s-macke: My favorite is the 366-byte C program emulator that can run Linux and Doom [0]. The VM implements an OISC - a One Instruction Set Computer [1]. [0] https://github.com/ioccc-src/winner/blob/master/2025/cable/p... [1] https://github.com/ioccc-src/winner/blob/master/2025/cable/R...
yayitswei: In case anyone was wondering, the IOCCC specifically permits LLM use in their guidelines. "The IOCCC has a rich history of remarkable winning entries created by authors who skillfully employed various techniques (often their own tools) to develop their code."
aquir: The website itself is obfuscated, it’s not easy to find the C sources at all!
223 points by davidbarker · 31 comments
Explore our hand-picked collection of out-of-copyright works, free for all to browse, download, and reuse. This is a living database with new images added every week.
What HN said:
rectang: There are lots of sites that provide images that somebody has claimed are public domain. But for significant use, you what you really need is provenance documentation. These folks seem to be more up-front about the issue than many sites I’ve seen: https://pdimagearchive.
samcgraw: This is just the kind of thing I love about the open internet: the insane amount of art available in archives, (e.g. https://www.britishmuseum.org/collection/, https://www.rijksmuseum.nl/en, the Met, etc). Adding this to the list for one of my side projects[0].
neilv: Does anyone know how easily you can do "copyright clearance" for these supposedly public domain images? For example, the page for the first image I clicked on said: Date 1833 Underlying Rights Public Domain Worldwide Digital Rights No Additional Rights...
mmh0000: I do not know how this site managed to break mousewheel scrolling so badly, but I am quite impressed.
192 points by kristoff_it · 121 comments
What HN said:
alkonaut: > It doesn’t matter that the language you use is memory-safe, if you didn’t design for correctness or have no process that will eventually lead you to fixing all bugs. After many years in the business I have come to a more pragmatic view.
flooow: I imagine it's a difficult time to be a Zig developer. In the near term, Bun choosing to switch from Zig to Rust specifically to fix all the memory errors seems to have done the Zig community some psychological damage.
ozgrakkurt: It is really inspiring to see other people passionate about programming. Really love the people building zig. Especially Andrew and Loris and some other people I saw in codeberg
BirAdam: I have voiced a similar thought more succinctly: “it is your worst software that will live forever.” The implication is that you should always strive to release software that isn’t overly buggy, isn’t slow, and is general a pleasure to use.
173 points by supermatou · 76 comments
What HN said:
Aurornis: > We found that high-dose supplementation was positively associated with 3 of 11 functions assessed: verbal memory, visual memory, and flexibility or set shift. However, the association with flexibility or set shift was not significant after correction for multiple testing.
niteshpant: I have thoughts on this 1. Yes, vitamin D actually controls a lot of bodily functions it’s very easily set aside as not a “main” factor when in reality it actually controls a lot 2.
jbellis: > the association with flexibility or set shift did not remain significant after false discovery rate correction this is right there in the abstract, isn't that the entire game?
cortic: Vitamin D excess can lower blood levels of vitamin K2, reducing bone development. Not sure if the levels would be high enough for this (difficult to find safe numbers for in utero), but the side effects, joint pains, stiffness, could have motivated the kids to stay indoors and f...
168 points by devenjarvis · 36 comments
Generate hands-on, multi-part technical tutorials on demand, with LLM skills tuned to make content approachable. Then you work through them yourself, by hand ✋ - devenjarvis/lathe
What HN said:
Arubis: For a somewhat hybrid approach here, have a look at https://github.com/DrCatHicks/learning-opportunities — the idea is to be used during “productive work” (so it’s not purely learning-oriented as with your repo here), and to interject as you work to ensure that you learn related...
d4rkp4ttern: A related idea is to have the LLM quiz you, Socratic-style about a topic of interest. It persists in asking questions at deeper levels until you arrive at the answer yourself.
dchuk: I’ve been using this general pattern - a custom cli app for deterministic tasks, skills for the agent harness, run the skills in the agent and it produces artifacts for you by using the cli and its own agentic reasoning - a lot lately for work.
rdksu: I have updated the popular /grill-me skill for this exact purpose! I had a very insightful grilling session yesterday on what exactly happens when you try to load an extremely large dataset in pandas, covering everything down to the last detail !
140 points by mmastrac · 12 comments
Technical articles and release notes about Symbolica, symbolic computation, numerical methods, and high-performance computer algebra.
What HN said:
mkl: Past discussion (2 years ago, 119 comments): https://news.ycombinator.com/item?id=40297423 This software does symbolic maths, and it's commercially licensed. The site still seems to be lacking comparisons to other computer algebra systems.
adius: As an alternative, I’m working on reimplementing Wolfram Language/ Mathematica in Rust: https://woxi.ad-si.com/ A lot of Wolfram Language code just works already!
timschmidt: I've been doing some symbolica-like things recently in the https://github.com/timschmidt/hyperreal ecosystem. Not a full CAS, just enough symbolic math to maintain precision through the calculations. Benchmarks against Symbolica and numerica here: https://github.
lcnbr: I’ve been a rust user of symbolica since 0.1 and it is insane how much nicer it is to use now. Builder patterns for constructing replacement rules (and now evaluators!), macros for namespacing symbols, and now the call trait to fill in for callables in rust.
128 points by kkm · 27 comments
Lossless compression of a target model
What HN said:
oceanplexian: A lot of this is over my head but why would you do compression when GPU time is the most expensive thing in the world right now? KV can be trivially stored on ram or even a spinning disk and retrieved on the order of milliseconds. See LM cache for vLLM for example.
zozbot234: The problem with this approach is that even recomputing a "draft" of the KV cache is still quadratic in context length. Maybe you can get some constant savings by always recomputing the earliest tokens, but it's not a good tradeoff as context sizes grow.
hypfer: TL;DR (and please correct me if I got it wrong): Tiny deterministic model predicts the K/V cache, prediction is compared with reality, delta is stored in vram. The other way round then just predicts the values again, applies the delta, and you have the full correct value while ju...
0-_-0: You can use the original model to compress the kv cache and get ∞x compression, since the prediction is perfect. The cost is time, and I don't see how this could be worth it.
122 points by supermatou · 58 comments
What HN said:
kalleboo: I've been recently working with Classic Mac OS programming[0] and just that memory model (also using dealing with the lack of virtual memory using opaque handles to memory that need to be locked when used) is painful enough[1] - having to deal with segment addressing on top of th...
jdw64: Sometimes I think that if it were the old days, I probably wouldn't have been able to program. I remember that these days we program on top of 64bit virtual addresses, but how did developers do it back then
summa_tech: Pretty good detail in this article! But what really surprises me is how some ideas just keep coming back. When I wrote a binary translator, I ended up having to keep a translated return stack to optimize RET opcodes.
boutell: In 1994 I was 2 years out of school. I'd written one windows shareware application and a whole lot of unix-y things. People were excited about the internet but most people didn't have access. Unix shell accounts via dialup were common though.
76 points by daesorin · 5 comments
What HN said:
darcien: Anyone have context on why this announcement from 2025 is shared? Seems like it's delayed and have not been released yet? https://blog.podman.io/2026/05/podman-6-release-delayed/
freedomben: I love podman and use it extensively, but am less sure how useful the machine functionality will be (to me). I use KVM (with virt-manager) extensively, and I could see how treating VMs as ephemeral like containers might be nice for some workflows, but for me I want my VMs to be...
AI/ML
Other
Programming
Science
Stories and comments sourced from Hacker News public API. Not affiliated with Y Combinator or Hacker News.