A couple million lines of Haskell: Production engineering at Mercury

246 unignorant 2026-05-03 00:20:06 UTC blog.haskell.org Llama Share 105

Ian Duncan, a Mercury engineer, discusses how Haskell helps maintain a large and complex codebase by encoding operational knowledge into types, making the safe path easy and the wrong thing difficult. He shares lessons learned from Mercury's experience with Haskell, including the importance of containing mutation, encoding incantations in types, and using durable execution frameworks like ...

This Month in Ladybird – April 2026

354 richardboegli 2026-05-02 21:20:06 UTC ladybird.org Llama Share 78

Ladybird browser made significant improvements in April, including faster JS-to-JS calls, improved HTML parsing, and enhanced UI features. The browser also fixed various bugs and added new features, such as a new Linux frontend, improved caching, and better performance on various websites.

Dav2d

490 dabinat 2026-05-02 18:20:06 UTC code.videolan.org Llama Share 135

Anubis is a server protection system using Proof-of-Work to prevent aggressive AI scraping. It requires modern JavaScript features, so disable JShelter or similar plugins for this domain.

Six Years Perfecting Maps on WatchOS

312 valzevul 2026-05-02 21:20:06 UTC david-smith.org Llama Share 73

The author spent six years developing a custom mapping experience for the Apple Watch, culminating in Pedometer++ 8's watchOS mapping support.

Unverified Evaluations in Dusk's PLONK

25 deut-erium 2026-05-03 07:20:06 UTC osec.io Llama Share 3

A critical soundness vulnerability was found in dusk-plonk, a PLONK implementation powering Dusk Network's ~$60M market cap. The vulnerability allowed a malicious prover to forge verifying proofs for arbitrary false statements, bypassing every constraint in the transaction circuit.

Show HN: Apple's Sharp Running in the Browser via ONNX Runtime Web

7 bring-shrubbery 2026-05-03 09:20:06 UTC github.com/bring-shrubbery Prefix Share 0

Enough disk space and RAM for the SHARP model (the exported ONNX sidecar is large, ~2.4 GB) For that reason, the app uses the hosted model by default. Uploading only the .onnx file directly in the browser usually will not work because the .onnx.data sidecar is separate. The easiest route is to follow the upstream SHARP setup first, then run this exporter script from this repo. The app runs ...

Windows quality update: Progress we've made since March

54 jovial_cavalier 2026-05-03 05:20:06 UTC blogs.windows.com Llama Share 97

The Windows Insider Program is improving with more transparency, control, and fewer disruptions, including changes to the navigation, Feedback Hub, and Windows Update. New features and improvements are being rolled out, such as unified updates, improved pause controls, and reduced AI presence, with a focus on performance, reliability, and user experience.

Do_not_track

350 RubyGuy 2026-05-02 18:20:06 UTC donottrack.sh Llama Share 113

Users can opt out of telemetry data collection by setting a standard environment variable. This variable should be checked by tools that collect telemetry or make non-essential network requests.

Neanderthals ran 'fat factories' 125,000 years ago (2025)

199 andsoitis 2026-05-02 21:20:06 UTC universiteitleiden.nl Llama Share 97

Neanderthals in central Germany 125,000 years ago extracted fat from bones by crushing and heating them, indicating complex resource management. This discovery pushes back the timeline for this behavior by tens of thousands of years, showing Neanderthals were adaptable and efficient in their food strategies.

Specsmaxxing – On overcoming AI psychosis, and why I write specs in YAML

135 brendanmc6 2026-05-03 07:20:06 UTC acai.sh Llama Share 131

The user created Acai.sh, a tool for writing and tracking feature specs, to improve software development and collaboration. It uses a feature.yaml format and ACIDs (Acceptance Criteria IDs) to reference requirements in code and tests.

Group averages obscure how an individual's brain controls behavior: study

13 hhs 2026-05-03 08:20:06 UTC med.stanford.edu Llama Share 0

Stanford Medicine research found individual brain patterns differ from group averages in children's brain scans. Analyzing individual brains provides insights into cognitive differences and may help develop better therapies for conditions like ADHD.

Sourcefeed – a pop-up RSS service

26 bjhess 2026-05-03 09:20:06 UTC sourcefeed.app Full Share 7

A standalone feed. No website, no newsletter. Just your words, delivered to your readers’ rss reader.

Care homes and hotels in Japan shut as expansion strategy unravels

46 mikhael 2026-05-03 02:20:06 UTC newsonjapan.com Llama Share 15

A hotel in Choshi, Chiba Prefecture, has remained closed since late 2025 due to an investigation into its operator's business model, which allegedly ties visa acquisition for Chinese investors to facility purchases. At least 24 facilities acquired by the operator are now shut or out of business.

Inventions for battery reuse and recycling increase seven-fold in last decade

206 JeanKage 2026-05-02 16:20:06 UTC epo.org Llama Share 18

Global battery demand is surging, with 14 million end-of-life batteries expected by 2040. Europe is positioning itself to lead the circular battery economy with innovation and policy support.

VS Code inserting 'Co-Authored-by Copilot' into commits regardless of usage

1234 indrora 2026-05-02 20:20:06 UTC github.com/microsoft Llama Share 641

User is trying to apply a suggestion to a GitHub conversation but is encountering various errors such as no code changes, deleted lines, and pull request queue. The suggestion cannot be applied at this time due to these restrictions.

Thoth – open-source Local-first AI Assistant

3 sydsachar 2026-05-03 09:20:06 UTC github.com/siddsachar Prefix Share 0

Thoth is a local-first AI assistant for personal AI sovereignty: a desktop agent with memory, tools, workflows, design creation, messaging, plugins, and optional cloud models while your durable data stays on your machine. It runs fully local through Ollama with 39 curated tool-calling models, or you can opt into OpenAI, Anthropic, Google AI, xAI, OpenRouter, and ChatGPT / Codex when you want ...

The agent harness belongs outside the sandbox

111 shad42 2026-05-02 22:20:06 UTC mendral.com Llama Share 82

The agent harness runs outside the sandbox for multi-user agents, allowing shared skills and memories in a database, and uses virtualization to route tool calls to the sandbox or database. This approach keeps the API surface the same as the model was trained on and avoids the need for custom tools like memory_read.

Clojurists Together – Q2 2026 Open Source Funding Announcement

112 dragandj 2026-05-02 22:20:06 UTC clojuriststogether.org Llama Share 12

Clojurists Together is funding 5 projects in Q2 2026, including improvements to Malli, SciNoj, Clojure LLM, Gloat, and PluMCP. These projects aim to enhance performance, usability, and functionality in various areas such as data analysis, AI, and native compilation for Clojure.

A more efficient implementation of Shor's algorithm

86 signa11 2026-05-02 23:20:06 UTC lwn.net Llama Share 22

Researchers at Google, UC Berkeley, and other institutions created a quantum circuit that can factor 256-bit elliptic-curve signatures using fewer qubits and gates than previous methods. They published a zero-knowledge proof that they know the circuit, but not the actual circuit itself.

Show HN: State of the Art of Coding Models, According to Hacker News Commenters

117 yunusabd 2026-05-02 22:20:06 UTC hnup.date Llama Share 59

A daily update captures Hacker News comments on AI-assisted coding models, selecting relevant posts and analyzing sentiment with Gemini. Results are logged to a Google Sheet for auditing and debugging.

Because it doesn't have to

55 zdw 2026-05-03 02:20:06 UTC blog.computationalcomplexity.org Llama Share 13

The internet stack's lower layers don't guarantee delivery, allowing simpler protocols like TCP to restart communication when it fails. Machine learning models can also make probabilistic guesses to solve complex problems, allowing for flexibility and better performance.

Investors pile into clean energy as Iran war drives push for energy security

12 JumpCrisscross 2026-05-03 10:20:06 UTC ft.com Full Share 0

The following information can help our support team to resolve this issue.

How fast is a macOS VM, and how small could it be?

249 moosia 2026-05-02 11:20:06 UTC eclecticlight.co Llama Share 93

A macOS VM on a Mac mini M4 Pro with macOS 26.4.1 runs at 98% of host speed, with GPU performance at 95%. A MacBook Neo can run a usable macOS VM with 2 virtual cores and 4 GB of memory.

A physics engine with incremental rollback for multiplayer games

90 BSTRhino 2026-05-02 20:20:06 UTC easel.games Llama Share 31

Easel's new custom-built physics engine supports large worlds by only snapshotting and rolling back changed objects. This allows for more complex games with thousands of objects.

Maryland to ban A.I.-driven price increases in grocery stores

166 doener 2026-05-03 02:20:06 UTC nytimes.com Full Share 127

Please enable JS and disable any ad blocker

AI, Intimacy, and the Data You Never Meant to Share

34 victorkulla 2026-05-03 03:20:06 UTC fshot.org Llama Share 2

Artificial intelligence is quietly entering intimate spaces with connected devices that learn user preferences through bio-feedback sensors. This raises concerns about data storage, access, and handling of intensely personal biometric information.

Windows API is Successful Cross-Platform API (2024)

81 phendrenad2 2026-05-03 03:20:06 UTC retrocoding.net Full Share 83

We're verifying your browser Website owner? Click here to fix

Systemd-manager-TUI: A TUI application for managing systemd services

7 thunderbong 2026-05-03 10:57:06 UTC github.com/matheus-git Prefix Share 1

This tool allows you to manage systemd services with ease. You can view logs, list services, view properties, edit unit files, and control their lifecycle: start, stop, restart, mask, unmask, enable, and disable using the D-Bus API. It also supports Vim-like navigation. It is possible to navigate between system and user units and choose to list only running services or all units. ...

Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge

298 bazlightyear 2026-05-03 04:20:06 UTC thinkpol.ca Llama Share 162

The AI Coding Contest's Word Gem Puzzle challenge saw Kimi K2.6 from Moonshot AI win with 22 match points, followed by MiMo V2-Pro from Xiaomi and GPT-5.5. The results showed a significant gap between the top two models and the rest, with Kimi's aggressive sliding strategy proving effective on larger grids.

The USB Situation

131 herbertl 2026-05-02 11:20:06 UTC randsinrepose.com Llama Share 162

I use a MacBook Pro with a Studio Display, connecting via a braided cable that handles video, power, and other Display connections. The cable's speed is limited by the port type, not the cable itself, so I recommend using a Thunderbolt 5 cable for faster speeds.