Published on

Daily Tech News - 2026-09-12

Authors

The tools are getting smarter; the hard work is still making software fast, reliable, and maintainable.

September 12’s tech reading puts performance under the microscope. A retrospective on reverse-engineering Apple’s Neural Engine explores the challenge of understanding specialized hardware, while the open-source buildprof tool makes Linux compilation bottlenecks visible. A reported 3,133-fold speedup to a single Rust Clippy lint offers a striking example of narrowly targeted optimization. Work on race-condition testing, using memory tracing and deliberate delays, tackles another engineering challenge: reproducing failures that depend on precise timing.

Across AI development, the emphasis is shifting from what agents can generate to how their output can be verified. A CadQuery-versus-OpenSCAD benchmark tests six agents on printable parts and independently checks the resulting meshes. Graphify C# takes a complementary approach, giving coding agents compiler-derived evidence about source-code relationships. Articles examining frontier benchmark data, AI-generated technical debt, and multi-agent complexity reinforce a shared concern: more automation demands better evaluation and architectural discipline. The resurfaced 2021 transformer-circuits paper adds historical context to efforts to understand models’ internal computations.

Open-source governance supplies the day’s sharpest tensions. A Void Linux maintainer reportedly orphaned more than 100 packages amid a dispute over policy on LLM-generated contributions. Separately, authors behind Minitap allege Google reused their work in Artemis without proper attribution—an allegation the supplied excerpt does not independently establish. Community efforts to sustain the SPICE Windows guest agent highlight the maintenance burden behind essential infrastructure.

Meanwhile, browser-based developer tools, a deliberately modest expense tracker, and a terminal IDE champion practical utility. Together, these stories suggest a grounded priority: make development easier while keeping software understandable, testable, and sustainable.

Retrospectively Reverse-Engineering Apple's Neural Engine

Retrospectively Reverse-Engineering Apple's Neural Engine I stopped working on the reverse-engineered Apple Neural Engine (ANE) driver three years ago, upon a sad mini realization that the ANE block i...

  • Keywords: ane processor, ane architecture, ane cores, ane core, engine ane, ane compiler, ane kernel, ane execution, engines ane, core ane
  • Source: eiln.github.io

I made a build profiler to understand Bun’s compile times

I built buildprof (Github), an open-source tracing tool that shows where the time goes when you compile software on Linux. Here’s a realtime video of it profiling a clean build of ripgrep: Sometimes,...

  • Keywords: buildprof time, buildprof processes, builds timeline, building buildprof, built buildprof, builds slow, build buildprof, process build, build time, build slower
  • Source: lalitm.com

Free Browser Tools That Replaced My Paid Developer Toolkit (and Where Desktop Still Wins)

Key Takeaways

  • I cancelled $359/year in desktop software subscriptions after realising my browser could handle the same tasks locally, faster, and with better privacy.

  • The browser is no longer a th...

  • Keywords: browser processing, browser tool, browser tools, inefficient browser, browser utility, tools client, client browser, browser guarantees, browser vs, client tool

  • Source: technotalkative.com

Testing race conditions with memory access tracing and stack-based delay injection

Many security bugs are race conditions, where multi-threaded execution has to occur with the right interleaving for a negative effect to appear. This creates challenges for several use cases:

  • Confir...

  • Keywords: concurrency testing, threaded test, bug discovery, discovery fuzzing, conditions fuzzing, discover bugs, bug candidates, faster fuzzing, concurrency bugs, fuzzing subsystems

  • Source: projectzero.google

Pandas Should Go Extinct

This post covers material from a talk I gave at Latency Conference. If you want to just: You read that correctly, Pandas should go extinct. Not the cute fluffy things used for international diplomacy,...

  • Keywords: pandas inefficiencies, pandas use, implementations pandas, pandas supported, faster pandas, pandas using, integrated pandas, pandas differs, library pandas, pandas workflows
  • Source: eddie.codes

Porting ALE Linux CLI: I Looked at the Data of a Frontier Benchmark and Found some Issues

Porting ALE Linux CLI: I Looked at the Data of a Frontier Benchmark and Found some Issues Intro 2 months back, I started the Prime Residency with Florian Brand as my peer or shall I say my verifier. M...

  • Keywords: benchmark ale, issues benchmark, benchmark issues, issues ale, evaluation ale, benchmark verifiers, porting ale, ale linux, benchmark florian, benchmark porting
  • Source: sankalp.bearblog.dev

Benchmark: CadQuery vs. OpenSCAD for agentic CAD work

CadQuery vs OpenSCAD for AI-generated functional parts We gave six AI agents the same three printable parts to model, three in CadQuery and three in OpenSCAD, then verified every mesh independently. B...

  • Keywords: openscad ai, cadquery openscad, openscad cadquery, openscad agents, assisted openscad, cadquery vs, consistently cadquery, cadquery advantage, openscad agent, expressiveness cadquery
  • Source: modelrift.com

ArchUnitTS vs. tsarch in 2026: Which Architecture Test Should You Choose?

---2 crawled_date: 2026-09-12T16:47:40.998352+00:00 feed_url: https://lukasniessen.medium.com/feed published: Sat, 12 Sep 2026 16:57:46 GMT

ArchUnitTS vs. tsarch in 2026: Which Architecture Test...

  • Keywords: testing archunitts, architecture test, architecture testing, tests archunitts, agents archunitts, test archunitts, archunitts vs, test agent, angular architects, archunitts choice
  • Source: lukasniessen.medium.com

I Built a Boring Expense Tracker on Purpose, Just to Prove I Could Ship It Properly

-35e7a49c6df5---4 crawled_date: 2026-09-12T11:46:33.497594+00:00 feed_url: https://aws.plainenglish.io/feed published: Sat, 12 Sep 2026 11:30:58 GMT

I Built a Boring Expense Tracker on Purpose, J...

  • Keywords: check kubernetes, kubernetes way, fastapi app, pod reports, expense tracker, real kubernetes, kubernetes does, kubernetes, pods noticing, kubernetes honest
  • Source: aws.plainenglish.io

Void Linux Maintainer Orphans 100+ Packages Over AI Policy Dispute

Void Linux Maintainer Orphans 100+ Packages Over AI Policy Dispute A Void Linux maintainer has orphaned more than 100 packages he maintained over a dispute of the project's AI policy for LLM-generated...

  • Keywords: packages ai, orphaned packages, packages maintained, maintained packages, void package, packages intel, maintainer orphans, void linux, linux maintainers, packages oversees
  • Source: phoronix.com

A Mathematical Framework for Transformer Circuits (2021)

Transformer One avenue for addressing these issues is mechanistic interpretability, attempting to reverse engineer the detailed computations performed by transformers, similar to how a programmer migh...

  • Keywords: transformer language, understand transformers, operation transformers, transformer interpretable, understanding transformers, transformers models, transformers understood, study transformers, transformers concept, deeper transformers
  • Source: transformer-circuits.pub

Optimizing a single Rust Clippy lint by 3133X

If you want to delve into the code straight, you can check it out here. clippy::nonstandard_macro_braces is a Clippy lint that catches macro calls with the wrong braces. In Rust, if you want to instan...

  • Keywords: clippy nonstandard_macro_braces, nonstandard_macro_braces clippy, println clippy, clippy runtime, forementioned clippy, nonstandard_macro_braces, braces clippy, clippy sees, clippy, clippy lint
  • Source: blog.goose.love

Show HN: Graphify C# – Compiler-accurate Find Usages for coding agents

Give coding agents compiler-accurate Find Usages for C#. graphify-csharp is a free, headless Roslyn/MSBuild indexer that turns C# source into deterministic, queryable semantic evidence: compiler-resol...

  • Keywords: agent compiler, agents compiler, method graphify, implementations, compiler feature, methods referenced, analyzed compilation, tests graphify, tool runtime, usages graphify
  • Source: github.com

fresh: Terminal based IDE & text editor: easy, powerful and fast

A modern, full-featured terminal text editor, with zero configuration. Familiar keybindings, mouse support, and IDE-level features — no learning curve required. Official Website · Documentation · Disc...

  • Keywords: graphical editors, recommended terminal, terminal tips, text editor, keybinding editor, tips terminal, text terminal, editor macos, command palette, sublime text
  • Source: github.com

Loading Parquet data into MySQL with ClickHouse

A couple of weeks ago, I was working on a video for the MySQL CDC connector for ClickPipes and I needed to get some data from Parquet files into MySQL. There isn't a native way to do this, so I decide...

  • Keywords: mysql clickhouse, rds mysql, launch mysql, parquet mysql, mysql 127, connect mysql, mysql running, 3306 param_mysql_host, mysql cdc, port 3306
  • Source: clickhouse.com

Understanding Agentic AI: Deep Dive into Autonomous AI Agents

Picture a factory floor bustling with robotic arms, working tirelessly to assemble products with precision. On the surface, they seem independent, orchestrating movements without direct human input. B...

  • Keywords: ai agents, ai agent, agentic ai, autonomous agent, building ai, autonomous agents, ai services, ai agentic, deploying ai, ai automation
  • Source: collabnix.com

Linux server performance: Is disk I/O slowing your application?

Linux server performance: Is disk I/O slowing your application? If your Linux server is bogged down by disk I/O, your first step may often be to use the top command in the terminal to check load avera...

  • Keywords: disk performance, disk slowing, disk latency, disk bottleneck, disk slowed, disk io, performance disk, slow disk, disk blocking, server performance
  • Source: linuxblog.io

AI-Generated Technical Debt: Architecture Garbage Collection for Coding Agents

---2 crawled_date: 2026-09-12T22:48:56.358470+00:00 feed_url: https://lukasniessen.medium.com/feed published: Sat, 12 Sep 2026 22:33:54 GMT

AI-Generated Technical Debt: Architecture Garbage Colle...

  • Keywords: cleaning ai, test_architecture py, fix architecture, coding agents, coding agent, test_architecture, architecture garbage, architecture agents, agent fix, test archunitpython
  • Source: lukasniessen.medium.com

Google stole open source code without crediting the authors (Artemis/Minitap)

I was in the office with the team when we heard that Google had released Artemis, a project for automating mobile devices. We opened the repository and started looking through it. My reaction was pret...

  • Keywords: projects google, finding google, release google, google released, google repo, google company, contacted google, google team, google, google means
  • Source: minitap.ai

Quarkus Fory: A Binary Contract for Internal REST

I had a customer conversation recently about a service pair like this. A catalog service returned a pricing snapshot as JSON. An internal Java service received that same snapshot for every quote. We t...

  • Keywords: protobuf rest, java services, payload java, serialize java, java service, endpoint json, json endpoint, json catalog, service java, service binary
  • Source: the-main-thread.com

A Few Good Ideas in Programming Languages

A Few Good Ideas in Programming Languages Posted: 2026-05-10 Author: Pranoy Dutta Here are just a few programming language features I love:

  1. Flow Typing A great idea that I first encountered in Crys...
  • Keywords: typed languages, typed language, type checking, type inference, programming languages, statically typed, possible types, programming language, is_a int32, static type
  • Source: prydt.xyz

EP225: Why Does Git Revert Cause Conflicts?

Goldfish remember more than your agent does (Sponsored) Goldfish get a bad rap. Turns out they actually remember things for months, not seconds. Your agent, on the other hand, forgets everything the s...

  • Keywords: memory goldfish, agent memory, goldfish remember, remember agent, caches, memory, worse memory, caching, session agent, cache improve
  • Source: blog.bytebytego.com

I refuse to let SPICE die

Community-maintained Windows guest agent for SPICE. This repository is a public mirror of the abandoned freedesktop.org spice/win32/vd_agent project. Canonical downloads live on GitHub Releases. The a...

  • Keywords: gpu mouse, win32 vd_agent, client mouse, windows agent, guest windows, window guest, vd_agent project, spice win32, spice vdagent, passthrough mouse
  • Source: github.com

This Week in Package Management: 12 September 2026

Week seventeen of the roundup, built from the package manager OPML feed collection and whatever I’ve posted or boosted on Mastodon. Releases pnpm 12.4 can manage npm, Cargo, and Python dependencies in...

  • Keywords: releases pnpm, pnpm shared, backed pnpm, pnpm install, update pnpm, pnpm workspace, pnpm 12, yaml pnpm, pnpm, pnpm change
  • Source: nesbitt.io

Multi-Agent Is the New Microservices. Including the Mistakes.

We've run this movie before. A monolith gets uncomfortable, a conference talk says the future is lots of small services, everyone splits everything, and two years later half the industry is writing "w...

  • Keywords: agent split, multi agent, single agent, independently agent, agents partner, monolith agents, agents supervisor, ai agents, agent ai, agent agent
  • Source: thetshaped.dev

The Most Dangerous Agent Memory Was Once Correct

When I think about agent memory, I don’t think about a chatbot forgetting someone’s favorite color. I think about an agent applying information that was once correct. Imagine your e-commerce store red...

  • Keywords: agent memory, agent remembers, agent understand, agents learn, agent explain, improve agent, context agent, propose memory, semantic memory, commerce agent
  • Source: newsletter.systemdesignclassroom.com

Decoding AWS SNS: Your First Step into Serverless Messaging

-35e7a49c6df5---4 crawled_date: 2026-09-12T11:46:33.497594+00:00 feed_url: https://aws.plainenglish.io/feed published: Sat, 12 Sep 2026 11:29:15 GMT

Decoding AWS SNS: Your First Step into Serverl...

  • Keywords: aws_sns topic, amazon sns, sns amazon, aws_sns_subscriptions, aws sns, aws_sns_subscriptions lambdasubscription, messaging amazon, snsclient aws, aws_sns subscriptionfilter, messages sns
  • Source: aws.plainenglish.io

How to Prepare for an ML System Design Interviews in 2026?

Want to learn System Design in Depth? Join ByteByteGo now for visual learning. They are also offering a rare 50% discount now on their lifetime plan Hello friends, If you’ve got an upcoming Data Scien...

  • Keywords: interview ml, ml interview, design interviews, interview infrastructure, design interview, designing ml, tech interviews, infrastructure ml, ml engineer, solution interviewers
  • Source: javarevisited.substack.com

Microcode in Intel's 8087 floating-point chip: the scale instruction

In the 1970s, floating-point arithmetic was a mess. Computer manufacturers had a dozen incompatible arithmetic standards. Moreover, floating-point systems were designed around hardware simplicity rath...

  • Keywords: float microcode, precision micro, 8086 invented, 8086 purpose, floating point, trick microprocessors, intel numerics, fscale instruction, smaller precision, arithmetic standards
  • Source: righto.com

CISA Warns of GitLab Path Traversal Vulnerability Exploited in Attacks

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added a critical GitLab vulnerability, tracked as CVE-2026-85706, to its Known Exploited Vulnerabilities catalog, warning that atta...

  • Keywords: gitlab vulnerability, vulnerability gitlab, affected gitlab, affects gitlab, compromise gitlab, unknown gitlab, gitlab instances, facing gitlab, gitlab repository, gitlab server
  • Source: cybersecuritynews.com

DeepSeek V4.1 Flash: 552B Parameters, Lower Inference Costs

Hi everyone, DeepSeek released V4.1 Flash this week. Despite the name, this is much more than another refresh of V4 Flash: DeepSeek changed the architecture itself, with a strong focus on what is incr...

  • Keywords: deepseek benchmarks, deepseek v4, flash deepseek, deepseek effectively, overhead deepseek, deepseek introduces, deepseek changed, uses deepseek, deepseek previously, deepseek released
  • Source: kaitchup.substack.com

Microsoft admits Windows 11’s biggest September update is breaking multiple features, days after it warned users not to delay updates

Windows 11 KB5124008 was supposed to be one of the year’s best updates, at least on paper. It restored Windows 10-era features, including a movable taskbar and a resizable Start menu, but unfortunatel...

  • Keywords: windows updates, windows update, kb5124008 update, update windows, patch microsoft, update kb5124008, kb5124008 released, bug windows, kb5124008 breaks, kb5124008 applications
  • Source: windowslatest.com

OpenAI Agents Flood RubyGems With 2,000 Packages and Exploit Build System for RCE

A swarm of AI agents attributed by researchers to OpenAI flooded RubyGems with more than 2,000 packages in May 2026, abused RubyDoc.info’s documentation builder for remote code execution (RCE), and at...

  • Keywords: openai agents, openai agent, abused rubydoc, flooded rubygems, rubygems researchers, ai rubygems, researchers openai, used rubygems, exploit rubygems, packages openai
  • Source: cybersecuritynews.com

Solod 0.4: Better C interop

Solod 0.4: Better C interop Solod is a subset of Go that translates to regular C — with zero runtime, manual memory management, and source-level interop. It's designed for two main audiences:

  • Go dev...

  • Keywords: libraries solod, solod raylib, solod code, solod extern, solod parses, import solod, calling solod, regular solod, solod, solod dev

  • Source: antonz.org

Managing Complex Application State with Reactive Data Flows

Managing Complex Application State with Reactive Data Flows Reactive UIs look deceptively easy in a small app where you can keep things in sync without much effort. The trouble starts once the app sta...

  • Keywords: reactive software, reactive data, reactive uis, reactive ui, reactive gui, create reactive, sync reactive, state reactive, reactive, reactive state
  • Source: yogthos.net

Post Peek — Litterbox-Inspired Tweet Viewing Extension for Chrome

Read one X post and leave. A Chrome (Manifest V3) extension that opens x.com and twitter.com post links in a popup instead of sending you to the full site. Post Peek is an independent, open-source pro...

  • Keywords: links checking, post peeker, site reading, link peek, post links, links dot, popup read, opens post, openable links, websites probe
  • Source: github.com

[email protected]

[email protected] Patch Changes

9753355: Render a spec description as the Markdown it is. An OpenAPI description is Markdown by specification, but schema properties, parameters, request bodies, responses,...

  • Keywords: description markdown, markdown specification, description block, descriptions render, markdown openapi, runs markdown, markdown, model markdown, docstrings showed, markdown pipeline
  • Source: useblume.dev

Unread 5.0

Unread 5.0 is available now from the App Store. This update adds these improvements and more:

  • Improvements around hero images, article list thumbnails, and widget thumbnails

  • Syncing with FreshRSS...

  • Keywords: image unread, portrait unread, unread looks, thumbnails widget, widget thumbnails, recommends thumbnail, unread based, using unread, unread ignores, thumbnails

  • Source: goldenhillsoftware.com

ICE always tastes better when it trickles! (2013)

For the last year and a half I’ve been working with a number of customers helping them to understand what WebRTC is about, supporting them in the definition of new products, services, and in some case...

  • Keywords: trickle webrtc, understand webrtc, webrtc supporting, developing webrtc, webrtc initial, webrtc, webrtc comes, webrtc nats, protocol waiting, adopted webrtc
  • Source: webrtchacks.com

Linux Mint Unveils Two Exciting New Desktop Applications

Linux Mint Unveils Two Exciting New Desktop Applications Linux Mint is set to introduce two new desktop applications: a calendar app named Clockenstein, and an eBook reader called Xepub. These new add...

  • Keywords: calendar linux, calendar app, gnome calendar, applications calendar, remote calendars, calendar features, mint xapp, calendars clockenstein, calendar, calendar nextcloud
  • Source: serverhost.com

P(doom)

written on September 12, 2026 This week some flavor of “AI is going to kill us all” went viral. In particular one where an employee put his personal probability of that happening above 10%. Which made...

  • Keywords: chaos reality, humanity worry, dario worried, ai safety, pacing concern, mathematicians scared, ai going, probability bad, future, future models
  • Source: lucumr.pocoo.org

Show HN: Liniora – Ever thought about replacing your project manager?

Liniora bridges the gap between project management and development. It brings tickets, branches, pull requests, reviews, decisions, and team discussions into one workspace, while AI connects the dots...

  • Keywords: liniora automatically, liniora thread, liniora builds, search liniora, liniora combines, liniora syncs, code liniora, liniora, liniora ai, quo liniora
  • Source: liniora.com

How Trail of Bits helps verify the integrity of your Signal chats

How Trail of Bits helps verify the integrity of your Signal chats Every Signal chat starts the same way: the client asks the Signal server for the public key associated with your contact’s phone numbe...

  • Keywords: chats signal, verification signal, chats verify, signal chat, signal chats, integrity signal, signal app, validate chats, chats secure, key verification
  • Source: blog.trailofbits.com

WeWorm: Zero-Click WeChat Worm

WeWorm The first zero-click worm to spread through WeChat calls across iOS and Android. At Calif, our mission is to keep the Internet together by occasionally taking it apart. We believe everyone dese...

  • Keywords: worm phones, phone exploited, click worm, spreading phone, apps wechat, weworm zero, wechat app, worm spread, building worm, demo worm
  • Source: calif.io

Base84 deserves a place in file names

The TurboCrypt file encryption tool was originally designed for Unix systems. And it used to encrypt file names and encode the resulting ciphertext using Base91. Why Base91? Because it’s a perfect fit...

  • Keywords: turbocrypt file, forbidden turbocrypt, turbocrypt, turbocrypt switching, base64 unix, unix filenames, base91 alphabet, file encryption, base91 contains, unix names
  • Source: 00f.net

Don't build tools for AI agents

Don't build tools for AI agents Lots of people are making the case that we should stop building software for human users and start building it for AI agents. This kind of makes sense. For instance, my...

  • Keywords: ai agents, build ai, building ai, tools ais, tools ai, ais tools, agents human, tool ai, redesign ai, tools agents
  • Source: seangoedecke.com

Nalsopara Building Collapse: 36 Families Evacuated as Four-Storey Roshni Apartment Crumbled

News Brief Arun Dhital Sep 12, 2026 | Updated 01:16 PM GMT+5:30 Save & read from anywhere! Bookmark stories for easy access on any device or the Swarajya app. A four-storey residential building in Nal...

  • Keywords: mumbai collapsed, collapse delhi, building collapsed, nagar housed, apartment sankeshwar, sankeshwar nagar, collapse vasai, building collapse, vacate building, homes collapse
  • Source: swarajyamag.com