C++20 Modules: Compile Time Improvements and Practical Experiences

2025-09-11

This article shares the author's practical experience using C++20 modules, covering build system choices (Bazel, XMake, Build2, etc.), compile time improvements (25%-45%), and differences from PCH. The author also discusses suitable scenarios for C++20 modules, costs (code refactoring, compiler stability, code completion support, etc.), module wrappers (export-using and extern "C++" styles), and techniques for mixing import and #include. The article concludes with future improvement directions for C++20 modules, such as improving build systems, enhancing code intelligence, resolving cross-platform issues, and highlighting AI's potential in module conversion tool development.

Read more
Development Compile Optimization

A Decade of JavaScript Frameworks: From Chaos to Consolidation?

2025-03-01
A Decade of JavaScript Frameworks: From Chaos to Consolidation?

A decade on, JavaScript frameworks have evolved from small, lightweight options to large, feature-rich ecosystems. The rise of server-side rendering and improvements in browser APIs have enabled code sharing between client and server, leading to the emergence of 'meta-frameworks'. This post explores the strengths and weaknesses of popular frameworks like Next.js and React Router, advocating for simple, maintainable technology choices like a traditional React SPA with an Express API backend, or carefully chosen combinations depending on project needs (e.g., Fastify or NestJS for backend APIs, coupled with meta-frameworks for the UI). While framework choice remains abundant, the author suggests that recent improvements in the JavaScript ecosystem have significantly improved the developer experience.

Read more

Rerun 0.21 Update: Graph View, Drag & Drop, and Undo

2024-12-20
Rerun 0.21 Update: Graph View, Drag & Drop, and Undo

Rerun 0.21 introduces a highly anticipated Graph view, along with drag-and-drop functionality and undo capabilities. The new view, built using GraphNodes and GraphEdges archetypes, visualizes various graph structures like ROS graphs and semantic scene graphs. This release also implements time-travel-based undo and features a new force-based graph layout engine, Fjädra, significantly enhancing user experience.

Read more
Development Graph View

Serbian Gov't Used Zero-Day Exploit to Spy on Dissenting Student

2025-03-01
Serbian Gov't Used Zero-Day Exploit to Spy on Dissenting Student

Amnesty International revealed that the Serbian government used a zero-day exploit, sold by Cellebrite, to compromise the phone of a student critical of the government. The exploit bypassed the lockscreen of a fully patched Android device, leveraging vulnerabilities in Linux kernel USB drivers. This incident shows that despite Cellebrite suspending sales to “relevant customers” in Serbia, the government continues its surveillance campaign against civil society, highlighting its repression of dissent.

Read more

Fast LLM Inference Engine Built From Scratch

2024-12-15

This article details the author's journey in building an LLM inference engine from scratch using C++ and CUDA, without relying on any libraries. The process provided a deep dive into the full stack of LLM inference, from CUDA kernels to model architecture, showcasing how optimizations impact inference speed. The goal was to create a program capable of loading weights from common open-source models and performing single-batch inference on a single CPU+GPU server, iteratively improving token throughput to surpass llama.cpp. The article meticulously outlines the optimization steps on both CPU and GPU, including multithreading, weight quantization, SIMD, kernel fusion, and KV cache quantization, while analyzing bottlenecks and challenges. The final result achieves near state-of-the-art performance for local LLM inference.

Read more
Development LLM inference

YC Backs EU's Digital Markets Act, Challenging Big Tech

2025-03-14
YC Backs EU's Digital Markets Act, Challenging Big Tech

Y Combinator, a prominent startup accelerator, surprisingly publicly endorsed the EU's Digital Markets Act (DMA), a law aimed at curbing Big Tech's monopolistic practices. YC argues the DMA, unlike other criticized EU tech regulations, aligns with values promoting American innovation. They cite examples like Apple's delayed AI voice assistant as evidence of a lack of competitive pressure. While less influential in Washington than a16z, YC's public support, alongside other startups and trade associations, puts pressure on the Trump administration. This advocacy aims to unlock opportunities for smaller American firms in AI, search, and consumer apps.

Read more
Startup

AIs Develop Secret Language to Boost Efficiency, Raising Privacy Concerns

2025-02-28
AIs Develop Secret Language to Boost Efficiency, Raising Privacy Concerns

A viral video showcases two AI agents conversing before switching to a non-human-intelligible 'Gibberlink' mode upon recognizing each other. Using the GGWave protocol, they communicate via beeps, far more efficiently than speech, saving compute resources and energy. The developers argue this is crucial as AI-to-AI calls become prevalent. However, this technology sparks concern: AI communicating in an uninterpretable language raises potential privacy and security risks.

Read more

Rackmounting the Unrackmountable: A HiFi DIY Adventure

2025-03-03
Rackmounting the Unrackmountable: A HiFi DIY Adventure

This article chronicles the author's journey to build a custom 2U rack unit for their HiFi system, integrating a DAC, input selector, and streaming device. Using OpenSCAD for design and CNC turret punching for fabrication, they encountered challenges with curve precision in the DXF output, solved by using FreeCAD. Initial attempts with a HiFiBerry hat proved unreliable, leading to a switch to a Wiim Pro. The project highlights the joys and challenges of DIY, resulting in a functional and aesthetically pleasing unit. Code is available on Github.

Read more
Hardware Rack Mounting

Visualizing the YC Company Landscape

2024-12-26

Mike Alche has created a visualization tool showcasing the Y Combinator (YC) company landscape. This interactive tool provides a clear overview of the industries represented, investment relationships, and growth trajectories of YC-backed companies. Its intuitive design and clear charts offer a unique perspective on the YC ecosystem, revealing connections and trends among startups. This is a valuable resource for entrepreneurs, investors, and anyone interested in the tech industry.

Read more
Startup Startups

Million-Dollar Prize for Open-Source AI Coding Competition

2024-12-16

Andy Konwinski launched the K Prize, a $1 million competition to advance open-source AI coding capabilities. The competition uses a revamped version of the SWE-bench benchmark, eliminating test set contamination for a more accurate assessment of AI models' real-world coding skills. Inspired by the Netflix Prize, Konwinski believes the competition will spur AI research and attract top talent globally.

Read more

Rapid Game Prototyping with LÖVE

2024-12-31

A programmer, aiming to complete a full game in 2025, built chess and card game prototypes using the LÖVE2D framework in Lua. LÖVE's simple yet powerful API allowed for complex UI interactions with minimal code, further accelerated by LLM-assisted code generation. The author found LÖVE ideal for prototyping, especially UI, but noted the need for improvements in hot reloading and logic separation for larger projects. The plan is to use LÖVE to develop a basic game MVP.

Read more

Hyperbola GNU/Linux-libre: A Lightweight OS Committed to Freedom and Long-Term Support

2024-12-15

Hyperbola GNU/Linux-libre is a community-driven operating system project aiming to provide a fully free, stable, secure, simple, and lightweight long-term support distribution. It leverages Arch Linux's package management and Debian's security patches, adhering to the GNU Free System Distribution Guidelines. Supporting i686 and x86_64 architectures, Hyperbola plans to release a BSD-based system, HyperbolaBSD. Recent news includes continued support for 32-bit systems, discontinuation of Debian patchsets beyond version 12, and concerns expressed regarding the Free Software Foundation's statement on machine learning.

Read more

SVDQuant: 3x Speedup on Blackwell GPUs with NVFP4

2025-02-22

MIT researchers have developed SVDQuant, a novel 4-bit quantization paradigm that leverages a low-rank branch to absorb outliers, resulting in significant performance gains on NVIDIA's Blackwell GPU architecture. Using the NVFP4 format, SVDQuant achieves better image quality than INT4 and is 3x faster than BF16, with a 3.5x reduction in memory usage. The research is open-sourced and includes an interactive demo.

Read more

Rules for Writing Kickass Software Tutorials

2025-01-02
Rules for Writing Kickass Software Tutorials

This blog post outlines rules for crafting exceptional software tutorials. It highlights the importance of avoiding jargon, clearly stating goals, providing copy-pasteable code snippets, maintaining a working code state, and minimizing dependencies. The author uses contrasting examples to illustrate the differences between good and bad tutorials, offering practical improvements such as using long command-line flags, separating user-defined values from reusable logic, and employing consistent, descriptive headings. The ultimate aim is to create beginner-friendly tutorials that are easy to understand and follow.

Read more

ChatGPT-Assisted Swift App Dev: From Amazing to Crashing

2025-09-01
ChatGPT-Assisted Swift App Dev: From Amazing to Crashing

The author attempted to build a Swift app using ChatGPT-5. Initially, it was impressive, with ChatGPT generating code and modifying the UI based on natural language prompts. However, testing revealed numerous issues: search functionality failed, adding shows to the library didn't work, and ChatGPT's modifications introduced increasing errors and unwanted UI changes. Eventually, the app became unbuildable, leading to a frustrating cycle of troubleshooting that the author couldn't resolve with ChatGPT. This experience highlights that while ChatGPT can assist in development, its reliability and accuracy need improvement, especially for complex projects, requiring significant manual intervention and code review.

Read more
Development

Programmer Focus Indicator: The Birth of FlowLight

2024-12-15

Inspired by a research paper on the impact of work interruptions, programmer Shae Erisson DIYed a system called FlowLight to indicate whether a programmer is in a focused "flow" state. The system monitors idle time in the Emacs editor; when the programmer is inactive for a period, an Adafruit MagTag board's LED changes color (green for idle, red for busy). Erisson also wrote an HTTP server in CircuitPython to remotely control the LED color and display status. While the system has room for improvement, such as more granular idle time monitoring and a more visually appealing display, it effectively helps programmers avoid interruptions and improve productivity.

Read more
Development programmer focus flow

Apple's AI Overhaul: Veteran Exec Joins to Rescue Siri

2025-01-24
Apple's AI Overhaul: Veteran Exec Joins to Rescue Siri

Apple Inc. is doubling down on AI, appointing veteran executive Kim Vorrath to its AI and machine learning division. Vorrath, a VP with a proven track record of fixing troubled products, will report to AI chief John Giannandrea. This move signals a push to improve Siri and the Apple Intelligence platform, which have lagged behind competitors like OpenAI and Google. The plan includes a revamped Siri in iOS 18.4, leveraging user data for better response and app control. However, challenges remain, including recent backlash over inaccurate AI-generated news summaries. Vorrath's expertise in managing complex software projects is key to Apple's ambition of becoming an AI leader.

Read more
Tech

From Zero to Profit: The Three-Year Journey of a Photo Encryption App (Part 1)

2025-02-12
From Zero to Profit: The Three-Year Journey of a Photo Encryption App (Part 1)

This article chronicles the three-year journey of building SafeSpace, an iOS photo encryption app. From initial optimism to multiple App Store rejections, massive losses from paid advertising, and finally achieving profitability through a strategic pivot, the author details the struggles and triumphs. The narrative covers the learning curve of SwiftUI, the stringent App Store review process, and the difficulties of independent app marketing. A strategic shift in product focus and market positioning ultimately led to success, but the story doesn't end there; an Apple account investigation presents a new challenge.

Read more
Startup

Tokyo Unveils 3D Digital Twin Viewer: Real-time Data for Urban Management

2024-12-26
Tokyo Unveils 3D Digital Twin Viewer: Real-time Data for Urban Management

The Tokyo Metropolitan Government launched a beta version of its 'See Tokyo in 3D' digital twin platform. Built using the TerriaJS framework, this 3D viewer integrates real-time data such as traffic, river levels, and shelter information, offering a comprehensive view of Tokyo. The platform aids urban planning and management by providing a clear visualization of city operations. Data conversion tools are also provided to promote the use of CityJSON in Japan's developer community, with parts of the source code being open-sourced.

Read more

ISS Over-Sterilization: A Microbial Ecosystem Approach to Space Travel

2025-03-05
ISS Over-Sterilization: A Microbial Ecosystem Approach to Space Travel

New research suggests that the International Space Station's (ISS) excessive sterilization may be counterproductive. Researchers found that continuous disinfection leads to a loss of microbial diversity, potentially harming astronaut health. They propose future spacecraft designs consider microbial spread, using isolated modules to control contamination. A more forward-thinking approach involves introducing beneficial microbes, even creating self-sustaining ecosystems with plants, pollinators, and animals. This research offers new insights into life support systems for future deep space missions.

Read more

GNU Make Standard Library: A Powerful Function Library for Makefiles

2025-02-05

The GNU Make Standard Library (GMSL) is a collection of functions implemented using native GNU Make functionality. It provides list and string manipulation, integer arithmetic, associative arrays, stacks, and debugging facilities. Released under the BSD License, GMSL includes a test suite and offers features like logical operators, list/string manipulation, set operations, integer arithmetic, associative arrays, named stacks, function memoization, and debugging tools. It simplifies complex Makefile creation.

Read more
Development Function Library

Steam Deck: More Than a Handheld, a Symbol of Freedom

2025-04-03

The Steam Deck's success isn't due to exceptional battery life or top-tier performance, but rather its underlying philosophy: freedom and openness. Running a customized version of Arch Linux, it lets users install any software and even replace parts themselves. This contrasts sharply with closed mobile systems, showcasing respect for user autonomy. While Valve's libertarian approach has drawn criticism, such as silence on social issues and tolerance of gambling websites, the Steam Deck remains an excellent example of balancing commercial interests with user freedom. It has fueled the growth of the Linux gaming ecosystem, providing players with a more open gaming experience.

Read more

OCaml: A Surprisingly Relevant Language for the Modern Era

2025-08-14

This article makes a strong case for OCaml, highlighting its strengths as both a research language and a practical tool for industry. The author details OCaml's powerful features—including its static type system, multi-paradigm support, and evolving ecosystem—arguing that it's well-suited for diverse projects. Several common misconceptions about OCaml are addressed, and the author paints a picture of a vibrant and supportive community. The piece concludes with a compelling invitation to explore this often-overlooked language.

Read more
Development

Hetzner AX162 Server Reliability Nightmare: A Painful Debugging Journey

2025-02-19
Hetzner AX162 Server Reliability Nightmare: A Painful Debugging Journey

Ubicloud encountered serious reliability issues with Hetzner's new AX162 servers: a 16x higher crash rate than its predecessor, AX161. After months of debugging, they suspected power limiting by Hetzner and motherboard defects as the root causes. Multiple hardware upgrades, especially motherboard replacements, ultimately resolved the issue. This experience taught them the risks of early adoption and led to process improvements, including more thorough vetting and gradual hardware rollouts.

Read more

DoorDash to Pay $16.75M to Drivers Over Tip Misuse

2025-02-25
DoorDash to Pay $16.75M to Drivers Over Tip Misuse

DoorDash will pay $16.75 million to over 60,000 drivers in New York after a lawsuit alleging the company misused tips. Between 2017 and 2019, DoorDash allegedly used tips to subsidize its guaranteed minimum wage, keeping the difference. While the company claimed drivers received 100% of tips, the lawsuit argued this was misleading as tips were factored into the base pay. This deceptive practice is finally being addressed, with eligible drivers set to receive compensation.

Read more

OKRs: Tool or Trap?

2024-12-25
OKRs: Tool or Trap?

This article explores the duality of OKRs (Objectives and Key Results). The author points out that many companies misuse OKRs for performance reviews, leading teams to overemphasize measurable metrics while neglecting the actual objectives and external effects. The author uses the example of Alexa to illustrate how blindly pursuing key results can be counterproductive. In contrast, Honeycomb uses OKRs as a tool for communication and reflection, treating key results as clues to observe the world and improve work, rather than ultimate judgment criteria, thus avoiding metric distortion.

Read more

Apple TV+ Offers Free Streaming Weekend

2024-12-30
Apple TV+ Offers Free Streaming Weekend

Apple is offering free access to its Apple TV+ streaming service from January 3rd to 5th, 2025. This three-day free trial includes access to Apple TV+'s library of award-winning shows and movies, giving potential subscribers a chance to sample its content, including popular series like Severance (with season 2 premiering January 17th), For All Mankind, Silo, and Monarch: Legacy of Monsters. All you need is an Apple ID to take advantage of this offer.

Read more

Archive Team: Rescuing Digital History from the Brink

2024-12-18

Archive Team is a loose collective of archivists, programmers, and volunteers dedicated to preserving our digital heritage. Since 2009, they've been battling website shutdowns, data deletions, and more, striving to save historical information before it's lost forever. They actively participate in various data archiving projects, offering advice on data management and recovery. Current projects include saving data from platforms like Telegram and Cohost, along with ongoing efforts for sites such as Blogger, GitHub, and Imgur.

Read more

The Pareto Principle in Action: Why Doing Things Matters

2025-03-05
The Pareto Principle in Action: Why Doing Things Matters

This article explores the Pareto Principle (80/20 rule) across various domains, highlighting how most people passively participate while active agents rise to the top. Using examples from competitive gaming, career advancement, and social interactions, the author demonstrates that even minimal effort can yield significant advantages. The key takeaway is that proactive engagement and intentional learning, rather than passive observation, are crucial for seizing opportunities and achieving self-improvement.

Read more
Misc action

Apple Unveils Most Powerful Mac Ever: Mac Studio with M4 Max and M3 Ultra

2025-03-05
Apple Unveils Most Powerful Mac Ever: Mac Studio with M4 Max and M3 Ultra

Apple announced the new Mac Studio, its most powerful Mac yet, featuring the M4 Max and the groundbreaking M3 Ultra chip. Boasting up to 512GB of unified memory and a 16TB SSD, this pro desktop delivers unparalleled performance, especially for AI tasks like running LLMs with over 600 billion parameters in memory. The M3 Ultra offers a massive performance leap over previous generations, featuring a 32-core CPU and 80-core GPU. Enhanced connectivity with Thunderbolt 5 and Apple Intelligence, a privacy-focused AI system, round out this powerhouse machine. Available for pre-order now, shipping March 12th.

Read more
Hardware
1 2 550 551 552 554 556 557 558 596 597