Search the archive

→ Get unlimited parallel test runs

→ Zero Flakes guaranteed

Learn more about their 90-day pilot

THE THREE BYTE FIX

Our World in Data, a charity org that shows charts about the world, received a complaint about slow map charts. After some profiling, a developer found that the k-means clustering algorithm the site was using was the reason. The issue was due to a 3-character typo in the JavaScript port of the algorithm. Fixing the typo resulted in a 20x speed improvement.

PRINTING MUSIC WITH CSS GRID

This article describes a new way to show music notation responsively on the web using CSS Grid. The system defines musical staves as grids, allowing notes to be positioned by pitch (on the y-axis) and time (on the x-axis). This approach is responsive, allows for easy swapping of clefs and time signatures, and makes sure elements like chords and lyrics are aligned properly.

"GOOD ENOUGH" IS GOOD ENOUGH!

Creating software that is “good enough” is much better than trying to create software that is perfect. Software can always be iterated upon after release. HTTP and HTML are examples of imperfect technologies that have staying power thanks to their continuous improvement.

THE BIGGEST PROBLEM IN TODAY'S WORK CULTURE - DISTRACTIONS

Distractions, usually from too many meetings, are the biggest problem in today's work culture. They prevent engineers from entering a “flow state” where they can do deep work and complete their tasks much faster. Managers can help by clustering meetings together and improving meeting culture. Engineers should protect their focus time and keep their phones away while in deep work.

SVELTE 5 RELEASE CANDIDATE

Svelte 5 is now in the Release Candidate stage, which means the design of the framework is mostly set. There will probably be no breaking changes, which means developers can now try it out without worries about future changes breaking their project.

EXTENSION.JS

Extension.js is an easy way to create cross-browser web extensions without any build configuration. It's plug-and-play and has built-in support for TypeScript, WebAssembly, React, and modern JavaScript.

PYINFRA

Pyinfra automates infrastructure using Python with features for ad-hoc command execution, service deployment, configuration management, and more.

AI LEADERBOARDS ARE NO LONGER USEFUL. IT'S TIME TO SWITCH TO PARETO CURVES

Current AI leaderboards aren't useful anymore because they don't account for the cost of running AI agents. Researchers should instead use Pareto curves to visualize the tradeoff between accuracy and cost. Surprisingly, simple baseline agents can be just as accurate as the complex, costly agents that top leaderboards.

HOW PERPLEXITY BUILDS PRODUCT

The AI-powered search engine Perplexity, which has tens of millions of users with less than 50 employees, uses AI internally to inform all aspects of building the company. It has small teams of 2-3 people that work in parallel to reduce any coordination costs. The company hires self-driven and very proactive engineers.

AMAZON CODEWHISPERER IS NOW CALLED Q DEVELOPER AND IS EXPANDING ITS FUNCTIONS

Amazon's AI coding assistant CodeWhisperer has been rebranded as Q Developer and now includes testing, transformation, and autonomous feature implementation.

RUN0, A SYSTEMD BASED ALTERNATIVE TO SUDO, ANNOUNCED (MASTODON THREAD)

run0 is a sudo-like command that provides a more secure and user-friendly way to temporarily get sudo privileges while handling issues like the size of the sudo binary.

7.CSS

A CSS framework for recreating the Windows 7 UI.

QUITTING ENGINEERING TO PURSUE ART FULL-TIME

Pablo Azócar made the leap from engineer to photographer to pen plotter artist.

> Soaking up the sun in Hawaii from $161 roundtrip

> Exploring Greece from $290 roundtrip

Just sign up, set your airports, and relax as deals like these

roll in:

CHATGPT'S AI ‘MEMORY' CAN REMEMBER THE PREFERENCES OF PAYING CUSTOMERS

OpenAI's Memory feature, which allows ChatGPT to store queries, prompts, and other customizations, is now available to ChatGPT Plus paying subscribers outside of Europe and Korea. Memory lets users tell ChatGPT to remember certain details and allows the chatbot to learn from conversations. Users have control over what ChatGPT retains and what is used for additional training. ChatGPT will tell users when memories are updated. The feature will eventually roll out ChatGPT Enterprise and Teams, but OpenAI has not specified when.

MUSK WINS CHINA'S BACKING FOR TESLA'S DRIVER-ASSISTANCE SERVICE

China's government has signaled its blessing for Tesla to roll out its advanced driver-assistance service in the country. Tesla will deploy its Full Self-Driving service based on mapping and navigation functions provided by Baidu. Being able to roll out FSD in China will help Tesla compete with major Chinese carmakers, some of which already provide higher-level driver-assistance features. Chinese officials have yet to respond to Elon Musk's request to transfer the data that Tesla cars collect in China to the US to further train the system.

CHINA READY TO LAUNCH LUNAR FAR SIDE SAMPLE RETURN MISSION

China's Chang'e-6 mission aims to land on the far side of the moon and collect up to 2,000 grams of lunar material to return to Earth. The mission will also include international scientific payloads from France, Sweden, Italy, and Pakistan. The launch time has yet to be disclosed, but navigation warnings indicate that the launch is expected early Friday. China plans to send a crewed mission to the Moon by 2030 and establish a permanent lunar base through the International Lunar Research Station program in the 2030s.

ATOMIC NUCLEUS EXCITED WITH LASER: A BREAKTHROUGH AFTER DECADES

The long-sought thorium transition has been found. This discovery promises revolutionary technological applications and could answer the question of whether the constants of nature are actually constant or whether they change in space and time. It makes it possible to combine classical quantum physics and nuclear physics. The technology can be used for precision measurements. It could create new types of clocks that would be significantly more accurate than the best atomic clocks available today and devices that analyze the gravitational field around Earth to provide indications of mineral resources or earthquakes.

Learn more about their 90-day pilot

GITHUB COPILOT WORKSPACE: WELCOME TO THE COPILOT-NATIVE DEVELOPER ENVIRONMENT

GitHub Copilot Workspace is a platform where any developer can go from idea to code to software in natural language. It allows developers to brainstorm, plan, build, test, and run code in natural language and tweak different Copilot-powered agents. GitHub Copilot Workspace is designed to enhance developer creativity and materially lower the barrier to entry to building software. This article provides an introduction to the platform along with an overview of how to use it.

DOKPLOY

Dokploy is a free self-hostable Platform-as-a-Service. It simplifies the deployment and management of applications and databases. Dokploy allows developers to deploy applications efficiently and securely on any VPS. It features Traefik integration, real-time monitoring, database backups, and more.

GPT2-CHATBOT

gpt2-chatbot recently appeared in the LMSYS Chatbot Arena, demonstrating greater compatibility than any previously known GPT-2 model. There is little information about the model listed on the site or elsewhere. This article explores the model and its capabilities. The model appears to be based on the GPT-4 architecture and consistently provides different outputs than models from other companies. It can be more performant than certain other models in specific domains. It is possible that this mystery model is in fact GPT-4.5.

GOOGLE LAYS OFF STAFF FROM FLUTTER, DART, AND PYTHON TEAMS WEEKS BEFORE ITS DEVELOPER CONFERENCE

Google has laid off staff across key teams like Flutter, Dart, Python, and others ahead of its annual I/O developer conference in May. The company says that the layoffs aim to help it simplify its structures to give employees more opportunities to work on innovative and important advances while reducing bureaucracy and layers. Some employees were asked to onboard their off-shore replacements. Affected employees will be able to apply for other open roles at Google.

THE INTERNET ARCHIVE'S LAST-DITCH EFFORT TO SAVE ITSELF

If a judge chooses to deny the Internet Archive's appeal, which is likely, the site may be faced with more lawsuits and forced to remove content of questionable copyright.

INSIDE THE ‘COLD WAR' AT TECHSTARS AS CEO MÄELLE GAVET HIRES, FIRES, FIGHTS TO FORCE CHANGE

Techstars experienced two weeks of chaos after Silicon Valley Bank's collapse - SVB's parent company was a major investor in Techstars and Techstars had a sizable deposit in SVB.

KAYTU

Kaytu reduces cloud costs for engineering, DevOps, and SRE teams by recommending optimal workload configurations based on actual usage.

A BRIEF HISTORY OF WEB DEVELOPMENT AND WHY YOUR FRAMEWORK DOESN'T MATTER

As long as you're providing value to people and/or having fun doing it, it doesn't matter what technical choices you make - things will change, but programmers will still be around writing useful programs.

META AND REASONABLE DOUBT

Investors' reaction to Meta's recent earning call shows skepticism about the company's short to medium-term future - this article looks at the company's long-term prognosis.

RECONCILING EFFECTIVE ALTRUISM AND E/ACC

Effective Altruism and Effective Accelerationism mostly just disagree on the topic of AI safety - there is agreement on nearly everything else.

COPILOT WORKSPACE

GitHub has announced a new project to allow language models to somewhat autonomously finish features and fix bugs.

OPENAI INTRODUCES “MEMORY” FEATURE FOR CHATGPT PLUS USERS

OpenAI has launched a new "Memory" feature for ChatGPT Plus users, allowing the AI to retain information across conversations, which can be managed via settings.

EFFORT ENGINE

Effort enables real-time adjustment of computations during LLM model inference on Apple Silicon chips, striking a balance between speed and quality. The method, which requires no retraining but requires conversion and precomputation, allows models to operate faster by loading fewer weights. The developers are seeking assistance from Swift/Metal engineers to optimize the implementation, which is available for download on GitHub.

OPEN GATO REPLICATION: JAT

GATO by DeepMind was touted to be a generalist agent. A team associated with Hugging Face has trained and evaluated JAT, which is a Jack-of-All-Trades model that performs reasonably on a wide variety of tasks.

FP6-LLM

Reducing floating point precision speeds up training but can be unstable. This paper shows that you can get full tensor core utilization in a new packing format without common instabilities or slowdowns from naive attempts.

STAR-CODER 2 15B INSTRUCT

This model is trained on synthetic and human data. Permissively licensed, it achieves 72.6 on humaneval. Its developers give great insights into how to replicate their data pipeline and apply principles to other problems where synthetic data might be useful.

path. Sign up free and start learning

WHISPER CPP CLI

A fully self-contained speech-to-text system built on top of Whisper by the folks who built Ruff.

MEMARY

Memary allows your agents to store a large corpus of information in knowledge graphs, infer user knowledge through our memory modules, and only retrieve relevant information for meaningful responses.

MISTRAL.RS

A fast LLM inference platform that supports inference on a variety of devices, quantization, and easy-to-use applications with an OpenAI API-compatible HTTP server and Python bindings.

JAX INTERNALS

Autodidax tutorial that shows the inner ideas in the growing Jax framework. It also introduces helpful jargon often used in the community.

AI ISN'T USELESS. BUT IS IT WORTH IT?

This article provides a critical examination of AI and machine learning, arguing that while AI tools can be useful for certain tasks, they often fail to live up to the grand promises made by AI companies. It highlights the significant gap between AI's practical use cases, like coding assistance and language generation, and the inflated expectations set by industry narratives. The article raises concerns about the ethical and environmental costs of AI development, along with the potential for misuse and overreliance on the technology.