Search the archive

implementation and the results. Read the article

HOW DOES SESSION REPLAY WORK?

Session replay provides qualitative views of user interactions that have already happened. It works by reconstructing the changes made to the DOM rather than literally recording the screen. This involves re-applying the CSS to the HTML and using open-source software like rrweb to record and replay these DOM changes.

A MILESTONE FOR TYPESCRIPT PERFORMANCE IN TANSTACK ROUTER

TanStack Router, a type-safe routing library for TypeScript, made performance improvements to address a bottleneck in TypeScript's language service when handling large route trees. The issue stemmed from the language service having to infer types for the entire route tree, even when only a specific route was referenced. To solve this, TanStack Router introduced a new approach for file-based routing where the route tree is explicitly declared instead of inferred, allowing the language service to infer types only for the relevant route.

INTERNING IN GO

Go's new `unique` package can improve memory usage by interning objects, specifically strings, to avoid storing duplicate copies. Interning, by reusing identical objects, reduces memory usage compared to other approaches. This post illustrates this through a program that processes a large text file, identifying all words starting with "B" and comparing the memory usage of different methods: storing references, cloning strings, and using the `unique` package.

REMIX'S CONCURRENT SUBMISSIONS ARE FUNDAMENTALLY FLAWED (WITHOUT CAUSAL ORDERING)

Remix's approach to concurrent submissions, using a "submission and revalidation" pattern, is fundamentally flawed. It leads to unnecessary latency and introduces race conditions where the UI can display outdated information, even after users have submitted changes. There are two solutions: Causal ordering, which makes sure requests are processed in the order they depend on each other, and Persistence, where a connection is maintained between the client and server to guarantee ordered processing.

THE HIDDEN COSTS OF OVER-COLLABORATION

Over-reliance on collaboration has hidden costs. Excessive collaboration can reduce overall output due to time spent in meetings and can create expensive interactions between teams that are not always necessary. It can also hide deeper-rooted problems, like inefficient workflows and unclear team boundaries.

STOP LEARNING TO GIVE FEEDBACK. LEARN TO RECEIVE IT

Receiving feedback well means approaching feedback with curiosity and seeking clarification when possible. It's also important to avoid immediate counterarguments when receiving feedback and instead create a safe environment for the feedback giver to be honest with you.

virtually for free.

NITRO

Nitro is a framework for developing high-performance native modules in React Native. It achieves this through hybrid objects, which are native objects implemented in C++, Swift, or Kotlin that provide direct access to native functionalities. Nitro uses a lightweight layer built on JSI and bridges directly to Swift and Kotlin while still prioritizing type safety and null-safety by using a code-generator to make sure that TypeScript specifications are the single source of truth for both JavaScript and native code.

ELECTRICO

Electrico is a lightweight, cross-platform alternative to Electron that uses Rust and Javascript to emulate Electron and Node APIs within embedded WebViews. This lets users run Electron apps without bundling Node.js and Chromium.

FARMHASH

This is a Node.js implementation of FarmHash, Google's family of high performance hash functions. It offers various hash and fingerprint functions with different bit lengths and seeding options.

WHY WORDFREQ WILL NOT BE UPDATED

wordfreq provided snapshots of language usage up to 2021 - it will not be updated anymore due to several reasons. First, the rise of generative AI has contaminated online data with AI-generated text, making it difficult to accurately represent human language usage. Second, previously free sources of data, like Twitter and Reddit, have become inaccessible or expensive, making it impractical to collect new data. Finally, its developer has become disillusioned with the current state of natural language processing.

A GENTLE GUIDE TO SELF-HOSTING YOUR SOFTWARE

This is a guide for beginners on self-hosting software and using Uberspace as a platform that emphasizes the software aspect rather than complex hardware setup. The author goes through the process of self-hosting two example apps.

REACT 19 CHEAT SHEET

A downloadable cheat sheet for all of the new features in React 19.

GITHUB TRICK TO FIND THE COMMIT THAT DELETED A FILE

You can use git log -- /path/to/deleted/file to find the commit that deleted a file in your local repo, and you can do the same on a hosted repo with https://github.com/⟦user⟧/⟦project⟧/commits/⟦branch⟧/⟦path/to/deleted/file⟧.

THE EMPIRE OF C++ STRIKES BACK WITH SAFE C++ PROPOSAL

A new proposal addresses C++'s memory safety vulnerabilities by introducing features similar to Rust, allowing developers to write safer code without needing to rewrite their projects in a different language.

APPLE MOBILE PROCESSORS ARE NOW MADE IN AMERICA. BY TSMC

Apple is the first customer for TSMC's new Arizona chip fabrication facility, producing its A16 mobile processors for the iPhone on American soil.

SNAP'S NEW SPECTACLES INCH CLOSER TO COMPELLING AR

Snap's fifth-generation Spectacles have a richer, more immersive display. They feel faster and last longer on a charge. The new Spectacles will not be sold but will be instead distributed to developers making AR lenses for Snapchat. Developers will have to apply for access through Lens Studio, Snap's desktop tool for creating AR software, and pay $99 per month to rent a pair of Spectacles for at least one year. This article takes a look at what it's like to wear the new Spectacles.

SALESFORCE'S NEW AI STRATEGY ACKNOWLEDGES THAT AI WILL TAKE JOBS

Salesforce says that its AI tools can handle tasks without human supervision. The company plans to charge $2 per conversion held by its new generative AI agents. Salesforce is insulating itself from customer job cuts by pricing its new AI features based on outcomes rather than the number of employees using it. Its new AI agents will allow companies to increase their workforce capacity during busy periods without having to hire additional full-time employees or gig workers.

MUSK'S NEURALINK GETS FDA'S BREAKTHROUGH DEVICE TAG FOR 'BLINDSIGHT' IMPLANT

Neuralink's experimental 'Blindsight' device, which is designed to enable even those who have lost both eyes and their optic nerve to see, has received the US FDA's breakthrough device designation. The 'breakthrough' tag is given to certain medical devices that provide treatment for or diagnosis of life-threatening conditions. It is aimed at speeding up the development and review of devices. Neuralink is currently testing an implant designed to give paralyzed patients the ability to use digital devices by thinking - it is unknown when it expects Blindsight devices to move into human trials.

WHY UNITED CHOSE SPACEX'S STARLINK TO POWER ITS FREE WI-FI

United Airlines has signed an agreement with SpaceX to bring Starlink internet services to its entire fleet and offer free Wi-Fi to all passengers. United is currently using a mix of four different providers, all with different capabilities and limitations. Starlink is the only service that provides global coverage with enough bandwidth. Its hardware is also easy to install, with the satellite terminals being much smaller than the ones United currently uses. The first passenger flight with Starlink Wi-Fi will take off sometime in 2025.

📥 Get Surfshark for 81% off with 3 months EXTRA

OPIK

Opik is an open source end-to-end large language model development platform. It allows developers to confidently evaluate, test, and monitor LLM applications. Opik can be used for tracing, annotations, evaluations, production monitoring, and more.

VOID

Void is an open source alternative to Cursor, an AI code editor. It allows developers to write code with AI tools and access powerful AI features while retaining full control over their data. Void has many built-in and community features - anyone can build their own AI integrations. Developers can choose to host their own models or use popular hosted models.

UNFORTUNATE STARTUP PIVOTS. A SHORT LIST

Going from B2B to consumer doesn't work as the founding team's strengths usually lie in selling to enterprise customers and having a deep network in a specific vertical. Adding more social features rarely helps. Neither does chasing trendy tech like AI or web3. Other common mistakes include building platforms prematurely and going from paid to free.

WHAT ARE THE BEST OPTIONS FOR AMAZON SDES THINKING ABOUT LEAVING OVER RTO POLICY (HACKER NEWS THREAD)

This thread is for SDEs thinking of leaving Amazon over the 5-day return-to-office announcement. It discusses the best options for work-from-home, flexibility, interesting projects, and salaries. There are a lot of good options out there, but workers may have to accept lower pay. Talented Amazon engineers should basically have the 'pick of the litter' and companies will likely be fawning over them. Amazon recruiters periodically ping previous employees to see if they're interested in returning, so it's worth leaving the company on good terms.

minutes)

23ANDME INDEPENDENT DIRECTORS QUIT BOARD OVER UNSATISFACTORY BUYOUT PLAN FROM CEO

23andMe went public in 2021 - its CEO has been trying to make the company private since April, but their proposal was deemed insufficient and not in the best interest of shareholders.

MUSK SAYS SPACEX WILL SUE FAA FOR "REGULATORY OVERREACH"

SpaceX is facing upwards of $630,000 in fines for allegedly violating launch requirements during two missions in 2023.

SNAPCHAT IS GETTING ITS BIGGEST REDESIGN IN YEARS

Snapchat is simplifying its interface from five tabs to three tabs to bring Stories closer to conversations, simplify content discovery, and allow quick access to the camera.

OPENTELEMETRY TRACING IN 200 LINES OF CODE

Tracing libraries are a lot simpler than you'd expect - it's just fancy logging combined with context propagation.

WHY TO NOT WRITE A BOOK

A book commits you to a single task for years, cutting you off from other opportunities, and there is a serious risk that it may not work out.

OPEN SOURCE NEEDS TO BE FINANCIALLY SYMBIOTIC

This post looks at how Eleventy evolved from a side project to an independently full-time community-funded open source project.

🌉 Register to attend @ The San Francisco Marriott Marquis

MISTRAL FREE API AND PRICE UPDATE

Mistral has released a free API tier, dramatically reduced its costs, improved the performance of its small model, and put its vision model in Le Chat.

CHALLENGERS ARE COMING FOR NVIDIA'S CROWN

Nvidia's dominance in AI chips has propelled it to immense market value, largely thanks to its GPU capabilities and CUDA software ecosystem. However, competitors like AMD, Intel, Cerebras, and SambaNova are developing innovative solutions to challenge Nvidia's supremacy in AI hardware. While Nvidia's lead remains secure for now, the landscape is dynamic, with multiple players striving to carve out their own niches in the AI market.

JINA EMBEDDINGS V3 (HUGGING FACE HUB)

The Jina series of embeddings are a high quality and powerful suite of models that can be used for embedding and retrieval. Its development team has released the next version of their model with improved performance and training.

TRUSTWORTHINESS OF RAG SYSTEMS

This work introduces a framework to evaluate the trustworthiness of Retrieval-Augmented Generation (RAG) systems across six key areas: factuality, robustness, fairness, transparency, accountability, and privacy.

ENHANCING RECOMMENDER SYSTEMS WITH BEEFORMER

The new beeFormer framework improves sentence Transformers by incorporating interaction data, making them more effective for recommender systems.

AI COMIC UNDERSTANDING

The last frontier of Visual Language Models is the ability to understand and reason about comics. This project is a survey and a call for research.

WORD LLAMA

A lightweight toolkit for fuzzy deduplication, reranking, and other NLP-based tasks. Optimized to run on the CPU.

SYLLABLE SEGMENTATION IN SPEECH MODELS

This project enhances speech representation learning by separating syllabic structures from speaker information in self-supervised models. By fine-tuning the HuBERT model with speaker perturbation techniques, researchers improved syllable segmentation, leading to better syllabic unit organization.

DATA PIPELINES ARE THE NEW AI SECRET SAUCE

With models being somewhat commoditized, much of the advantage in AI comes from the data. It also, by extension, comes from the pipeline that ingests and creates the data. This post discusses the challenges and opportunities associated with data pipelines in the modern age.