As a worsening air leak persists in the ISS’s Zvezda service module, NASA has implemented precautionary emergency measures. Five crew members are currently sheltering inside the SpaceX Crew Dragon, wearing their spacesuits and prepared for an immediate, rapid departure. Meanwhile, the two Russian cosmonauts remain in the Russian segment to continue repair efforts, with their designated Soyuz spacecraft serving as their emergency exit. While the crew is split between the two vessels as a safety precaution, NASA emphasizes that this is not an official evacuation order. If a full departure becomes necessary, the crew will return to Earth according to their original flight assignments: the four Crew-12 astronauts will return via Dragon to splash down off the U.S. coast, while the remaining three crew members will utilize the Soyuz spacecraft to land in Kazakhstan. NASA and Roscosmos are continuing to monitor the situation and work toward a permanent repair.

This tool evaluates your ad-blocker’s effectiveness by testing three areas: **cosmetic filtering** (removing empty ad placeholders), **script blocking**, and **host blocking** (preventing connections to trackers and ads). When you run the test, the tool monitors which connections fail or which elements are removed. You then receive a color-coded score based on your performance. **Important Notes:** * **Compatibility:** If the test doesn't accurately reflect your setup, it doesn't necessarily mean your protection is failing; some ad-blockers, VPNs, or DNS configurations are simply incompatible with this specific testing method. * **Scoring:** A green score indicates satisfactory protection. You do not need a perfect 100% to be secure. * **Optimization:** For those seeking higher scores, the author suggests using specific host lists (like the "Turtlecute" list) or combining browser-based blockers like uBlock Origin with privacy-focused browsers or DNS solutions like NextDNS. Ultimately, this project is designed as a user-friendly, quick way to verify your ad-blocking setup while acknowledging that multiple tools and configurations are valid for maintaining online privacy.

Bundler 4.0.13 introduces "cooldown," a security feature designed to mitigate supply-chain attacks by preventing the immediate installation of newly released gems. Because many attacks rely on the brief window between a malicious version being published and developers unknowingly installing it, cooldown allows users to set a mandatory aging period (e.g., 7 days) before a new version becomes resolvable. The feature is opt-in and highly flexible, configurable via the `Gemfile`, global settings, environment variables, or command-line flags. It leverages the `created_at` timestamp now provided by the Rubygems.org v2 index. When a cooldown is active, `bundle install` will bypass versions that haven't aged sufficiently, while `bundle outdated` provides visibility into held-back updates. Crucially, cooldown includes an "escape hatch": users can set `--cooldown 0` to immediately install critical security patches or emergency fixes. This feature complements existing security measures like 2FA and trusted publishing, adding a layer of protection that ensures code has been scrutinized before it is pulled into your project. Existing `Gemfile.lock` files remain undisturbed, ensuring that this new policy does not unexpectedly disrupt your current environment.

Russian users attempting to access blocked websites have recently encountered error messages in Chinese: "This URL has been blacklisted." Investigations reveal that these messages originate from TSPS (Technical Means of Countering Threats) hardware, the "black boxes" installed by Roskomnadzor to enforce internet censorship. The technical analysis indicates that these systems use Deep Packet Inspection (DPI) technology supplied by China Unicom, a state-owned telecom giant and key developer of China’s "Great Firewall." This equipment is legally mandated to provide the Chinese Ministry of State Security with full access to data and system code, raising major concerns regarding espionage and the security of Russia's critical infrastructure. The irony is that the "sovereign Runet" project, intended to ensure independence from foreign technology, has effectively outsourced Russian internet surveillance to Chinese infrastructure. Due to this reliance on sophisticated foreign DPI systems, standard VPN protocols are easily detected and blocked. To counter this, advanced stealth protocols like VLESS Reality are necessary to mask traffic and bypass these high-level filtering measures, protecting users from both domestic censorship and potential foreign digital surveillance.

Programmers often complain that colleagues are more willing to write documentation for AI tools like Claude than for their human coworkers. However, the author suggests a shift in perspective: treat the documentation generated by AI as a valuable project asset. The author now instructs Claude to maintain “handoff” documents throughout a project’s lifecycle. At the end of each project, they ask Claude to generate a structured, high-level summary of the problem solved and the changes implemented. After reviewing and editing these summaries—ensuring they meet the same standard as human-written documentation—the author commits them to the repository. This process offers a significant efficiency boost: a task that would take an hour manually takes only seconds for the AI to draft and minimal time for the human to review. By archiving these AI-generated summaries, developers leave behind a useful paper trail for future team members, effectively bridging the gap between machine-assisted development and collaborative knowledge sharing. The author advises treating these AI outputs as professional work, ensuring they are verified and integrated into the project's permanent record.

mouseless Lighting-fast mouse control with the keyboard Enable JavaScript This site requires JavaScript to function properly. Please enable JavaScript in your browser settings.

This article challenges the Reaganomics-era belief that extreme wealth accumulation acts as a "rising tide" for the economy. By comparing the billionaire class of 1988 to today’s super-wealthy, the author highlights an astronomical growth in fortunes that far outpaces inflation. Using vivid illustrations—such as the vast difference in time and scale between a million and a billion—the author argues that modern wealth concentration has moved beyond "economic utility" into a realm of societal distortion. The piece suggests that once wealth reaches a certain threshold, the individual loses touch with reality, as their extreme capital renders them immune to failure, consequences, and the social feedback loops that typically foster moral development. Ultimately, the author questions whether we truly need individuals with hundred-billion-dollar net worths to drive innovation. They propose that society should establish wealth caps to prevent individuals from exerting excessive political and economic control. The author concludes that shifting away from this trajectory requires electing leaders who prioritize the common good over billionaire interests, advocating for a return to a society where accountability and shared prosperity take precedence over unchecked accumulation.

Fully remote Our global team operates remotely with modern tools, valuing outcomes over hours. Open-source We champion transparency, community and developer friendliness. Developer tool Work for your peers, on pain points you know, with a fast feedback loop. Technical challenges From DevExp to high scale to API fragmentation, we work on hard problems. Expert-led We are developer infrastructure veterans from Uber, Netlify, Algolia. Traction With fast-growing revenues & usage, it’s a perfect time to join.

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Redis 8.8 introduces significant performance gains and several powerful new features designed to improve developer efficiency and application scalability. **Performance Improvements** Redis 8.8 delivers substantial end-to-end throughput increases across core data types. Notably, **XREADGROUP** sees up to 83% improvement, **Sorted Set** operations (ZADD, ZRANGE) improve by up to 74%, and **MGET** performance increases by up to 68%. Additionally, persistence and full replication are now up to 60% faster. **Key Features** * **Array Data Structure:** A new, high-performance, index-addressable container that supports dynamic resizing, sparse storage, and built-in server-side aggregation, making it ideal for sliding windows and log management. * **Window Counter Rate Limiter:** New `INCREX` command simplifies rate limiting by providing server-side windowing, expiration control, and boundary enforcement without requiring Lua scripts. * **Streams NACKing:** The `XNACK` command allows consumers to explicitly release messages, prioritizing them for immediate re-delivery to other consumers. * **Subkey Notifications:** Hashes now support field-level notifications for events like expiration or deletion. * **JSON & Time Series:** Users can now specify floating-point precision (BF16–FP64) for JSON arrays and execute multiple aggregators in a single time series query. * **Sorted Set COUNT:** A new aggregator for union/intersection operations that scores elements based on set membership frequency.

Enable JavaScript and cookies to continue

**lowfat** is a lightweight, local-first CLI tool designed to minimize AI token costs by filtering unnecessary CLI output before it reaches your AI agent. By streamlining verbose command responses, it ensures your tokens are spent only on essential data. **Key Features:** * **Composable & Extensible:** Follows a UNIX-style pipe philosophy, allowing users to mix built-in filters with custom plugins. * **User-Centric:** Provides transparent history tracking, allowing you to monitor savings and customize filters based on your specific workflows. * **Versatile Integration:** Works seamlessly with tools like Claude Code, OpenCode, and Pi agent. It can be used via direct prefixing, shell initialization, or agent-specific configuration hooks. * **Granular Control:** Adjust compression aggressiveness (from "lite" to "ultra") and easily develop custom plugins using a dedicated DSL. * **Privacy-Focused:** Operates entirely locally with no telemetry, ensuring you maintain full ownership of your data. Installation is simple via `cargo` or Homebrew. Whether you are optimizing a CI/CD pipeline or everyday terminal tasks, `lowfat` provides the insights and control necessary to keep your AI interactions efficient and cost-effective.

In this poignant reflection, Kevin Jack McEnroe explores his evolving relationship with his mother, actress Tatum O’Neal. Following a stroke caused by a drug overdose, O’Neal is no longer the volatile, addicted figure of his childhood; instead, she has become, in some respects, the stable and attentive mother he always longed for. McEnroe reflects on the profound toll of loving an addict—the constant anxiety, the blurred boundaries of codependency, and the lifelong burden of responsibility. He candidly discusses the challenge of setting firm limits, such as refusing to buy her vapes, which helps him define his own autonomy while teaching her the healthy boundaries she lacked as a celebrity accustomed to always getting her way. Ultimately, McEnroe finds peace by accepting his mother’s complexities and his own. He no longer feels the need to shield himself from his upbringing or rescue his mother from her choices. By embracing his capacity for deep love while maintaining emotional distance, he has learned to “steady his boat,” successfully navigating the difficult balance between providing compassion and preserving his own well-being.

To achieve technological sovereignty, the European Commission has unveiled a comprehensive, multi-layered strategy designed to strengthen Europe’s digital autonomy across the entire value chain. This integrated approach builds upon existing projects—such as AI Factories—through four key initiatives: * **The Chips Act 2.0:** Focuses on bolstering the semiconductor ecosystem, supply chain resilience, and domestic demand. * **The Cloud and AI Development Act (CADA):** Aims to leverage cloud and AI technologies to transform industrial ecosystems and improve societal outcomes. * **The EU Open Source Strategy:** Designed to minimize external dependencies across the entire technology stack. * **Strategic Roadmap for Digitalisation and AI in Energy:** Targets the digital transformation of the energy sector. By interconnecting these initiatives, the Commission aims to create a mutually reinforcing framework that secures Europe’s technological future from infrastructure to software.

Establishing a secure connection... Please enable JavaScript to continue. Request ID: 4a2d83dbec1ef61e7f7ead87a2accd17

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

**databow** is a new, open-source command-line tool designed to unify database interactions. Built in Rust for speed and efficiency, it provides a single, consistent interface for querying any database that supports an ADBC (Arrow Database Connectivity) driver. By leveraging the Apache Arrow ecosystem, databow eliminates the need to juggle multiple, fragmented CLI tools (like `psql` or `snowsql`) with varying syntaxes and quirks. Whether you are working with transactional, analytical, lakehouse, or time-series databases, databow offers a standardized experience. **Key features include:** * **Unified Interface:** Connect to 30+ databases (e.g., PostgreSQL, Snowflake, DuckDB, BigQuery) using a single tool. * **Modern REPL:** Includes syntax highlighting, multiline input, and clean, aligned output. * **Flexible Usage:** Ideal for both interactive shell use and automation within scripts and pipelines. * **Efficient Export:** Supports direct exporting of query results to formats like CSV, JSON, and Arrow. * **Streamlined Profiles:** Simplify connections by saving database configurations as profiles. You can install databow via `uv tool install databow` or `cargo install databow`. For more information and to explore the growing list of supported drivers, visit the project documentation.

John Archibald Wheeler famously described gravity as a cycle: matter tells space how to curve, and space tells matter how to move. While Einstein’s general relativity explains this via the geometry of space-time, it fails at the quantum scale, such as inside black holes. To bridge this gap, physicists are using the "holographic principle," which maps 3D space-time onto a 2D surface of quantum particles. Previous research identified quantum "entanglement" as the connective tissue that gives space its structure. However, these early models resulted in an "inert" space—one that could exist but could not bend or react to matter. Recent breakthroughs by researchers like Charles Cao have identified a missing ingredient: "magic." In quantum computing, magic refers to non-Clifford gates that introduce complexity into quantum states. By incorporating these gates into quantum error-correcting codes, scientists have shown that "magic" acts as a fabric softener, allowing space to become flexible. This suggests that gravity is a direct manifestation of quantum mechanics, arising when space is encoded with enough complexity to allow matter and geometry to interact. While still in early development, this approach frames space-time as an emergent property of quantum information.

In this post, Safia Abdalla explores the enduring legacy of ANSI escape codes, a standard originating in 1979 that remains the foundation for terminal interactivity today. Originally developed to allow "dumb" CRT terminals to process instructions for text formatting, colors, and cursor movement within plain-text streams, the standard works by inserting specific escape sequences (starting with `\x1b[`) that modern emulators interpret as commands rather than visible text. While the standard began with basic 8-color support, it has evolved to accommodate modern 24-bit RGB color and complex UI elements like progress bars, spinners, and full-screen applications such as Vim. Today, developers use libraries like `chalk` or `Spectre.Console` to abstract these codes, allowing for the creation of sophisticated, interactive command-line interfaces. Ultimately, Abdalla emphasizes that understanding these 50-year-old standards provides valuable insight into how current software design choices can scale over decades, while highlighting the fun of experimenting with these "incantations" to build modern terminal experiences.

ESP32 Bit Pirate is an open-source, multi-protocol hacking tool that transforms ESP32-S3 devices into versatile diagnostic and development platforms. Inspired by the legendary Bus Pirate, this firmware supports an extensive array of digital protocols, including I2C, SPI, UART, 1-Wire, JTAG, and CAN, while adding robust radio capabilities like Wi-Fi, Bluetooth, RFID, Sub-GHz, and Infrared. Users can interact with the device through a web-based interface, serial terminal, or a standalone mode (on supported hardware like the M5 Cardputer). It features advanced scripting support via Python, allowing for automation of tasks such as EEPROM dumping, signal analysis, and protocol sniffing. The ecosystem includes dedicated hardware docks and expanders that ensure compatibility with legacy Bus Pirate accessories. Designed for educational, diagnostic, and interoperability testing, the firmware is highly extensible and user-friendly, with one-click installation via a web flasher. Comprehensive documentation and community-driven scripts are available to assist with device manipulation and data logging. *Note: The tool is for authorized use only. Users must strictly adhere to 3.3V/5V logic levels and local telecommunications regulations to avoid device damage or legal issues.*

The Ladybird project is transitioning to a closed-contribution model where only project maintainers can submit code. Public pull requests will no longer be accepted, and all currently open requests will be closed. This decision stems from the changing landscape of open-source development, specifically the rise of AI tools. Previously, code contributions served as a reliable proxy for trust and effort. However, because AI can now rapidly generate sophisticated code, these submissions no longer offer the same assurance of good faith. As Ladybird approaches its first alpha release, the team must prioritize security and architectural integrity; they contend that only those directly responsible for maintaining the browser should introduce changes that could impact user security. While Ladybird remains open source and will continue to welcome community input through bug reports, security feedback, and design discussions, it will no longer maintain a formal process for external code submissions. The maintainers emphasize that this shift is necessary to ensure they can take full responsibility for the browser as it prepares for real-world usage.

A new DNA synthesis method called "Sidewinder" promises to bridge the gap between AI-driven genomic design and physical construction. While generative AI tools like Evo 2 can now design complex genetic sequences at high speeds, laboratory assembly has remained a slow, expensive, and error-prone bottleneck. Developed by a team led by Caltech’s Kaihang Wang, Sidewinder allows scientists to assemble multiple genetic sequences simultaneously with unprecedented precision. By using unique molecular "barcodes" on DNA fragments—akin to page numbers in a manuscript—the method ensures fragments link to their intended partners, virtually eliminating assembly errors. Recent innovations, including software that automates barcode design and the use of cost-effective raw materials, enable researchers to build long, complex DNA constructs in days rather than months. This "step change" in synthetic biology accelerates the development of engineered microbes for drug, fuel, and chemical production. The researchers have co-founded a company, Genyro, to commercialize the technology, though they intend to keep the platform accessible to the academic community, potentially paving the way for the assembly of entire synthetic genomes.

This project explores interfacing a Raspberry Pi Pico (RP2350) with a Z80 CPU to monitor and interact with its address and data buses. The author utilizes a custom breakout PCB to connect the RP2350’s GPIOs to the Z80’s 16-bit address and 8-bit data buses, as well as its essential control lines (/RD, /WR, /IORQ, etc.). A key focus is understanding Z80 bus timing, specifically the clock requirements, to ensure synchronization. The author uses the RP2350’s high-speed GPIO capabilities to poll the Z80 bus, effectively emulating discrete TTL logic chips to capture data. A demonstration proves the concept by using an RC2014 computer to perform I/O writes, which the RP2350 detects to update a local LED display. While currently using simple, non-synchronized polling on one core, the author intends to advance this into a more robust interface. Future goals include implementing bidirectional I/O, memory-mapped device emulation, and utilizing the RP2350’s PIO (Programmable I/O) blocks for autonomous bus management, potentially allowing the RP2350 to control the Z80 clock for precise, instruction-aware single-stepping.

In this post, the author experiments with writing a concise article, breaking their habit of producing overly detailed, long-form content. The focus is on implementing "linear cosine palettes," a simple procedural method for generating continuous color palettes in R. Originally proposed by Inigo Quilez and adapted for R by Mike Cheng, the technique uses a cosine function—$f(t) = \mathbf{a} + \mathbf{b} \ \cos(2 \pi(\mathbf{c} t + \mathbf{d}))$—to create smooth, visually appealing color transitions based on randomly selected base vectors. The author provides a compact R function to implement this approach and demonstrates how these palettes look when rendered as strips. To test the practical application of these palettes, the author applies them to two generative art systems: a rectangle subdivision method and a Lissajous curve generator. Despite minimal optimization, the palettes yield high-quality results. The author concludes that the method is remarkably effective and—perhaps most importantly—manages to successfully complete the exercise of writing a brief, focused blog post.

**isUpMap** is a real-time status dashboard that monitors the health of over 80 popular internet services. Using a live, JavaScript-powered heatmap, the platform provides immediate visibility into whether key tools are operational, degraded, or currently down. The service covers a broad spectrum of critical infrastructure, including: * **AI:** OpenAI, Anthropic, xAI, Groq, and more. * **Developer & Cloud:** GitHub, AWS, Vercel, Cloudflare, and Docker. * **Business & Productivity:** Slack, Discord, Stripe, Notion, Figma, and Shopify. Beyond current status checks, isUpMap tracks incident history and per-service uptime statistics. Users can browse the full directory to verify the status of specific tools or enable desktop notifications to receive instant alerts when a service experiences an outage. It is an essential tool for developers and professionals who rely on constant uptime for their daily operations.

The author explores the feasibility of "local-first" AI by fine-tuning open-source LLMs (Llama 3.1 8B and Qwen 2.5 7B) to mimic the writing style of 1990s Microsoft technical documentation. Using over 37 million words from the "Bitsavers" archive, the author cleaned the data and applied QLoRA (Quantized Low-Rank Adaptation) via cloud-based GPUs to create specialized adapters. This process allowed the models to adopt a distinct, period-correct technical tone, successfully applying 90s-style documentation structures even to modern or fictitious concepts. The experiment revealed that fine-tuning is an effective, affordable way to imbue models with a specific stylistic "voice" or in-house writing standard. However, the author emphasizes that this is not a turn-key solution: it requires high-quality data curation, careful management of training parameters (like rank and epochs), and a balance between model expressiveness and factual reliability. Ultimately, the author concludes that while these fine-tuned models are impressive impersonators capable of augmenting a writer’s workflow, they lack the human judgment necessary to replace a technical writer.

An investigation by *The Intercept* has revealed that the U.S. military is using "La Tilde," a new AI-powered content mill, to disseminate propaganda across Latin America. Operated as a messaging platform for U.S. Special Operations Command South (SOCSOUTH), the site poses as a modern news brand while publishing a blend of personal finance tips and articles that aggressively promote U.S. military operations. The site relies heavily on AI-generated text and imagery, often characterized by poor quality and visual errors. It hides its government backing behind a small, easily overlooked disclosure on its "About" page, a tactic mirrored by other Pentagon-sponsored propaganda networks. Despite claiming to employ dozens of reporters, the site appears to operate without a professional masthead, instead utilizing digital marketing subcontractors to push narratives that favor U.S. geopolitical interests, such as joint military exercises and controversial regional interventions. Experts note that while the content is "lazy" and the AI execution is shoddy, such platforms allow the military to quickly shift and scale influence operations, raising significant concerns regarding transparency and the use of automated propaganda to shape foreign public opinion.

A new documentary chronicling the 40-year history of C++ has premiered on YouTube. Reflecting on the film’s release, Herb Sutter celebrates C++’s journey from its humble origins at Bell Labs to its current status as the fastest-growing major programming language, boasting a 90% increase in users over the last 3.5 years. The documentary provides a comprehensive look at the language's evolution, featuring insights from key figures including Bjarne Stroustrup, Alexander Stepanov, Brian Kernighan, and John Romero. It covers critical milestones, such as the invention of the Standard Template Library (STL), the formation of the ISO standards committee, the "C++ winter" of the early 2000s, and the resurgence of modern C++ starting with C++11. By detailing the language's impact on industries ranging from gaming and high-frequency trading to particle physics at CERN, the film serves as an essential synopsis of one of computing’s greatest success stories. Sutter recommends the documentary as a "must-watch" for anyone interested in the history and future of software development.