HackerNews Digest

The paper “Longest Straight Line Paths on Water or Land on the Earth” (R. Chabukswar & K. Mukherjee, submitted 9 Apr 2018, revised 2 Jul 2018) addresses two complementary geospatial optimization problems: (1) determining the maximal distance a vessel can travel along a straight line on Earth without intersecting any land, and (2) finding the longest straight‑line route a vehicle can drive without crossing a significant water body. The authors model these problems as global optimization tasks complicated by the fractal nature of coastlines, islands, and inland lakes. They apply a branch‑and‑bound algorithm to efficiently search the solution space and compute the optimal paths. Submission history shows four version updates, with file sizes around 5.5 MB. The work contributes a systematic computational approach to previously heuristic analyses of uninterrupted straight‑line travel on the planet.

RISC‑V has been added to CPython as an official tier‑3 platform. RISC‑V is an open instruction set architecture, distinct from proprietary ISAs such as x86 and ARM, and its ecosystem is projected to expand significantly. Support was achieved through community contributions: testing on physical RISC‑V hardware, fixing architecture‑specific bugs, improving build processes, and reviewing patches, with notable assistance from the RISE Project, which supplied machines and buildbots, and from the Sovereign Tech Agency fellowship. Future work includes integrating RISC‑V testing directly into CPython’s continuous‑integration pipeline via RISE’s RISC‑V Runners, aiming for faster feedback and earlier detection of platform‑specific issues. Long‑term goals are to promote RISC‑V to tier‑2 status, develop architecture‑specific optimizations, and broaden support across the Python ecosystem by encouraging users to test CPython on diverse RISC‑V hardware and report findings.

Tencent released the open‑source Hy4 preview, a next‑generation large language model with 770 billion total parameters, 49 billion active parameters, and a context window exceeding 1 million tokens. The model is accessible via WorkBuddy, CodeBuddy, Yuanbao, ima, Tencent Cloud TokenHub, and OpenRouter; it is free on WorkBuddy and CodeBuddy for two weeks, with Hy3 access extended to 30 Sept. Internal blind testing by 163 experts on 203 engineering tasks gave Hy4 an average score of 2.99/4.00, marginally ahead of GLM‑5.3 (2.92) and Kimi K3 (2.94). Hy4 shows gains in software engineering (long‑context planning, debugging, front‑end visual quality), office productivity (complex workflow, financial analysis, cross‑document collaboration), game development (prototype generation from natural‑language prompts), and scientific research (AI, molecular dynamics, condensed‑matter physics, mathematics). The model participated in its own training‑method optimization and performed autonomous inference‑system improvements, raising throughput by 31.8 %. API pricing is $0.834 per M input tokens, $2.501 per M output tokens, and $0.042 per M cache‑hit tokens. A subsequent Hy4 series rollout is planned.

The author observes a heightened sensitivity to software bugs—seeing hundreds weekly—while most users miss them, a phenomenon termed “bug blindness.” This awareness often leads to identifying issues ranging from moderate (spam‑filled search results) to severe (products that only work via non‑intuitive work‑arounds). Examples include poor search engine outputs, widely disliked platforms like Blackboard, and performance‑cheating in Discourse. The author notes that fans and developers frequently exhibit “quality blindness,” dismissing flaws because they align with internal optimism or personal habits.

Key points: - Repeated exposure cultivates habitual work‑arounds that mask underlying defects. - Dog‑fooding helps but can be undermined by ingrained shortcuts. - Overcoming quality blindness requires deliberate feedback and willingness to notice flaws. - The author has successfully taught peers to notice bugs, leading to lasting improvements in their perception of software quality. - Similar blind spots appear in other domains, such as advertising effectiveness and user perception of ads.

XCENA and Samsung’s MX1 is a CXL.mem device that combines up to 2 TB DDR5, a PCIe 6/CXL 3.2 ×8 host interface (128 GB/s bidirectional) and eight downstream PCIe 6 lanes for SSDs presented as memory. The board (≈90 W total) hosts 3072 in‑order RISC‑V cores (1.1 GHz) organized into 24 subsystems (128 cores each). Each cluster of 32 cores shares a 128 KB L2 data cache (VIPT) and a 128 KB L2 instruction cache; four cores share an 8 KB instruction cache. Two ARM Cortex‑A53 cores manage control. A custom vector processing engine per subsystem provides FP16/FP32 dot‑product throughput of ~3 TFLOPS (≈128 FLOP per cycle). The ASIC is fabricated on Samsung’s 4 nm process, consuming ~40 W plus DRAM power. Memory hierarchy includes 4 KB L1 data cache, a 1024‑entry TLB (64 KB pages) and a map cache for SSD‑backed pages. The programming model resembles OpenCL/CUDA with a shared virtual address space and host‑device pointer compatibility. SSDs can be cached in DDR5, pinned, or prefetched to mitigate latency; RAID can increase bandwidth. Near‑memory compute exploits internal bandwidth and reduces CXL traffic, enabling >200 GFLOPS effective performance despite a 3 TFLOPS peak.

FreeCORE is an independently maintained operating system built on FreeBSD that continues the TrueNAS CORE 13.3 codebase. It provides a dedicated update path, allowing existing TrueNAS CORE 13.3 installations to upgrade directly to FreeCORE 15.0‑U1 in place. Version 15.0‑U1 is designated as stable, and after the upgrade users remain on the FreeCORE development train for subsequent updates. The project therefore offers a seamless migration from the upstream TrueNAS CORE release to a self‑hosted, FreeBSD‑based distribution while preserving compatibility with the original system’s features and configuration.

California’s legislature approved Assembly Bill 1856, amending the Digital Age Assurance Act to exempt open‑source operating systems from its age‑verification requirements, effective before the law’s Jan 1 2027 start date. The bill redefines “operating system provider” to exclude any entity distributing software under licenses that allow copying, redistribution and modification—specifically GPL, MIT, BSD, and Apache—removing distributions such as Debian, Fedora, Ubuntu, Arch and BSD families from compliance. It also excludes non‑stand‑alone components (libraries, dependencies) delivered via package managers, and storefronts for extensions that run inside host applications, taking browser‑extension stores out of scope. The amendment deletes the original “user” definition that labeled every device owner a child, and adds a provision prohibiting age‑signal requests from OS providers or app stores unless legally required, providing a safe harbor for inaccurate signals. Windows, macOS, iOS and Android remain subject to age‑collection at account setup, with a July 1 2027 deadline for pre‑Jan 1 devices. The status of SteamOS remains unclear; GrapheneOS, distributed under MIT/Apache, is now fully exempt.

Tether is an open‑source project that bridges iPhone functionality to Linux desktops, aiming to replicate Apple’s Continuity features where technically feasible. Initial development focused on clipboard synchronization, enabling copy/paste between macOS and Linux (Wayland) and establishing a foundation for later capabilities. Security is built‑in: communication between the iOS app and the Linux daemon (tetherd) uses mutual TLS, with regular audits (Opus, Fable) for vulnerabilities. Subsequent additions include file transfer, OTP handling via WebExtension integrations for Zen Browser (Firefox) and Betterbird (Thunderbird), and, after reverse‑engineering Bluetooth protocols (ancs4linux, BlueFerry), support for iMessage, SMS, notifications, and contact sync. The Bluetooth implementation was written in C++ under an MIT license to avoid copyleft restrictions of related GPL projects. Tether’s roadmap emphasizes community contributions and aims to provide a reliable, privacy‑focused Continuity experience for Linux users with iPhones.

Benjamin Franklin repeatedly used pseudonymous personas to influence public opinion and advance reforms.

• Silence Dogood (teenage letters) let him bypass his brother’s publishing embargo and comment on women’s hardships.
• Polly Baker dramatized a legal case of an unwed mother, exposing gender bias in colonial courts.
• Richard Saunders authored Poor Richard’s Almanack, mixing astronomical data with maxims that shaped American folk wisdom.
• Americanus and Benevolus appeared in the Pennsylvania Gazette and British papers, respectively, to argue colonial rights, critique the transport of convicts, and counter the Stamp Act, using satire and reasoned rebuttals.
• Near his death, Franklin, as an abolitionist, submitted a petition to Congress and mocked pro‑slavery rhetoric with a fabricated Algiers speech.

In 1769, collaborating with his cousin Captain Timothy Folger, Franklin charted the Gulf Stream (“river in the ocean”), enabling ships to avoid the warm eastward current, cutting trans‑Atlantic voyages by up to two weeks and accelerating trade. His alter egos provided anonymity, rhetorical flexibility, and a platform for social critique while his scientific work directly impacted navigation and commerce.