Digest · 14 stories · 27 sources

2026-07-10

Updated

Suggest a story

Top stories

  1. Security confirmed Chat Control 1.0 scanning derogation survives EU Parliament vote On 2026-07-09 the European Parliament voted on the extension of the interim derogation (Regulation 2021/1232, called "Chat Control 1.0") that permits providers to voluntarily scan communications for child sexual abuse material. The vote ran under the ordinary legislative procedure second reading, where an absolute majority of at least 361 MEPs is required to reject or amend the text. Reporting states 314 MEPs voted to reject, 47 short of that threshold, so the derogation proceeds and runs until 2028-04-03. The derogation covers voluntary scanning on services that are not end-to-end encrypted (reported to include Gmail, Facebook Messenger, Instagram DMs, Skype, Snapchat, iCloud Mail, and Xbox) and does not mandate breaking end-to-end encryption.
  2. AI confirmed Meta ships Muse Spark 1.1 and opens the Meta Model API in public preview Meta Superintelligence Labs released Muse Spark 1.1 on 2026-07-09, a multimodal model aimed at agentic work: multi-agent orchestration with a 1 million token context and parallel subagent delegation, computer use, coding tasks (bug diagnosis, feature work, large migrations), and image, video, and audio input. Meta launched a public preview of the Meta Model API for developer access, with the model also available in "Thinking" mode in the Meta AI app and at meta.ai. Listed API pricing is 1.25 USD per 1M input tokens, 4.25 USD per 1M output tokens, and 0.15 USD per 1M cached input tokens. The announcement gives comparative claims but publishes no benchmark numbers.
  3. AI confirmed Tencent releases Hy3, a 295B open Mixture-of-Experts model Tencent Hunyuan released Hy3 on 2026-07-06 under Apache 2.0, a Mixture-of-Experts model with 295B total parameters and about 21B active per token and a 256K context window, positioned for agentic and long-context coding work. Tencent states the release follows a late-April preview and feedback from more than 50 internal product teams, and reports it rivals open models with 2 to 5 times the parameter count. Weights are on Hugging Face and ModelScope, and the model is free on OpenRouter until 2026-07-21.
  4. Languages discussion pgrust reimplements PostgreSQL in Rust and passes the regression suite pgrust (by GitHub user malisper) is a from-scratch reimplementation of PostgreSQL in Rust targeting compatibility with Postgres 18.3. The README states it passes 100% of the Postgres regression suite (more than 46,000 regression queries), is disk-compatible with existing Postgres 18.3 data directories, and can boot from an existing installation. The README also states it is not production-ready and not performance-optimized, that PL/Python, PL/Perl, and PL/Tcl are not generally compatible yet, and that the project used AI-assisted programming for deeper server changes. License is AGPL-3.0.

Conferences and events

  1. Event developing EuroPython 2026 starts in 3 days (2026-07-13) EuroPython 2026 runs 2026-07-13 to 2026-07-19. It is the main community Python conference and covers language, packaging, typing, performance, and ecosystem tracks.
  2. Event developing ICML 2026 is active through 2026-07-11 The International Conference on Machine Learning runs 2026-07-06 to 2026-07-11. The main conference and workshops cover training methods, reasoning, evaluation, and interpretability.

Developer tools

  1. Dev tools discussion Interview with Mitchell Hashimoto on Ghostty and Zig A published interview with Mitchell Hashimoto covers the design of the Ghostty terminal emulator and his experience building it in Zig, including tradeoffs in the language and the project's approach to performance and platform support.

Linux and kernel

  1. Linux/Kernel developing Initial patches boot the Apple M4 on Linux Developer Yureka Lilian posted the first device trees and bindings to boot Apple Silicon M4 systems on Linux, reported by Phoronix on 2026-07-09. The M4 bring-up is closer to the M3 than the M2-to-M3 step. Most changes are in the m1n1 bootloader, which no longer sets CPU configuration bits because iBoot now sets and locks them. The patches only reach a bootable state without working peripherals or a usable desktop, and SMP boot depends on idle=nop patches and remains unstable.

Infrastructure

  1. Infrastructure confirmed IERS confirms no leap second at the end of 2026 The International Earth Rotation and Reference Systems Service issued Bulletin C 72 on 2026-07-06, announcing that no leap second will be introduced at the end of December 2026. No positive or negative leap second is scheduled. The difference between UTC and TAI stays at UTC minus TAI equal to -37 seconds, unchanged since 2017-01-01.
  2. Infrastructure confirmed Prometheus 3.13.1 fixes a TSDB range-query correctness bug Prometheus released 3.13.1 on 2026-07-10, a bugfix release for the 3.13 LTS line. It fixes a TSDB defect where the head-chunk cache could return samples from the wrong chunk, or spurious not-found errors, on range queries after a head-chunk truncation.

Engineering posts

  1. Engineering post discussion Good tools are invisible Ginger Bill (creator of the Odin language) argues that the best tools recede into the background and let the user focus on the work rather than on the tool. The post contends that toolmakers should favor sensible defaults and low-friction usability over maximal configurability, and that a steep learning curve is a cost to be justified, not a virtue. It uses text editors, GUI versus terminal applications, and Linux desktop configuration culture as examples, drawing on 15 years of the author's own editor use.
  2. Engineering post discussion Write code like a human will maintain it The post argues that code-quality discipline matters more, not less, when a large language model writes the code. Its claim is that LLMs reproduce the patterns already present in a codebase, so merging duplicated or poorly structured AI-generated code trains the assistant to repeat those shortcuts on later requests, compounding the problem. The author recommends holding AI-written code to the same review standard as human-written code rather than assuming a future tool will clean it up.

Hacker News

  1. Pulse discussion Show HN: colibri streams model weights from SSD to run GLM 5.2 on low-RAM machines A Show HN project (colibri) runs the large GLM 5.2 model on a memory-constrained machine by streaming weights from an SSD or NVMe drive rather than holding them in RAM. The thread reached 494 points.
  2. Pulse discussion HN thread on pgrust questions AI-assisted rewrites and test-based reliability The 451-comment thread on pgrust (covered in Top stories) surfaces a recurring argument about database reliability.