<h2>Must know</h2>
<ul>
<li><strong>Canonical joins Rust Foundation as Gold Member</strong>: Canonical is doubling down on Rust, joining the Foundation as a Gold Member to influence the language’s direction in Linux infrastructure. The company plans to replace core GNU utilities with Rust-based alternatives (uutils) starting in Ubuntu 25.10, signaling a major shift in software supply chain security for the most popular Linux distro. This move aims to reduce memory-safety vulnerabilities in regulated environments and includes a focus on securing <a href="http://crates.io" target="_blank" rel="noopener nofollow">crates.io</a> dependencies. <a href="https://app.daily.dev/feed-by-ids?id=6DSn3Gmgy&amp;id=clix7apkf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vite 8 adopts Rust-based Rolldown for massive speed gains</strong>: The web development ecosystem is seeing a performance leap as Vite 8 replaces its dual esbuild/Rollup architecture with Rolldown, a unified bundler written in Rust. This change delivers 10–30x faster production builds while maintaining full plugin compatibility, addressing long-standing developer experience bottlenecks. Meanwhile, TypeScript 6 has been released as a bridge toward a native Go-powered compiler coming in version 7. <a href="https://app.daily.dev/posts/QG6z68w6t" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux 7.0-rc5 and the pghot memory subsystem</strong>: Linus Torvalds has released the fifth candidate for Linux 7.0, which includes a new ‘pghot’ subsystem from AMD for intelligent hot-page tracking. Benchmarks on Zen 5 servers show measurable performance gains in tiered memory scenarios by unifying how the kernel detects and promotes frequently accessed pages. The release also cleans up legacy code, including the removal of UDP Lite, which reportedly yields a 10% networking speedup. <a href="https://app.daily.dev/feed-by-ids?id=1tWnyX7nj&amp;id=9z2uCLEQw&amp;id=POd1HGEP1&amp;id=QlGhzzwwW" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Qt 6.11 introduces hardware-accelerated 2D graphics API</strong>: The new Qt Canvas Painter offers an HTML5-inspired API that outperforms the old QPainter OpenGL backend. <a href="https://app.daily.dev/posts/JY6Nxd1Co" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wix migrates 1,000 MySQL servers to Graviton with zero downtime</strong>: The team used Ansible automation to move 160 clusters to ARM, cutting CPU utilization by 50% and reducing costs. <a href="https://app.daily.dev/posts/mlsIwgQ0W" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Flash-MoE runs 397B parameter models on a MacBook Pro</strong>: A pure C/Metal engine achieves 4.4 tokens/second by streaming active experts from SSD using parallel pread calls. <a href="https://app.daily.dev/posts/xsGaEEFUf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NixOS gains traction for deterministic AI agent environments</strong>: Developers are using Nix flakes to provide LLM agents with exact tool versions in isolated, reproducible shells.</li>
<li><strong>Dapr Agents v1.0 brings Python AI frameworks to Kubernetes</strong>: The GA release focuses on infrastructure resilience, offering durable workflows and persistent state across 30+ databases. <a href="https://app.daily.dev/posts/uKbm4srSE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Aperture by Tailscale centralizes LLM API key management</strong>: The new gateway uses Tailscale identity to authenticate requests, reducing credential sprawl across engineering teams. <a href="https://app.daily.dev/posts/yy7yOwfVy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MkDocs ecosystem fractures after maintainer governance failure</strong>: The popular documentation tool has split into three forks (ProperDocs, MaterialX, Zensical) following a private rewrite by the founder. <a href="https://app.daily.dev/posts/GedRmjUpE" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Elon Musk’s Terafab semiconductor pipe dream</strong>: Musk is promising a ‘Terafab’ facility to build chips for Tesla and SpaceX, despite having zero experience in the most complex manufacturing process on Earth. Given his track record of ‘Full Self-Driving’ timelines and recent liability for misleading investors, this looks like another attempt to pump stock with sci-fi promises. Building a fab requires more than just a 200-gigawatt power plan; it requires physics and supply chains that don’t care about tweets. <a href="https://app.daily.dev/feed-by-ids?id=qNbywR4ri&amp;id=rLaUkgjrd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The rise of Generative Engine Optimization (GEO)</strong>: SEO agencies are already pivoting to ‘GEO,’ selling tools to help brands ‘optimize’ their presence in AI-generated answers from ChatGPT and Perplexity. It’s the same old snake oil in a new bottle, promising to game black-box probabilistic models that don’t even guarantee the same output twice. If you’re paying for ‘citation intelligence’ to trick an LLM, you’re just funding the next generation of digital noise. <a href="https://app.daily.dev/posts/6t97IdU4o" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The death of model.fit() and the birth of vibe-checking</strong>: Data scientists are being told their new job is ‘Evaluation-Driven Development,’ which is a fancy way of saying ‘reading chat logs to see if the bot messed up.’ While the industry rebrands manual QA as ‘curating golden datasets,’ the actual engineering rigor is being replaced by ‘inter-rater agreement metrics.’ It’s a desperate attempt to make the unpredictable nature of LLMs feel like a disciplined science. <a href="https://app.daily.dev/posts/seweaAUF7" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
