<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>News on Golang4U</title><link>https://golang4u.com/news/</link><description>Recent content in News on Golang4U</description><generator>Hugo</generator><language>en</language><lastBuildDate>Sat, 24 Oct 2026 09:00:00 +0000</lastBuildDate><atom:link href="https://golang4u.com/news/index.xml" rel="self" type="application/rss+xml"/><item><title>KDE Plasma 6.9 Stabilizes KWin's Wayland Picture-In-Picture Support</title><link>https://golang4u.com/news/2026-10-25-kde-plasma-6-9-stabilizes-kwin-s-wayland-picture-in-picture-/</link><pubDate>Sat, 24 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-25-kde-plasma-6-9-stabilizes-kwin-s-wayland-picture-in-picture-/</guid><description>&lt;p&gt;With the wonderful KDE Plasma 6.8 having shipped just more than one week ago, KDE developers are beginning to go heavy on new features for Plasma 6.9&amp;hellip;&lt;/p&gt;
&lt;p&gt;&lt;a href="https://www.phoronix.com/news/KDE-Plasma-6.9-Wayland-PiP" target="_blank" rel="noopener noreferrer"&gt;Read on the source site&lt;/a&gt;&lt;/p&gt;</description></item><item><title>Linux 7.4 Workqueue Changes Bring A '10 To 100 Times Faster' Optimization</title><link>https://golang4u.com/news/2026-10-24-linux-7-4-workqueue-changes-bring-a-10-to-100-times-faster-o/</link><pubDate>Fri, 23 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-24-linux-7-4-workqueue-changes-bring-a-10-to-100-times-faster-o/</guid><description>&lt;p&gt;Adding to all the exciting performance optimizations and other improvements for Linux 7.4 is an exciting workqueue improvement. It&amp;rsquo;s not too often getting to talk about &amp;ldquo;10 to 100 times faster&amp;rdquo; software improvements&amp;hellip;&lt;/p&gt;
&lt;p&gt;&lt;a href="https://www.phoronix.com/news/Linux-7.4-Workqueue" target="_blank" rel="noopener noreferrer"&gt;Read on the source site&lt;/a&gt;&lt;/p&gt;</description></item><item><title>Linux 7.4 Begins Landing Patches To Speed Up The Kernel Build Process</title><link>https://golang4u.com/news/2026-10-23-linux-7-4-begins-landing-patches-to-speed-up-the-kernel-buil/</link><pubDate>Thu, 22 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-23-linux-7-4-begins-landing-patches-to-speed-up-the-kernel-buil/</guid><description>&lt;p&gt;There are notable patches by Arm engineer Lorenzo Stoakes that have begun landing in Linux 7.4 for speeding up the Linux kernel build process. This work can provide some significant reductions in the Linux kernel build time&amp;hellip;&lt;/p&gt;
&lt;p&gt;&lt;a href="https://www.phoronix.com/news/Linux-7.4-Kbuild" target="_blank" rel="noopener noreferrer"&gt;Read on the source site&lt;/a&gt;&lt;/p&gt;</description></item><item><title>AMD GAIA 0.25 Brings Several Improvements For Local AI Agents On Ryzen AI</title><link>https://golang4u.com/news/2026-10-21-amd-gaia-0-25-brings-several-improvements-for-local-ai-agent/</link><pubDate>Tue, 20 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-21-amd-gaia-0-25-brings-several-improvements-for-local-ai-agent/</guid><description>&lt;p&gt;AMD GAIA 0.25 is out this week as the latest feature release for this open-source project that began as &amp;ldquo;Generative AI Is Awesome&amp;rdquo; and is now going on a tag line of &amp;ldquo;AMD&amp;rsquo;s open-source agent factory&amp;rdquo; in building/using local AI agents on AMD Ryzen AI hardware&amp;hellip;&lt;/p&gt;
&lt;p&gt;&lt;a href="https://www.phoronix.com/news/AMD-GAIA-0.25-Released" target="_blank" rel="noopener noreferrer"&gt;Read on the source site&lt;/a&gt;&lt;/p&gt;</description></item><item><title>Linux 7.3 Squeezes Even More Performance Out Of The AMD Threadripper PRO 9995WX</title><link>https://golang4u.com/news/2026-10-18-linux-7-3-squeezes-even-more-performance-out-of-the-amd-thre/</link><pubDate>Sun, 18 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-18-linux-7-3-squeezes-even-more-performance-out-of-the-amd-thre/</guid><description>&lt;p&gt;On October 16, 2026, Michael Larabel published the latest in a long-running Phoronix series: benchmarks of the &lt;strong&gt;Linux 7.3&lt;/strong&gt; kernel state on a &lt;strong&gt;96-core AMD Ryzen Threadripper PRO 9995WX&lt;/strong&gt; workstation (192 threads), comparing it against the shipping Linux 7.2. The headline, in the author&amp;rsquo;s words, is that the kernel &amp;ldquo;looks great&amp;rdquo; — and the numbers back it up. I/O tests for &lt;strong&gt;PCIe Gen 4/5 NVMe&lt;/strong&gt; storage posted clear gains, and that I/O improvement propagated into real application workloads: &lt;strong&gt;PostgreSQL&lt;/strong&gt;, &lt;strong&gt;MariaDB&lt;/strong&gt;, and &lt;strong&gt;TigerBeetle&lt;/strong&gt; (the Go-based financial-transactions database) all ran faster on 7.3, network throughput came up, and the kernel&amp;rsquo;s own micro-benchmarks — &lt;code&gt;fork&lt;/code&gt;, &lt;strong&gt;&lt;code&gt;io_uring&lt;/code&gt;&lt;/strong&gt;, and &lt;strong&gt;mutex&lt;/strong&gt; contention — all moved. Compilation workloads were &amp;ldquo;subtly but consistently faster,&amp;rdquo; and CPU-based AI inference showed small gains.&lt;/p&gt;</description></item><item><title>Go 1.27 is released</title><link>https://golang4u.com/news/2026-10-17-go-1-27-is-released/</link><pubDate>Sat, 17 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-17-go-1-27-is-released/</guid><description>&lt;p&gt;Go 1.27 was released in August 2026, and it is one of the larger releases in the language&amp;rsquo;s history: &lt;strong&gt;generic methods&lt;/strong&gt; finally exist, the standard library gained a &lt;strong&gt;&lt;code&gt;uuid&lt;/code&gt; package&lt;/strong&gt;, a new &lt;strong&gt;post-quantum&lt;/strong&gt; signature scheme made it into &lt;code&gt;crypto/tls&lt;/code&gt; and &lt;code&gt;crypto/x509&lt;/code&gt;, a long-anticipated &lt;strong&gt;&lt;code&gt;encoding/json/v2&lt;/code&gt;&lt;/strong&gt; package landed (and quietly became the implementation behind the v1 &lt;code&gt;encoding/json&lt;/code&gt;), the runtime got &lt;strong&gt;size-specialized memory allocation&lt;/strong&gt; and a &lt;strong&gt;generally-available goroutine-leak profile&lt;/strong&gt;, and the toolchain gained a batch of new &lt;code&gt;go fix&lt;/code&gt; modernizers.&lt;/p&gt;</description></item><item><title>Go Power: Microsoft’s Bold Bet on Faster TypeScript Tools</title><link>https://golang4u.com/news/2026-10-17-go-power-microsoft-s-bold-bet-on-faster-typescript-tools/</link><pubDate>Sat, 17 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-17-go-power-microsoft-s-bold-bet-on-faster-typescript-tools/</guid><description>&lt;h1 id="go-power-microsofts-bold-bet-on-faster-typescript-tools" class="headerLink"&gt;
 &lt;a href="#go-power-microsofts-bold-bet-on-faster-typescript-tools" class="header-mark"&gt;&lt;/a&gt;Go Power: Microsoft’s Bold Bet on Faster TypeScript Tools&lt;/h1&gt;&lt;p&gt;In March 2025, Anders Hejlsberg — co-creator of TypeScript — announced &lt;strong&gt;Corsa&lt;/strong&gt;, an initiative to port the TypeScript compiler and its language service to a native implementation &lt;strong&gt;written in Go&lt;/strong&gt;. The numbers behind the bet: &lt;strong&gt;build times down by roughly 10x&lt;/strong&gt;, project load &lt;strong&gt;8x faster&lt;/strong&gt;, and about &lt;strong&gt;half the memory usage&lt;/strong&gt; of the JavaScript implementation, with a language service that makes completions, quick info, go-to-definition, and find-all-references dramatically snappier.&lt;/p&gt;</description></item><item><title>Introducing the pkg.go.dev API</title><link>https://golang4u.com/news/2026-10-17-introducing-the-pkg-go-dev-api/</link><pubDate>Sat, 17 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-17-introducing-the-pkg-go-dev-api/</guid><description>&lt;p&gt;For years the only way to get data &lt;em&gt;out&lt;/em&gt; of pkg.go.dev was to scrape the HTML. Dependency dashboards, CI checkers, internal package browsers, and — increasingly — AI assistants that need to reason about the Go ecosystem all ended up parsing &lt;code&gt;&amp;lt;a&amp;gt;&lt;/code&gt; tags from rendered package pages. That was never going to survive a refactoring. On May 21, 2026, the Go team shipped the fix: an official, documented, versioned &lt;strong&gt;REST API&lt;/strong&gt; for pkg.go.dev, currently served under the &lt;code&gt;/v1beta&lt;/code&gt; path, with a machine-readable &lt;strong&gt;OpenAPI specification&lt;/strong&gt; published alongside it.&lt;/p&gt;</description></item><item><title>Linux Lands Fix for Rare, Varying SMP Boot Hangs Affecting Kernels Since 2024</title><link>https://golang4u.com/news/2026-10-17-linux-lands-fix-for-rare-varying-smp-boot-hangs-affecting-ke/</link><pubDate>Sat, 17 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-17-linux-lands-fix-for-rare-varying-smp-boot-hangs-affecting-ke/</guid><description>&lt;h1 id="linux-lands-fix-for-rare-varying-smp-boot-hangs-affecting-kernels-since-2024" class="headerLink"&gt;
 &lt;a href="#linux-lands-fix-for-rare-varying-smp-boot-hangs-affecting-kernels-since-2024" class="header-mark"&gt;&lt;/a&gt;Linux Lands Fix for Rare, Varying SMP Boot Hangs Affecting Kernels Since 2024&lt;/h1&gt;&lt;p&gt;Kernel bugs that are &lt;em&gt;rare&lt;/em&gt; and &lt;em&gt;vary by hardware&lt;/em&gt; are the hardest kind to diagnose, because the reproduction is not a unit test you can run on a loop — it is a race that only fires when the scheduler happens to place one thread on one CPU at one exact millisecond of boot. That is exactly the class of bug Linus&amp;rsquo;s tree just fixed ahead of Linux 7.3, and it is a clean case study in &lt;strong&gt;concurrency timing&lt;/strong&gt; that maps one-for-one onto Go: a set of worker threads that start too early, race a global setup phase, and intermittually deadlock the system.&lt;/p&gt;</description></item><item><title>Platform-independent SIMD in Go</title><link>https://golang4u.com/news/2026-10-17-platform-independent-simd-in-go/</link><pubDate>Sat, 17 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-17-platform-independent-simd-in-go/</guid><description>&lt;p&gt;Before 1.26, the only way to get SIMD performance out of a Go program was to write Go assembly, a bar high enough that plenty of code that &lt;em&gt;could&lt;/em&gt; vectorize simply never did. The experimental &lt;code&gt;archsimd&lt;/code&gt; package (1.26) lowered that bar for &lt;code&gt;amd64&lt;/code&gt;, and 1.27 extends it to &lt;code&gt;arm64&lt;/code&gt; NEON and &lt;code&gt;wasm&lt;/code&gt;. But architecture-specific SIMD has a second problem that assembly never fully solves: &lt;strong&gt;the platforms disagree about everything&lt;/strong&gt; — vector sizes, which operations exist, how masks are represented, which features need runtime detection. &lt;code&gt;amd64&lt;/code&gt; has 128/256/512-bit vectors; &lt;code&gt;arm64&lt;/code&gt; has 128-bit NEON and 128–2048-bit SVE; &lt;code&gt;wasm&lt;/code&gt; has a single fixed 128-bit size. &lt;code&gt;AVX512&lt;/code&gt; uses dedicated mask registers; &lt;code&gt;NEON&lt;/code&gt; and &lt;code&gt;wasm&lt;/code&gt; get by with vector bitmasks; &lt;code&gt;SVE&lt;/code&gt; allocates one bit per byte with the least-significant bit governing the operation.&lt;/p&gt;</description></item><item><title>Results from the 2026 TAB election</title><link>https://golang4u.com/news/2026-10-17-results-from-the-2026-tab-election/</link><pubDate>Sat, 17 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-17-results-from-the-2026-tab-election/</guid><description>&lt;p&gt;On October 16, 2026, LWN reported a small headline with outsized consequences for anyone who runs Go on Linux: the results of the &lt;strong&gt;2026 election for the Linux Foundation Technical Advisory Board (TAB)&lt;/strong&gt; were announced. &lt;strong&gt;Kees Cook, Dave Hansen, Miguel Ojeda, and Shuah Khan&lt;/strong&gt; have all been &lt;strong&gt;re-elected&lt;/strong&gt;, and &lt;strong&gt;Laurent Pinchart&lt;/strong&gt; is &lt;strong&gt;joining the board&lt;/strong&gt;.&lt;/p&gt;
&lt;p&gt;The announcement is three sentences long. The reason it matters is that the TAB is the top rung of the governance chain that every line of the Linux kernel passes through before it lands — and the people who are re-elected there decide what the kernel is allowed to be. If your production Go servers, your &lt;code&gt;golang.org/x/&lt;/code&gt; dependencies, your containers, and your eBPF programs all sit on top of the Linux kernel, the TAB is where the rules get made. This post unpacks what the TAB actually is, how the kernel&amp;rsquo;s governance chain works, what the 2026 roster change means, and what it implies for Go developers in practice.&lt;/p&gt;</description></item><item><title>Type Construction and Cycle Detection</title><link>https://golang4u.com/news/2026-10-17-type-construction-and-cycle-detection/</link><pubDate>Sat, 17 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-17-type-construction-and-cycle-detection/</guid><description>&lt;h1 id="type-construction-and-cycle-detection" class="headerLink"&gt;
 &lt;a href="#type-construction-and-cycle-detection" class="header-mark"&gt;&lt;/a&gt;Type Construction and Cycle Detection&lt;/h1&gt;&lt;p&gt;Go&amp;rsquo;s static typing is a big part of why it is a good fit for production systems that have to be robust. Every time &lt;code&gt;go build&lt;/code&gt; runs, the compiler first parses each package into an AST and hands that AST to the type checker. In Go 1.26, the language team overhauled a part of that type checker that most programmers never think about: how it &lt;em&gt;constructs&lt;/em&gt; types, and how it detects when a type&amp;rsquo;s definition is circular. The result: a simpler construction algorithm, and a class of corner-case &lt;strong&gt;compiler panics&lt;/strong&gt; that now produce ordinary type errors.&lt;/p&gt;</description></item><item><title>Arch-specific SIMD in Go</title><link>https://golang4u.com/news/2026-10-15-arch-specific-simd-in-go/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-arch-specific-simd-in-go/</guid><description>&lt;p&gt;Go 1.26 gave amd64 an experimental SIMD API. Go 1.27&amp;rsquo;s headline change: the same API now lands on &lt;strong&gt;arm64 (NEON) and wasm&lt;/strong&gt;, and the design rules that will shape the API for years are settled. If you have ever wanted to touch machine-wide registers from Go without writing assembly, this is the post.&lt;/p&gt;
&lt;h2 id="what-simd-actually-is" class="headerLink"&gt;
 &lt;a href="#what-simd-actually-is" class="header-mark"&gt;&lt;/a&gt;What SIMD actually is&lt;/h2&gt;&lt;p&gt;Single Instruction, Multiple Data: one instruction operates on a wide register (128, 256, 512 bits) holding several 8- to 64-bit elements, in parallel. &lt;code&gt;add 8 float64 values&lt;/code&gt; is one instruction. For uniform, element-parallel work — crypto, signal processing, data shuffling, numeric kernels — this is a real multiplier, not a micro-optimization.&lt;/p&gt;</description></item><item><title>Better HTTP/1 connection reuse</title><link>https://golang4u.com/news/2026-10-15-better-http-1-connection-reuse/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-better-http-1-connection-reuse/</guid><description>&lt;p&gt;100 sequential &lt;code&gt;http.Get&lt;/code&gt; calls against &lt;code&gt;index.golang.org&lt;/code&gt;, HTTP/1 only: &lt;strong&gt;710 ms on Go 1.26, 150 ms on Go 1.27&lt;/strong&gt;. Same program, same machine, same code — the 5x comes from a behavioral change in &lt;code&gt;net/http&lt;/code&gt; about what happens when you &lt;em&gt;don&amp;rsquo;t read a response body&lt;/em&gt;. This is the most-likely-to-quietly-help-you item in the 1.27 release, and it is also a lesson in how HTTP/1 connection reuse actually works.&lt;/p&gt;
&lt;h2 id="what-a-new-http1-connection-really-costs" class="headerLink"&gt;
 &lt;a href="#what-a-new-http1-connection-really-costs" class="header-mark"&gt;&lt;/a&gt;What a new HTTP/1 connection really costs&lt;/h2&gt;&lt;p&gt;Every HTTP version runs over TCP, usually over TLS. The setup ladder before your first request byte flies:&lt;/p&gt;</description></item><item><title>Claude Code / AI Shows Potential For Reverse-Engineering Hardware For Linux Support</title><link>https://golang4u.com/news/2026-10-15-claude-code-ai-shows-potential-for-reverse-engineering-hardw/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-claude-code-ai-shows-potential-for-reverse-engineering-hardw/</guid><description>&lt;h1 id="claude-code-reverse-engineered-a-scanner-the-80-of-hardware-re-that-is-tedious" class="headerLink"&gt;
 &lt;a href="#claude-code-reverse-engineered-a-scanner-the-80-of-hardware-re-that-is-tedious" class="header-mark"&gt;&lt;/a&gt;Claude Code Reverse-Engineered a Scanner: the 80% of hardware RE that is tedious&lt;/h1&gt;&lt;p&gt;Phoronix&amp;rsquo;s report is short on &lt;em&gt;hype&lt;/em&gt; and long on &lt;em&gt;evidence&lt;/em&gt;: &lt;strong&gt;Christian Schaller of Red Hat&lt;/strong&gt; — a long-time Linux driver reverse engineer — used &lt;strong&gt;Claude Code&lt;/strong&gt; to reverse-engineer the &lt;strong&gt;Pustek 8200i&lt;/strong&gt; negative-photo scanner (a niche device with &lt;em&gt;no&lt;/em&gt; Linux driver) into a &lt;strong&gt;working SANE driver&lt;/strong&gt;, then wrote a Rust GNOME app to use it. The &lt;em&gt;method&lt;/em&gt; is the durable part: Claude Code drove a &lt;strong&gt;Windows VM running the vendor&amp;rsquo;s own driver&lt;/strong&gt;, &lt;strong&gt;captured the USB traffic&lt;/strong&gt;, and &lt;strong&gt;reverse-engineered the protocol&lt;/strong&gt; from the captures.&lt;/p&gt;</description></item><item><title>Code That Passes Every Test Can Still Break the Next AI Agent That Touches It</title><link>https://golang4u.com/news/2026-10-15-code-that-passes-every-test-can-still-break-the-next-ai-agen/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-code-that-passes-every-test-can-still-break-the-next-ai-agen/</guid><description>&lt;h1 id="code-that-passes-every-test-can-still-break-the-next-ai-agent-that-touches-it" class="headerLink"&gt;
 &lt;a href="#code-that-passes-every-test-can-still-break-the-next-ai-agent-that-touches-it" class="header-mark"&gt;&lt;/a&gt;Code that passes every test can still break the next AI agent that touches it&lt;/h1&gt;&lt;p&gt;The New Stack&amp;rsquo;s framing of Go in the agent era is a &lt;em&gt;reframe&lt;/em&gt;, not a &lt;em&gt;hype&lt;/em&gt; cycle: &lt;strong&gt;the language that wins the agent era is the one a &lt;em&gt;non-author&lt;/em&gt; can &lt;em&gt;read, understand, and fix&lt;/em&gt; at the lowest cost&lt;/strong&gt; — and that is the property Go currently has the most of. The counterintuitive part: it is &lt;em&gt;not&lt;/em&gt; because AI writes Go best (it writes every language competently), but because a &lt;em&gt;second&lt;/em&gt; agent — or the human on call at 3am — can &lt;em&gt;read&lt;/em&gt; Go better than it can read the alternatives. The &lt;strong&gt;test suite is the floor&lt;/strong&gt; that makes the first agent&amp;rsquo;s work look right; the &lt;strong&gt;reading cost is the ceiling&lt;/strong&gt; that decides whether the second agent can maintain it at all.&lt;/p&gt;</description></item><item><title>Docker, K8s, and Terraform: The Go Ops Triad That Actually Ships</title><link>https://golang4u.com/news/docker-k8s-terraform-the-go-ops-triad-that-actually-ships/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/docker-k8s-terraform-the-go-ops-triad-that-actually-ships/</guid><description>&lt;p&gt;The three tools every Go deploy runs on are all written in Go: &lt;strong&gt;Docker&lt;/strong&gt; builds the image, &lt;strong&gt;Kubernetes&lt;/strong&gt; runs the image, &lt;strong&gt;Terraform&lt;/strong&gt; provisions the infrastructure the image runs on. That is the whole ops pipeline — &lt;code&gt;build → run → provision&lt;/code&gt; — and the reason it is boring is that each step is a Go binary with a stable CLI and a well-defined exit code. The tools that do not ship (Ansible, Chef, Puppet, Salt) are all config-push frameworks: agent-based, stateful, and one more thing to keep alive. The triad is pull-based, declarative, and stateless — which is exactly what Go&amp;rsquo;s &lt;code&gt;go build&lt;/code&gt; culture likes.&lt;/p&gt;</description></item><item><title>Go 1.25: The Release That Quietly Removes an Entire Class of Bugs</title><link>https://golang4u.com/news/go-1-25-the-release-that-quietly-removes-an-entire-class-of-bugs/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/go-1-25-the-release-that-quietly-removes-an-entire-class-of-bugs/</guid><description>&lt;p&gt;Most Go releases add a thing: a syntax, a package, a flag. Go 1.25 is the release that stopped adding features and started &lt;em&gt;removing&lt;/em&gt; the things that cause 3am pages — no new syntax, no new standard-library package, no new &lt;code&gt;GODEBUG&lt;/code&gt; flag you have to remember. Just a series of small, surgical changes to the compiler and the runtime that take whole &lt;em&gt;categories&lt;/em&gt; of bug off the table.&lt;/p&gt;
&lt;p&gt;The thesis of this post: &lt;strong&gt;a release&amp;rsquo;s value is measured by the bugs it makes impossible, not the features it adds.&lt;/strong&gt; Go 1.25 is the first release I would call &amp;ldquo;the boring one&amp;rdquo; in the good way — it is the release you upgrade to and then never think about again, because a whole class of &amp;ldquo;the service is slow and I find out in three weeks&amp;rdquo; and &amp;ldquo;the test passes 9 times out of 10&amp;rdquo; bugs stops being yours to chase.&lt;/p&gt;</description></item><item><title>Go Experts: 'I Don't Want to Maintain AI-Generated Code'</title><link>https://golang4u.com/news/2026-10-15-go-experts-i-don-t-want-to-maintain-ai-generated-code/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-go-experts-i-don-t-want-to-maintain-ai-generated-code/</guid><description>&lt;h1 id="go-experts-i-dont-want-to-maintain-ai-generated-code" class="headerLink"&gt;
 &lt;a href="#go-experts-i-dont-want-to-maintain-ai-generated-code" class="header-mark"&gt;&lt;/a&gt;Go Experts: &amp;ldquo;I Don&amp;rsquo;t Want to Maintain AI-Generated Code&amp;rdquo;&lt;/h1&gt;&lt;p&gt;The New Stack ran a conversation with &lt;strong&gt;two long-time Go programmers&lt;/strong&gt; on the future of their language in the age of AI agents — and the headline is the punchline of the title: &lt;strong&gt;&amp;ldquo;I don&amp;rsquo;t want to maintain AI-generated code.&amp;rdquo;&lt;/strong&gt; The &lt;em&gt;generation&lt;/em&gt; cost of AI code is the easy half; the &lt;em&gt;maintenance&lt;/em&gt; cost is the expensive half, and it is the half that decides which languages win the agent era. The conversation is worth reading in full, but the durable lesson is a &lt;em&gt;systems&lt;/em&gt; one: &lt;strong&gt;a language&amp;rsquo;s fitness for AI agents is measured by how cheap it is for a &lt;em&gt;human&lt;/em&gt; (or a later agent) to read, understand, and fix the code the first agent wrote — not by how fast the first agent wrote it.&lt;/strong&gt;&lt;/p&gt;</description></item><item><title>Ubuntu 26.10 released</title><link>https://golang4u.com/news/2026-10-15-ubuntu-26-10-released/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-ubuntu-26-10-released/</guid><description>&lt;p&gt;Ubuntu 26.10 &amp;ldquo;Stonking Stingray&amp;rdquo; shipped on 15 October 2026 — six months after the 26.04 LTS — and it is the first Ubuntu release with &lt;strong&gt;RISC-V desktop images&lt;/strong&gt; and a &lt;strong&gt;dedicated amd64v3 image&lt;/strong&gt;. For a Go shop that cross-compiles, a distro release is not &amp;ldquo;the OS updated.&amp;rdquo; It is a new target in the matrix, a new kernel under the runtime, and a new set of toolchain questions. This post frames the release through the Go lens.&lt;/p&gt;</description></item><item><title>Ubuntu 26.10 Released With Linux 7.3 + GNOME 51, Rust Coreutils Migration Complete</title><link>https://golang4u.com/news/2026-10-15-ubuntu-26-10-released-with-linux-7-3-gnome-51-rust-coreutils/</link><pubDate>Thu, 15 Oct 2026 09:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-ubuntu-26-10-released-with-linux-7-3-gnome-51-rust-coreutils/</guid><description>&lt;p&gt;Ubuntu 26.10 &amp;ldquo;Stonking Stingray&amp;rdquo; is out — the six-month, non-LTS update after 26.04, per &lt;a href="https://www.phoronix.com/news/Ubuntu-26.10-Released" target="_blank" rel="noopener noreferrer"&gt;Phoronix&lt;/a&gt;: &lt;strong&gt;Linux 7.3&lt;/strong&gt; in near-final shape, &lt;strong&gt;GNOME 51&lt;/strong&gt;, the &lt;strong&gt;Rust &lt;code&gt;uutils&lt;/code&gt; coreutils migration completed&lt;/strong&gt;, TPM-backed full-disk encryption, and — the line that matters for this site — &lt;strong&gt;RISC-V desktop images and an optimized amd64v3 image&lt;/strong&gt; for the first time. This is the &amp;ldquo;new distro, new targets&amp;rdquo; post: what a Go toolchain shop does the week a distro lands.&lt;/p&gt;</description></item><item><title/><link>https://golang4u.com/news/2026-10-15-kde-releases-documentary/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-kde-releases-documentary/</guid><description>&lt;p&gt;title: &amp;ldquo;KDE Releases Its 30th Anniversary Documentary: A Story of Free-Software Underdogs&amp;rdquo;
image: /images/covers/2026-10-15-kde-releases-documentary.svg
date: 2026-10-15T09:00:00+00:00
author: &amp;ldquo;Punit Naran&amp;rdquo;
description: &amp;ldquo;KDE&amp;rsquo;s 30th-anniversary documentary is a systems story: a 30-year project that had to keep a whole desktop stack coherent, whose freeness depended on a commercial company&amp;rsquo;s license change, and whose side project (KHTML) out-competed the web. The Go lens: a 40-line stdlib fetcher that verifies the announcement is live, plus the three lessons a 30-year-old codebase carries.&amp;rdquo;
tags: [&amp;ldquo;Go&amp;rdquo;, &amp;ldquo;Community&amp;rdquo;, &amp;ldquo;Linux&amp;rdquo;, &amp;ldquo;KDE&amp;rdquo;, &amp;ldquo;History&amp;rdquo;]
draft: false&lt;/p&gt;</description></item><item><title/><link>https://golang4u.com/news/2026-10-15-lpddr5-5467-vs-6400-vs-7467-mt-s-memory-performance-benchmar/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-lpddr5-5467-vs-6400-vs-7467-mt-s-memory-performance-benchmar/</guid><description>&lt;p&gt;title: &amp;ldquo;LPDDR5 5467 vs. 6400 vs. 7467 MT/s: what the memory speed actually costs a laptop&amp;rdquo;
image: /images/covers/2026-10-15-lpddr5-5467-vs-6400-vs-7467-mt-s-memory-performance-benchmar.svg
date: 2026-10-15T09:00:00+00:00
author: &amp;ldquo;Punit Naran&amp;rdquo;
description: &amp;ldquo;Phoronix benchmarks the Star Labs StarFighter Ultra (Core Ultra 9 285H, 8x8GB LPDDR5-7467) at three memory speeds. The vendor default is 6400, not the rated 7467. The Go lens: a ~60-line stdlib Go microbenchmark is the cheapest &amp;rsquo;lab instrument&amp;rsquo; a laptop with a speed knob in the BIOS can have, and the interesting number is the perf-per-watt ratio, not the raw delta.&amp;rdquo;
tags: [&amp;ldquo;Go&amp;rdquo;, &amp;ldquo;Community&amp;rdquo;, &amp;ldquo;Hardware&amp;rdquo;, &amp;ldquo;Memory&amp;rdquo;, &amp;ldquo;Benchmarks&amp;rdquo;]
draft: false&lt;/p&gt;</description></item><item><title/><link>https://golang4u.com/news/2026-10-15-microsoft-joins-google-in-backing-go-for-ai-agents-openai-an/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-microsoft-joins-google-in-backing-go-for-ai-agents-openai-an/</guid><description>&lt;p&gt;title: &amp;ldquo;Microsoft Joins Google in Backing Go for AI Agents — OpenAI and Anthropic Lag&amp;rdquo;
image: /images/covers/2026-10-15-microsoft-joins-google-in-backing-go-for-ai-agents-openai-an.svg
date: 2026-10-15T09:00:00+00:00
author: &amp;ldquo;Punit Naran&amp;rdquo;
description: &amp;ldquo;Go has become the lingua franca of cloud infrastructure, and now Microsoft (alongside Google) is building its agent framework in Go — while OpenAI and Anthropic&amp;rsquo;s agent tooling is still Python-first. The Go lens: the agent infrastructure layer is a 10MB static binary with a goroutine per agent; the Python model layer is a 100MB runtime. The boundary is an API, not a language. The Go agent runtime, in ~60 lines of stdlib, is what the two companies are standardizing.&amp;rdquo;
tags: [&amp;ldquo;Go&amp;rdquo;, &amp;ldquo;Community&amp;rdquo;, &amp;ldquo;AI&amp;rdquo;, &amp;ldquo;Cloud&amp;rdquo;, &amp;ldquo;Agents&amp;rdquo;]
draft: false&lt;/p&gt;</description></item><item><title/><link>https://golang4u.com/news/2026-10-15-platform-independent-simd-in-go/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-platform-independent-simd-in-go/</guid><description>&lt;p&gt;title: &amp;ldquo;Platform-Independent SIMD in Go: the write-once vector layer&amp;rdquo;
image: /images/covers/2026-10-15-platform-independent-simd-in-go.svg
date: 2026-10-15T09:00:00+00:00
author: &amp;ldquo;Punit Naran&amp;rdquo;
description: &amp;ldquo;Go 1.27 adds an experimental, fully portable, size-agnostic simd package — a Highway-style &amp;lsquo;write once, run on every arch&amp;rsquo; vector API that emulates on non-SIMD platforms so the code always runs. The Go lens: why size is a &lt;em&gt;runtime&lt;/em&gt; property (16 lanes on AVX-512, 4 on NEON), the Load/Store/MulAdd core, the ToArch()/FromArch escape hatch for the one missing op, and an innerProduct you can compile with GOEXPERIMENT=simd.&amp;rdquo;
tags: [&amp;ldquo;Go&amp;rdquo;, &amp;ldquo;Community&amp;rdquo;, &amp;ldquo;SIMD&amp;rdquo;, &amp;ldquo;Performance&amp;rdquo;, &amp;ldquo;1.27&amp;rdquo;]
draft: false&lt;/p&gt;</description></item><item><title/><link>https://golang4u.com/news/2026-10-15-vendor-hooks-in-the-android-kernel/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://golang4u.com/news/2026-10-15-vendor-hooks-in-the-android-kernel/</guid><description>&lt;p&gt;title: &amp;ldquo;Vendor hooks in the Android kernel: the Go way of thinking about a kernel extension point&amp;rdquo;
image: /images/covers/2026-10-15-vendor-hooks-in-the-android-kernel.svg
date: 2026-10-15T09:00:00+00:00
author: &amp;ldquo;Punit Naran&amp;rdquo;
description: &amp;ldquo;At the 2026 Linux Plumbers Conference, Todd Kjos (Google) ran the Android-track session on vendor hooks — the mechanism that lets phone vendors keep out-of-tree kernel changes close to mainline without a full fork. The Go lens: a vendor hook is a one-branch indirection, a named call point, and a small registration function. That is an &lt;em&gt;idiom&lt;/em&gt; — and Go has a perfectly portable 40-line model of it (a function-pointer table, a registration, a call-site check) you can compile, run, and extend on any platform. The kernel is a fork, the Go model is a function pointer.&amp;rdquo;
tags: [&amp;ldquo;Go&amp;rdquo;, &amp;ldquo;Community&amp;rdquo;, &amp;ldquo;Linux&amp;rdquo;, &amp;ldquo;Android&amp;rdquo;, &amp;ldquo;Kernel&amp;rdquo;, &amp;ldquo;LPC&amp;rdquo;]
draft: false&lt;/p&gt;</description></item></channel></rss>