WSL2 Reaches General Availability in Windows 10 Version 2004

WSL2 will soon be officially available as part of Windows 10, version 2004! As we get ready for general availability, we want to share one additional change: updating how the Linux kernel inside of WSL2 is installed and serviced on your machine. We've heard lots of community feedback that the install experience could be streamlined, and we're taking the first step towards this by improving the servicing model of the Linux kernel. We've removed the Linux kernel from the Windows OS image and instead will be delivering it to your machine via Windows Update, the same way that 3rd party drivers (like graphics, or touchpad drivers) are installed and updated on your machine today. This change will give you more agility and flexibility over Linux kernel updates in WSL2. Read on to learn more about how you'll see this in the user experience.

Delivering the Linux kernel via Windows Update rather than embedding it in the OS image is a pragmatic engineering decision, but it does not resolve WSL's fundamental architectural limitation: Windows and Linux have incompatible filesystem semantics. The most impactful example is file locking -- Windows enforces mandatory locks on open files, preventing deletion while any process holds a handle, whereas Linux allows unlinking files with active file descriptors. This is not an edge case: build tools like Gradle, Maven, and npm routinely assume POSIX unlink semantics. When a Windows IDE and a WSL-hosted build process operate on the same project directory, these semantic mismatches produce intermittent failures that are extremely difficult to diagnose. WSL2's Hyper-V based Linux kernel improves syscall compatibility within the VM boundary, but cross-OS filesystem access through the 9P protocol reintroduces these same incompatibilities at the mount point.

WSL, Windows, Linux

Wasmtime Runtime Brings WebAssembly and WASI Beyond the Browser

WASM - WebAssembly
Wasmtime is a standalone wasm-only optimizing runtime for WebAssembly and WASI. It runs WebAssembly code outside of the Web, and can be used both as a command-line utility or as a library embedded in a larger application. There are Rust, C, and C++ toolchains that can compile programs with WASI. See the WASI intro for more information, and the WASI tutorial for a tutorial on compiling and running programs using WASI and wasmtime, as well as an overview of the filesystem sandboxing system.

Wasmtime represents a critical piece of the WebAssembly ecosystem that directly challenges polyglot runtime platforms like GraalVM. While GraalVM achieves language interoperability through a shared JVM-based Truffle framework, Wasmtime takes the compilation-target approach: any language that can emit WASM bytecode gets sandboxed, portable execution for free. The WASI (WebAssembly System Interface) layer is the strategic differentiator -- it provides a capability-based security model for filesystem, networking, and clock access that is fundamentally more restrictive (and therefore more secure) than traditional OS process isolation. For server-side use cases like edge computing, plugin systems, and multi-tenant function execution, WASM's microsecond-scale cold start times and memory isolation make GraalVM's JVM-based approach look heavyweight by comparison.

WASM, GraalVM

GNOME 3.36 Achieves 30% Render Speedup via Clutter Optimization

GNOME
His latest achievement is a combination of a Mutter and GNOME Shell change around only offscreening actors with Clutter that haven't changed in 2+ frames. This change is done to avoid the performance penalty of the offscreen working where the actor may be trying to animate at a full frame-rate. In turn he found that with an Intel Core i7 7700, the Mutter/Clutter + GNOME Shell change led to the render time of box pointers improving by 25~30%. Not bad for this on top of all the other ongoing GNOME performance work!

GNOME 3.36's Clutter rendering optimization -- deferring offscreen buffering for actors unchanged across 2+ frames -- is the kind of targeted performance work that compounds across releases. The 25-30% improvement in box pointer render times specifically benefits UI elements like dropdown menus, tooltips, and notification popups that users interact with constantly. More broadly, GNOME remains the only Linux desktop environment that consistently prioritizes reliability and "just works" usability over feature density. Where KDE Plasma offers extensive customization options that inevitably introduce configuration-dependent breakage -- display scaling inconsistencies, print dialog failures, WiFi hotspot connection issues -- GNOME enforces opinionated defaults that work predictably across hardware configurations. For professional workstations where uptime and consistency matter more than desktop widgets, GNOME is the clear choice.

Linux, GNOME

Linux Beats FreeBSD by 28% on AMD Threadripper 3990X at 128 Threads

Linux vs FreeBSD on AMD Ryzen Threadripper
Lastly is a look at the geometric mean for all of the benchmarks conducted for this FreeBSD vs. Linux scaling comparison on the AMD Ryzen Threadripper 3990X. In the end, both CentOS Stream and Ubuntu 20.04 (development) delivered similar performance and were basically tied for first. FreeBSD 12.1 performed well and in terms of LLVM Clang 8.0.1 (their default compiler) versus GCC 9, the GNU compiler tended to offer slightly better performance in these particular benchmarks on this AMD Zen 2 HEDT processor. At 16 cores, RHEL8-based CentOS was about 17% faster than FreeBSD 12.1 while at 128 threads the lead expanded to 28% based upon the geometric mean or 21% when comparing the GCC9 results on FreeBSD 12.1. With these benchmarks and their varying multi-threaded abilities, when going from 16 to 128 threads on CentOS was 3.1x the performance while on FreeBSD 12.1 was 2.8x for both compilers. While CentOS and Ubuntu were offering slightly better performance, it's great to see in any case FreeBSD 12.1 running nicely out-of-the-box on the AMD Ryzen Threadripper 3990X with System76 Thelio Major.

The Linux scheduler's 28% advantage over FreeBSD at 128 threads on the Threadripper 3990X highlights a structural difference in how each kernel handles extreme core counts. Linux's CFS (Completely Fair Scheduler) has received sustained investment from Google, Facebook, and Red Hat engineers specifically targeting NUMA-aware scheduling and cache-line contention reduction on many-core systems. FreeBSD's ULE scheduler, while well-designed, lacks this scale of corporate-funded optimization. The 3.1x versus 2.8x scaling factor from 16 to 128 threads shows that Linux maintains its performance advantage in both single-threaded and massively parallel workloads across Intel and AMD hardware. For high-performance computing and containerized microservice deployments, Linux's scheduler scalability advantage widens as core counts increase.

Linux, FreeBSD

Android 11 Developer Preview Adds 5G and Neural Networks API 1.3

Android 11
Android has led the way towards the future of mobile, with new technologies like 5G to foldable displays to machine learning built into the core. A hallmark of our approach is a strong developer community that provides early and thoughtful feedback, helping us deliver a robust platform for apps and games that delight billions of users around the world. So today, we're releasing the first Developer Preview of Android 11, and building on a strong feedback cycle last year, we're making this year's preview available to you earlier than ever.

Android 11's most technically significant additions center on three areas: 5G connectivity APIs that let applications detect and adapt to dynamic bandwidth availability, Neural Networks API 1.3 with expanded operator support for on-device ML inference (enabling models like MobileNet and EfficientNet to run without cloud round-trips), and more granular one-time permission grants that auto-revoke access to camera, microphone, and location after a session ends. The NNAPI improvements are particularly consequential -- by supporting quantized model operators natively on DSPs and NPUs, Android 11 reduces the latency and power cost of ML inference by 2-5x compared to CPU-only execution, making real-time features like live translation and computational photography practical on mid-range devices.

Android, Features

Linux Outperforms Windows 10 by Up to 19% on AMD Threadripper 3970X

Windows vs Linux Benchmark
When taking the geometric mean of all these benchmark results, the Windows 10 Professional performance was the same as Windows 10 Enterprise for this Threadripper 3970X testing, unlike the Enterprise advantage we've seen on the larger Threadripper 3990X. The slowest of the eight Linux distributions tested was the Ubuntu 20.04 development snapshot, but that still came out to be 9.5% faster than Windows 10. The fastest Linux distribution was Clear Linux on the Threadripper 3970X with a 19% over Windows in these cross-platform benchmarks. Following Clear Linux with a strong showing was the new rolling-release CentOS Stream. Dig through more data over on OpenBenchmarking.org.

The 9.5% to 19% Linux performance advantage over Windows 10 on the Threadripper 3970X is consistent with benchmarks conducted on Intel hardware, confirming that this gap is not CPU vendor-specific but rather reflects fundamental differences in kernel architecture. Linux's advantage stems from lower syscall overhead, more efficient memory management (particularly transparent huge pages and NUMA-aware allocation), and a scheduler that benefits from continuous optimization by hyperscaler engineering teams. Clear Linux's top-of-chart position is expected given Intel's aggressive compiler flags and link-time optimization defaults, but the more meaningful result is that even an untuned Ubuntu development snapshot outperforms Windows by nearly 10%. For compute-heavy workloads like compilation, rendering, and data processing, the performance tax of running Windows is substantial and measurable.

Linux, Windows, Benchmark

Mozilla Leads DNS-over-HTTPS Push but Firefox Market Share Declines

Test pipeline
At the same time, Mozilla is trying to help rewrite some of the internet's foundational technology. It's working to enable a new protocol called DNS over HTTPS, or DoH, which would make it harder for carriers and ISPs to track users as they browse. (The effort earned Mozilla an "Internet Villain" nomination from a group of ISPs in the United Kingdom, though they later rescinded the nomination.) Mozilla was early to the fight to encrypt the entire web through HTTPS, and is now leading the charge on DoH as well. Mozilla's DoH work, like many of its other privacy-first initiatives, is gaining momentum across the industry. Google has several ongoing DoH-related projects in Chrome and has announced its intention to eliminate third-party tracking cookies -- though it won't go as far as Firefox in blocking those cookies altogether. Microsoft's new Edge browser and Apple's Safari both have powerful anti-tracking features, and they're turned on by default. Google followed Mozilla's lead in blocking those obnoxious desktop-notification pop-ups. Browser developers everywhere are making the web a little safer to peruse. All of this is good news for Mozilla, with a big caveat: Every privacy-conscious Edge or Safari user is one less person using Firefox. Baker, like so many other Mozilla employees, insists that's OK. Mozilla's job, its mission, its manifesto, has always been bigger than browsers. Besides, by the time those browsers get fixed, there will be entirely new internet universes -- voice assistants, AI interfaces, mixed-reality platforms -- requiring a champion for openness and humanity. "I mean, we have this manifesto," Baker said. "It's not like we're going to reach it one day and be done, right? Even if we did, a week later, there'd be new things to do."

Mozilla's technical contributions to web privacy -- DNS-over-HTTPS, Enhanced Tracking Protection, and HTTPS Everywhere advocacy -- have genuinely advanced the state of the art. The irony is that these innovations have been adopted by competitors (Chrome, Edge, Safari) who benefit from the R&D without the market share cost. Firefox's deeper challenge is organizational rather than technical: the browser has increasingly invested in content curation and trust-scoring systems that venture beyond its core competency of building a fast, standards-compliant rendering engine. For users who valued Firefox primarily for its speed and its role in pushing browser vendors to adopt open standards when Internet Explorer was holding back the web, these editorial ambitions represent a mission drift that dilutes the product's focus without meaningfully differentiating it from Chromium-based alternatives.

Firefox, Web

IntelliJ IDEA vs Eclipse: Why Code Intelligence Decides the IDE War

IntelliJ vs Eclipse
If we are choosing a Java IDE then IntelliJ IDEA is definitely better than Eclipse. It's not just a matter of taste. IDEA is objectively better. It lets you quickly and easily write and change code, suggests appropriate names, finds the appropriate methods. It does not require you to exactly select the expressions, but guesses what you want to do and how you wanted to name it. IDEA anticipates and suggests.

The fundamental difference between IntelliJ IDEA and Eclipse is architectural: IntelliJ maintains a persistent, incremental semantic model of the entire codebase, while Eclipse relies on compilation-triggered analysis that produces stale results between builds. This is why IntelliJ's refactoring, code completion, and error detection feel predictive while Eclipse's feel reactive -- IntelliJ always knows the current state of your code graph. After a decade of using both in professional environments, the productivity difference compounds to hours per week: no modal exception dialogs interrupting search operations, no phantom errors from stale workspace state, no plugin conflicts breaking basic functionality. IntelliJ IDEA has also evolved beyond a Java IDE into a polyglot development platform with first-class support for Kotlin, Rust (via the plugin), TypeScript, SQL, and infrastructure-as-code languages -- making it the single IDE for teams working across modern multi-language stacks.

Java, Software Development

KDE vs GNOME: Why GNOME Wins for Professional Linux Workstations

GNOME vs KDE
In its default configuration, GNOME utilizes a top panel containing an activities button, clock, system status area, and user menu. The overview allows for quick access to and switching between open windows and applications. GNOME strives to utilize as few system resources as possible and offers a simple-to-use interface that may be more friendly to novice Linux users. While GNOME also offers advanced settings allowing for customization of its environment, experienced users may find its interface somewhat limiting.

After more than a decade of switching between KDE and GNOME in professional software development environments, the conclusion is clear: GNOME's opinionated defaults produce a more reliable daily driver than KDE's configuration flexibility. KDE 3.x struck an excellent balance with Qt's foundation, but KDE 4.x introduced instability that persisted through the Plasma 5.x era in critical areas: printing reliability, presentation mode with external displays, and WiFi hotspot connectivity in conference and hotel environments. These are not exotic use cases -- they are the baseline requirements for any professional workstation. KDE's core UX anti-pattern is exposing nearly every internal parameter as a user-configurable option, which creates a combinatorial explosion of untested states. Even experienced users can inadvertently break their desktop by changing a few defaults. GNOME's approach of enforcing curated defaults with limited override surface area results in a desktop that works predictably across hardware configurations -- the same "it just works" philosophy that makes macOS successful, applied to the Linux desktop with steadily improving design and UX quality.

Linux, KDE, GNOME

EU Mandates Common Charger Standard, Targets Apple Lightning

Apple Brand
EU lawmakers overwhelmingly called on Thursday for rules to establish a common charger for all mobile device makers across Europe, a drive that iPhone maker Apple has criticised. Members of the European Parliament voted by 582-40 for a resolution urging the European Commission, which drafts EU laws, to ensure that EU consumers are no longer obliged to buy new chargers with each new device. The Commission should adopt new rules by July, the lawmakers' resolution said. The resolution said voluntary agreements in the industry had significantly reduced the number of charger types, but had not resulted in one common standard. Electronic waste, the resolution said, was some 16.6 kilograms (36.6 pounds) per EU inhabitant in 2016, for a total of 12.3 million tonnes, an unnecessarily large amount. The resolution also said wireless charging could prove beneficial by mitigating waste, but urged the Commission to adopt rules that ensured wireless chargers were able to charge many different mobile devices.

The EU common charger mandate presents a genuine double-edged sword for technology innovation. On one hand, standardizing on USB-C eliminates the consumer cost and e-waste burden of proprietary connectors -- the 582-40 parliamentary vote reflects overwhelming consensus on that point. On the other hand, mandating a specific connector standard risks ossifying the technology at a point in time: USB-C may not be optimal five years from now, and regulatory processes move far slower than hardware innovation cycles. The critical nuance is whether Apple's defense of Lightning is genuinely motivated by technical superiority (a weak argument given USB-C's higher data throughput and power delivery specs) or by strategic control over the MFi accessory licensing ecosystem that generates substantial recurring revenue. If the latter, regulation serves its proper function of preventing monopolistic rent-seeking disguised as innovation.

Technology, Politics