标签：#X

Website performance is no longer just a nice-to-have feature—it's a critical factor for user experience, SEO, and business success. Even a one-second delay in page load time can reduce conversions and increase bounce rates. Whether you're building a portfolio, SaaS application, eCommerce platform, or business website, these optimization techniques can make a significant difference. Optimize Images Images are often the largest assets on a webpage. Use modern formats like AVIF or WebP, compress images, and serve responsive image sizes to reduce bandwidth usage. Self-Host Fonts Third-party font requests add latency. Self-hosting fonts, preloading critical font files, and serving only the required character subsets can dramatically improve loading performance. Remove Unused CSS & JavaScript Shipping unnecessary code increases download size and execution time. Tree shaking, code splitting, and removing unused styles help keep your bundle lean. Enable Caching Configure long-term browser caching for static assets and use hashed filenames for cache busting. This allows returning visitors to load your website much faster. Use Lazy Loading Images, videos, and iframes that aren't immediately visible should load only when needed. Native lazy loading is supported by modern browsers and is easy to implement. Optimize Core Web Vitals Google's Core Web Vitals measure how users experience your website. Focus on: Largest Contentful Paint (LCP) Interaction to Next Paint (INP) Cumulative Layout Shift (CLS) Improving these metrics benefits both SEO and user satisfaction. Minify Assets Minify HTML, CSS, and JavaScript files before deployment. Smaller files transfer faster and improve overall performance. Use a CDN Serving assets from edge locations around the world reduces latency and improves loading times for global visitors. Prioritize Accessibility Accessible websites provide a better experience for everyone and often align with SEO best practices. Use semantic HTML, descriptive labe

Week 16 shipped the AI background remover: Rust-native, ort + ISNet + libvips, no Python. That was the feature. Week 17 was not about writing more of it. It was about the boring, high-leverage part that most side projects skip: turning one working feature into pages that can actually rank, and wiring those pages together so search engines can find them. No new engine code this week. Just leverage on what already existed. Here is what that actually looked like. The problem: a hub with nothing pointing at it The background remover lives at /remove-background . That is the hub. The plan was classic hub-and-spoke: one general tool page, then use-case spokes that each target a specific intent (removing a signature background, prepping an Amazon product photo, and so on). I built two spokes this week. But halfway through, I looked at how internal links actually worked on the site and found the real problem: nothing linked from the hub to the spokes. The spokes linked back to the hub in their body text, but the hub had no idea they existed. Neither did the ~180 converter pages. Tool links on the site were hardcoded in a frontend constant, roughly: export const IMAGE_TOOLS = [ { label : " Compress JPG " , href : " /compress/jpg " , tool : " compress " }, { label : " Resize Image " , href : " /resize-image " , tool : " resize " }, { label : " Crop Image " , href : " /crop-image " , tool : " crop " }, { label : " Images to PDF " , href : " /images-to-pdf " , tool : " convert " }, ] as const ; That list covered the converter tools. It did not include the background remover or its spokes at all. So the new pages were orphans: reachable only through the sitemap, with no internal links carrying any signal to them. For a domain that is still young and still earning Google's trust, orphan pages get discovered slowly and rank even slower. The fix: one constant as the source of truth Instead of hardcoding links in three different places, I made a single constant describe the whole cl

Rockstar's long-awaited Grand Theft Auto VI is launching November 19th, 2026 for PlayStation 5 and Xbox Series S/X consoles. The game will be available digitally at launch, with physical cases containing codes, not discs, so your options for preordering are straightforward. The preloading phase for the game begins on November 12th, which should give you […]

Installing Ubuntu Server 24.04 LTS in an offline mode (no LAN cable or WiFi connected) leaves you with an unmanaged network interface which requires manual configuration post-install. The interface enp4s0 is unmanaged by systemd-networkd and also the service itself is disabled. NOTE: This guide applies to Ubuntu server. On Ubuntu desktop you have the NetworkManager service installed and the steps would be different. To fix the problem follow the steps below: Create a Brand New Netplan Configuration File Look into /etc/netplan . You will probably see the dir is empty. This means the installer completely gave up on configuring the network and didn't create any profiles. Create a new file, e.g. vim /etc/netplan/01-netcfg.yaml . Paste the following configuration network : version : 2 renderer : networkd ethernets : enp4s0 : dhcp4 : true This makes the interface managed by networkd instead of the desktop NetworkManager. Fix permissions The new file needs to be readable only by root so fix it's permissions chmod 600 /etc/netplan/01-netcfg.yaml . Enable network service Enable the network service and make it start on boot: systemctl enable systemd-networkd systemctl start systemd-networkd Run netplan Finally, we need to tell Ubuntu to use our new configuration and activate the network netplan generate netplan apply Network should be up! ip a

When you delete files in your virtualbox VM in order to free up space on the host filesystem, this space is not automatically reclaimed. In order for the host system to see the changes you need to rewrite the free space with zeroes. Follow the below steps to perform this operation: Install zerofree package. It is needed to rewrite the free space with zeroes. Mount the filesystem as "readonly". This is needed for the tool to be able to perform it's task. If you're working with the "/", easiest way to mount it as readonly is to edit the kernel parameters. Edit /etc/default/grub . Find the GRUB_CMDLINE_LINUX_DEFAULT line. Add init=/bin/bash to it reboot Run zerofree -v /dev/sdX . This could run for some time, depending on the size of your disk. After it's done, run exec init to finish booting up. Shutdown the VM in order to be able to run the next command which requires a lock on the VDI volume. On the Windows host run VBoxManage.exe modifymedium "path\to\disk.vdi" --compact

A few years ago, writing code was the difficult part. Today, AI can generate an API, build a React component, explain Dynamic Programming, fix bugs, and even suggest architecture—all within seconds. So here's a better question. If AI can generate code, what exactly are companies hiring humans for? The answer isn't typing speed. It isn't memorizing syntax. And it certainly isn't copying solutions faster than someone else. The value of a developer is shifting. And learning platforms need to shift with it. The Developer Role Is Changing Modern software engineering is becoming less about writing every line manually and more about making good engineering decisions. Can you understand a problem before solving it? Can you identify why one solution is better than another? Can you improve AI-generated code instead of accepting it blindly? Can you build something that is maintainable, scalable, and useful? These questions matter more today than they did five years ago. AI Reduced the Cost of Writing Code One of AI's biggest achievements is reducing repetitive work. That's a good thing. Developers spend less time writing boilerplate and more time focusing on higher-level thinking. But this creates a new challenge. When everyone has access to the same AI tools, writing code becomes less of a differentiator. Thinking becomes the differentiator. Learning Needs to Evolve Too Many learning experiences still revolve around one objective: Solve another problem. Complete another lesson. Earn another badge. Those activities still matter. But in an AI-first world, they aren't enough on their own. Learners also need opportunities to connect concepts, apply knowledge, build projects, and understand why solutions work—not just that they work. Where PyNyx Takes a Different Direction PyNyx is being built around a broader learning journey rather than a collection of isolated activities. Instead of separating learning into unrelated pieces, the platform connects multiple stages of growth. Stru

Hello, everyone. When listening to a conversation, we naturally keep track of who is speaking. A program has a harder job: beyond finding speech, it must also determine where one speaker gives way to another. Today, I will use an ONNX version of Pyannote Segmentation 3.0 to detect speaker changes in a two-person conversation and split the recording into one WAV file per utterance. What I Tested This lab uses FFmpeg to decode a roughly 14-second conversation into a 16 kHz mono waveform. It then combines the Pyannote segmentation model with simple post-processing to produce contiguous speaker segments. I wanted to verify: Whether six alternating utterances can be separated into six segments Whether the detected speaker indexes remain consistent throughout the recording Whether ONNX Runtime can process the audio faster than real time using only its CPU execution provider Whether every segment can be saved as a separate WAV file The complete code and reproducible environment are available in the pyannote-scd lab in kiarina/labs . This test performs segmentation using the model's speaker indexes. It does not compare speaker embeddings or run clustering, so it is not a complete speaker diarization pipeline that identifies the same person throughout a long recording. Reproducing the Lab You will need: mise uv FFmpeg curl The following commands fetch only this lab, download the shared test audio, and run it: git clone --depth 1 --filter = blob:none --sparse \ https://github.com/kiarina/labs.git cd labs git sparse-checkout set .gitignore .mise/tasks Makefile mise.toml \ 2026/07/04/pyannote-scd make download-test-assets mise -C 2026/07/04/pyannote-scd run On the first run, the task downloads the full-precision onnx/model.onnx file from onnx-community/pyannote-segmentation-3.0 on Hugging Face. uv then prepares the Python dependencies and runs the detector. How Speaker Segments Are Detected The input is this shared test asset: assets/mp3/conversation_2speaker_14s_16k.mp3 The re

I had just aced my mentor’s Docker exam, so of course I thought I understood containers. I had said all the right words: namespaces, cgroups, images, layers, PID 1, Kubernetes Pods. Then I typed my first serious command and Linux reminded me that knowing the nouns is not the same thing as building the thing. $ sudo unshare -p 1 test unshare: failed to execute 1: No such file or directory That was the opening scene. I had not even built anything yet. I had typed the flags wrong and accidentally asked unshare to execute a program called 1 . This was going to be less “implement Docker” and more “let the kernel correct my confidence, one error at a time.” v1: namespaces, or the first time PID 1 lied to me The first version was supposed to be easy: run a process in a new PID namespace and prove it sees itself as PID 1. So I ran the command the way I thought it worked: $ sudo unshare --pid bash # echo $$ 25184 That was not PID 1. That was just embarrassing. The rule I had missed is simple: PID namespaces apply to children. The process that calls unshare --pid does not magically become PID 1. You need to fork. The first child born into the new namespace becomes PID 1. So the working version was: $ sudo unshare --pid --fork bash # echo $$ 1 That one line changed the tone. I was inside a different process universe. The shell thought it was process 1. Signals felt different. Orphans came home to it. Then I ran ps , and got humbled again. # ps -o pid,ppid,comm PID PPID COMMAND 25310 25304 bash 25344 25310 ps That made no sense at first. I was PID 1, but ps was showing host-looking PIDs. The next reveal: ps does not ask the kernel some pure “what processes exist?” question. It reads files. If /proc still points at the host procfs, your tools will tell you the host story. So I remounted /proc from inside the namespace: # mount -t proc proc /proc # ps -o pid,ppid,comm PID PPID COMMAND 1 0 bash 7 1 ps That was when it clicked. The namespace did not become real to my eyes until /pr

I broke my own project on purpose, twice, before it worked. Here's the story. Round one: the debate club My first idea for this hackathon sounded great in my head. Six AI agents, each with a "role" — security, architecture, performance, whatever — and they'd debate each other across multiple rounds before agreeing on a final answer. Like a mini panel of experts arguing it out. I built it. I ran it against some vulnerable test code. It came back with 127 findings. I got excited for about four minutes. Then I actually read them. Maybe three were real. The other 124 were the agents politely agreeing with each other about problems that didn't exist, or restating the same bug five different ways because five different agents happened to notice it. Precision was somewhere around 2%. Worse than a single model working alone. That stung a little, not going to lie. I'd spent days on the debate logic. Round two: quieter, and better So I ripped it apart. No more debate rounds. No more six agents shouting over each other. I went down to four, gave each one exactly one job, and — this is the part that actually fixed things — made them depend on each other in order instead of all firing at once. One agent maps out the code first. Two others use that map to look at security and quality separately. A last one compares what they found, throws out duplicates, and — importantly — actually checks the line numbers against the real file instead of trusting the AI's word for it. Same test file. This time: real vulnerabilities, correctly flagged, nothing made up. Point it at clean code afterward and it correctly said nothing was wrong, which honestly felt like a bigger win than finding the bugs did. The annoying lesson I wanted this project to feel impressive. More agents, more debate, more "look how sophisticated this is." What actually worked was the boring answer: fewer agents, clear roles, one checking the other's work instead of everyone talking at once. I named the final version Synod

I like neofetch / fastfetch , but I wanted one with a genuinely empty dependency graph — nothing pulled from crates.io. So I built purefetch : a small system-info fetcher written entirely in Rust using only std plus raw Linux syscalls. Disclosure up front: purefetch was built largely with AI assistance (Claude Code). I directed the design, and every change was reviewed and tested — including running it on four architectures under QEMU — but most of the code is AI-generated. I'd rather be honest about that than pretend otherwise. _,met$$$$$gg. ooonea@unicorn ,g$$$$$$$$$$$$$$$P. ────────────── ,g$$P" """Y$$.". OS Debian GNU/Linux 13.5 (trixie) x86_64 ,$$P' `$$$. Host ThinkPad P53 (20QQS0JD01) ',$$P ,ggs. `$$b: Kernel 6.12.94+deb13-amd64 `d$$' ,$P"' . $$$ Uptime 6 days, 15 hours, 30 mins $$P d$' , $$P Packages 2477 (dpkg), 1 (flatpak) $$: $$. - ,d$$' Shell zsh 5.9 $$; Y$b._ _,d$P' Display 1920x1080 (eDP-1) Y$$. `.`"Y$$$$P"' DE GNOME 48.7 `$$b "-.__ WM Mutter (Wayland) `Y$$ Terminal kitty 0.41.1 `Y$$. CPU Intel(R) Core(TM) i7-9850H @ 4.60 GHz `$$b. GPU Quadro RTX 3000 `Y$$b. Memory 15.28 GiB / 62.61 GiB (24%) `"Y$b._ Swap 0 B / 8.00 GiB (0%) `""" Disk (/) 8.52 GiB / 489.57 GiB (2%) Locale en_US.UTF-8 Battery 76% (Not charging) Zero dependencies, really No libc crate, no sysinfo , no nix , no color crate — nothing from crates.io. Almost everything is just reading and parsing /proc and /sys . The result is a single ~484 KiB binary that builds offline. The only things std can't do are statfs (disk usage) and ioctl (terminal size / tty check). Instead of pulling in a binding crate, those are issued as raw Linux syscalls via core::arch::asm! : #[cfg(target_arch = "x86_64" )] unsafe fn syscall3 ( n : usize , a1 : usize , a2 : usize , a3 : usize ) -> isize { let ret : isize ; core :: arch :: asm! ( "syscall" , inlateout ( "rax" ) n as isize => ret , in ( "rdi" ) a1 , in ( "rsi" ) a2 , in ( "rdx" ) a3 , out ( "rcx" ) _ , out ( "r11" ) _ , options ( nostack ), ); ret } Four arch

Every meditation app I have tried wants something from me. Headspace wants me to maintain a streak. Calm wants me to listen to a Daily Jay. Insight Timer wants me to join a group. One after another, apps designed to reduce my stress started creating new forms of it. The Feature Trap Here is what happened to meditation apps between 2015 and 2026: 2015: "Just meditate 10 minutes a day." 2018: "Track your streak! You do not want to break it, do you?" 2021: "Compare your stats with friends. See who meditated more this week." 2024: "AI-generated personalized guided meditation based on your emotional state, delivered at the optimal time based on your circadian rhythm." Wait — was not the whole point to stop optimizing everything? Subtraction as a Feature I switched to OneZen last month. Here is what I noticed: No onboarding. Open the app. Breathe. Close the app. That is the entire user flow. No streaks. I missed three days last week and the app did not shame me. It did not even notice. It just opened to the same calm screen, waiting, as if three days was the same as three hours. No gamification. No XP points. No badges. No "you are in the top 14% of meditators this month." Because meditation is not a competition you can win. What Subtraction Feels Like The first week was uncomfortable. I kept checking if I had "done it right." There was nothing to check. No dashboard. No stats. Just me and my breath. By week two, something shifted. Meditation stopped being a task on my to-do list and started being... just breathing. I was not practicing to maintain a number. I was practicing because it felt good. This is what minimalism actually means. Not fewer pixels. Less cognitive load. Less obligation disguised as features. The Bigger Idea OneZen's philosophy applies far beyond meditation apps: The best productivity tool is the one with the fewest notifications. The best social network is the one that respects when you leave. The best habit tracker does not exist — because the ha

A gente sempre ouve falar que o sistema operacional impede que um processo veja a memória do outro ou que o programa fale diretamente com o hardware, mas normalmente não explicam o "como". Eu sempre achei isso meio mágico até que eu resolvi ir atrás da resposta, e é bem interessante. Vou me basear na arquitetura x86, mas é provável que outras arquiteturas sejam parecidas. O problema: a CPU Pra CPU não existe processo, kernel, sistema operacional. Existe só endereços de memória de onde ela lê a próxima instrução e executa. Se a CPU pode falar direto com a RAM, SSD, teclado, mouse, tela... O que me impede de escrever um programa pra ler suas senhas e tokens direto da RAM? Ou de ler arquivos e alterar arquivos sensíveis direto no SSD? Por outro lado, se o kernel fiscalizasse cada instrução que da CPU antes dela executar, isso seria extremamente lento... Outro problema: os interrupts Se a CPU só executasse sequencialmente, seu sistema poderia executar várias coisas e esquecer de checar se uma tecla foi apertada, se o mouse mexeu, etc... Então certos eventos interrompem o que quer que a CPU esteja fazendo para serem tratados assim que possível. Alguns exemplos de interrupt são: Teclas do teclado pressionadas ou soltas Botões e movimento do mouse Timers Operações de disco assíncronas Pacotes de rede recebidos/transmitidos Uma solução: rings Os processadores da arquitetura x86 tem o esquema de rings. Pense em rings como grau de limitação. Ring 0 significa limitação zero, ou seja, acesso a todas as instruções da CPU e consequentemente acesso total ao hardware e memória. O kernel roda em ring 0, ou kernel mode. O kernel assim que é carregado configura todos os interrupts handlers da CPU para executar o handler apropriado do kernel, em kernel mode, claro. Em ring 3 a CPU fica limitada e não pode fazer instruções consideradas privilegiadas. E obviamente em ring 3 a CPU não consegue se colocar em ring 0 sozinha, pois dessa forma qualquer programa conseguiria se pôr em ring 0. O

House of the Dragon, Adventure Time: Side Quests, and The Man Will Burn are just a few of the TV shows you should be watching right now.

Seven days of flow. Fixed a peer identity binding issue in py-libp2p, automated my Neovim lockfile merges, and added a massive batch of notes on xv6 and Category Theory. 10 commits and a solid PR in the works. TL;DR I managed to hit a perfect seven-day streak this week, balancing some deep-dive p2p networking work with necessary maintenance on my local environment. The highlight was opening a PR in py-libp2p to tighten up how PeerRecords are handled in the Kademlia DHT. On the side, I spent time automating the annoying parts of my Neovim config and dumping a fresh batch of notes into my knowledge base. 10 commits, 204 lines added, and a much cleaner workflow to show for it. What I Built Neovim Configuration & CI I’m a firm believer that your editor should work for you, not the other way around. My nvim repo saw a lot of action this week—9 commits in total—but most of it was under-the-hood maintenance. I’ve been leaning on Lua to keep things snappy, and this week was about ensuring my plugin ecosystem doesn't rot. I pushed several updates to keep plugins at their latest versions, but the real "quality of life" improvement was adding a chore to auto-resolve lazy-lock.json merge conflicts. If you’ve ever worked on your Neovim config across multiple machines, you know the headache of the lockfile drifting. I set up a flow to prioritize incoming changes, which saves me from manually triaging JSON diffs every time I pull from dev . It’s a small tweak, but it removes a recurring friction point in my daily flow. I also spent time cleaning up the root of the config, with about 37 additions and 33 deletions—refactoring is a constant process when you live in your terminal. The Knowledge Base I also put some serious time into main-notes . I’m currently going deep on a few different subjects, and I use this repo as my "second brain." I added 167 lines of new material across 13 files, covering a pretty diverse range of topics: xv6 (the re-implementation of Unix V6), Category Theo

TL;DR: Most extensions block Google's AI Overviews by hiding the panel with a content script after it renders — fragile, flickery, and always a step behind Google's markup changes. A better approach: force udm=14 at the network layer with declarativeNetRequest , so the AI Overview never loads. The content script becomes a backstop, not the main mechanism. One Chrome API mystery — AI Mode being invisible to four different extension APIs — shows why the DOM was never the right layer. Google puts an AI Overview at the top of most search results now, and a lot of people would rather it didn't. So there's a whole shelf of Chrome extensions that remove it. Almost all of them work the same way, and I think that way is a mistake. The obvious approach, and why it's a trap The default move is DOM-hiding: inject a content script, wait for the AI Overview panel to render, find it by class name or attribute, and set display: none . It's the first thing anyone reaches for, and it works — until it doesn't. The problems are all baked into the approach. You're reacting after the render, so there's a flash of AI content before your script catches it. You're matching against Google's markup, which is obfuscated and reshuffled constantly, so every layout change is a silent breakage. And you're paying for DOM churn on a page you don't control. You end up in a permanent game of catch-up against a page that changes whenever Google feels like it. The deeper issue is that you're operating one layer too high. The panel is a symptom . By the time it's in the DOM, the work is already done — the server decided to send it, the page rendered it, and now you're scrambling to un-render it. If you can move the decision earlier, none of that scramble has to happen. The thesis: prevent it at the network layer Google Search takes a parameter, udm , that selects which result vertical you get. udm=14 is the plain "Web" results view — the classic list of links, no AI Overview, no AI Mode. It's Google's ow

Cursor hopes to continue offering third-party AI models after it's acquired by SpaceX, testing the relationships between frontier AI labs.

A former software manager claims Wisk rushed software testing ahead of a crucial 2025 flight test.

Plex is pushing customers to newer features and more frequent payments.

What happens when a machine no longer needs to be trained to see something new? That's the quiet question sitting underneath this week's news, buried next to a less invasive brain implant and a handful of robots getting tougher for the real world. Neuralink says it's completed its first "transdural" brain implant, a surgical approach built to reduce trauma during the procedure. As someone who spends a lot of time thinking about how you get sensors close to a human eye without hurting anyone, I find these less-invasive-implant strategies worth watching, because the surgical-risk problem is basically the same one we wrestle with in ophthalmic hardware. Vision is getting less invasive too, in its own way. Roboflow rolled out text-prompt object detection built on SAM3 (Meta's latest segmentation model): you type the class of object you want "forklift," "cracked tile," whatever, and it returns boxes and masks without you collecting a single training image first. That's a real shift. For most of computer vision's history, teaching a model to recognize something new meant labeling hundreds of examples before you could even start; this collapses that step into a sentence. The same week brought several applied builds using the same detect-then-orchestrate pattern: a drone system that patrols for intrusions, a pipeline that inspects transmission lines for damaged cables, and an airport tool that spots foreign debris on the tarmac. The Robot Report's roundup of June's biggest robotics stories leaned heavily on humanoid robots companies going public, new deployments, and production milestones stacking up faster than would have seemed plausible a few years ago. Apptronik unveiled its Apollo 2 humanoid alongside a dedicated data-collection facility built so the robot keeps learning after it's deployed, not just during initial training which quietly answers one of the harder questions in robotics: how do you keep a system improving once it's out of the lab? X Square Robot raised e

Amazon says it now has enough satellites operating in low-Earth orbit to light up its Starlink internet competitor. With last night's launch, Amazon Leo has 396 satellites deployed, which is "enough to support continuous service across initial latitudes," according to Chris Weber, VP heading up business and product for Amazon Leo. That puts the company […]