标签：#Go

The engineering team at Stripe recently described how they automated database incident recovery by modeling their global infrastructure as a graph. Using graph search algorithms together with state machines, the team computes and executes remediation plans automatically. By Renato Losio

Hello everyone! This is the second article about the development of RUSEON-core, a Zero-Copy video streaming server for AI platforms and Edge video infrastructure. In the first article , I talked about the fundamental reason why we decided to create our own server in the first place. I also covered the main problem with most similar solutions — the "thundering herd" — and how we managed to squeeze out 8 Gbps on a single CPU core. By the way, I forgot to mention in that article that besides simple streaming, we also record the streams in fMP4 format. It’s stored locally for N amount of time, and it can fly off to an S3 bucket (depending on how long the clients want to keep the recordings). This article is precisely about a non-obvious (well, at least to me, maybe for someone else it's an everyday thing) problem related to data storage and its specifics across all Operating Systems. So, let's dive in. We rolled out our first release to production (100 cameras), made the clients happy, and started working. About an hour passed, and the alerts started flying. I SSH into the server, open htop, and see there's only 100 MB of free RAM. Uh-oh. I should clarify that the production server had 32 gigs of RAM. The expected behavior was that the CPU is chilling, the network card is chewing through the traffic, RAM usage is around 250-300 MB, and the disks are not heavily loaded. So, when you see numbers like that in htop, you start blaming yourself and your crooked hands that wrote this piece of "garbage". But still, we decided to go to Google, ChatGPT, and the like. Fortunately, the answer was found quickly, and we stopped beating ourselves up. The code was absolutely not the culprit; Linux itself ate the memory. If you've ever written tons of data to a disk, I think you already know what’s going on. There is an "invisible enemy" known as the Page Cache. That was exactly the root of this problem. How does the Page Cache work and what to do with it? When your function that is su

Multi-Repo to Monorepo: How I Automated 6 Go Microservice Releases and Then Made It 15x Faster Amandeep Singh Amandeep Singh Amandeep Singh Follow Aug 7 Multi-Repo to Monorepo: How I Automated 6 Go Microservice Releases and Then Made It 15x Faster # go # devops # automation # monorepo 6 reactions 1 comment 10 min read

Mediax: FFmpeg's Cooler Cousin 🚀 FFmpeg is powerful, but the syntax? A nightmare. So I built Mediax. 36 intuitive verbs. Interactive TUI. One-line install. Convert, compress, trim, crop, rotate, add watermarks... all with simple commands. bash mediax convert input.mov output.mp4 mediax compress large.mp4 small.mp4 --quality medium mediax trim video.mp4 clip.mp4 --start 00:01:30 --duration 10 Just run mediax for the interactive terminal interface. One-line install curl -sSL https://raw.githubusercontent.com/robert-sarah/mediax/main/install.sh | sh Built with Go Single binary — no dependencies Cross-platform: Windows, macOS, Linux Open source (MIT) 36 verbs for everything Conversion: convert, compress, gif Audio: extract-audio, mute, volume, replace-audio Video: trim, crop, resize, rotate, flip, concat, split Effects: speed, reverse, blur, sharpen, fade-in, fade-out Advanced: watermark, subtitle, thumbnail, template, batch Examples Resize for Instagram mediax template video.mp4 insta_video.mp4 --platform instagram Extract audio from 1:30 to 1:40 mediax extract-audio video.mp4 clip.mp3 --start 00:01:30 --duration 10 Detect video issues mediax wtf video.mp4 Why I built it I was tired of copying FFmpeg commands from Stack Overflow and breaking files because I forgot a flag. Mediax makes FFmpeg accessible for beginners and faster for pros. Try it today Repo: github.com/robert-sarah/mediax Contributions welcome! 🙏

When building a nested reply system, most developers worry about deep tree complexity or messy data structures. For Vlox , I took a different approach: keeping things flat, fast, and secure by reusing a single Mongoose schema with smart atomic limits. Here is a deep dive into how I engineered a production-ready, race-condition-safe reply mechanism using MongoDB transactions, strict type sanitization, and automated limits. How It Works 🛠️ User Action: A user clicks the reply icon and submits their reply. The Payload: Vlox's system sends 3 fields via the endpoint /api/v1/reply/comment/post/:id : id : The post ID (passed as a URL parameter). rootCommentId : The ID of the root comment being replied to. reply : The raw text entered by the user. Sanitization & Validation: The incoming reply is instantly converted to a trimmed string. It then passes through two critical validation checks: Existence Check: The reply must exist. (If a malicious actor sends a payload without a body, the string literally evaluates to "undefined" and gets blocked). Length Limit: The reply must be under 201 characters, enforcing the standard comment limit. Atomic Transactions: If the validation checks pass, the system initiates a Mongoose transaction to execute the following steps safely: Permission Check: It verifies if the user has permission to reply by checking the post's status via await schemas.Posts.findOne(hotQueries.find_user_post(id, req.session.userId)); . Creation: If permissions are valid, it creates a new reply. (Fun fact: It reuses the exact same schema as standard comments!) The Reply Schema Structure: The reply object functions just like a normal comment, with two distinct exceptions: It does not contain a repliesCount field. It includes an extra rootId field, which explicitly points to the ID of the root comment being replied to. Concurrency & Caps: To guarantee that a single comment never receives more than 10 replies while simultaneously incrementing the counter, the system r

Google recently changed how it refers and assigns names to hacking groups. TechCrunch spoke with one of the world’s foremost experts on tracking hackers to understand why companies give hackers codenames.

Building a personalization engine for a multi-format content feed, without machine learning, and the testing process that forced us to rebuild it. TL;DR We run a collaborative cultural curation platform (think: user-submitted recommendations for movies, books, games, music, and long-form posts, all mixed into one feed) on a fairly ordinary PHP + MySQL stack. Over about a year we went through two full generations of the feed ranking algorithm. The first version solved the obvious problem (stop being purely chronological) but quietly failed at real personalization. The second version fixed that by rethinking what "user taste" even means, moving scoring out of SQL and into application code, and adding a layer of post-ranking business rules. This post walks through both generations, why the second one had to happen, and how we actually tested and calibrated a feed ranking system without a data science team or an ML pipeline. No exact weights, table names, or formulas below — just the engineering story. The starting problem: one feed, five content shapes Before personalization is even on the table, a multi-format feed has a normalization problem. Movies, books, games, music, and editorial posts live in different tables, with different columns, different publishing cadences, and engagement numbers on completely different scales. "1,000 likes" on a music post and "1,000 likes" on a book review are not the same signal. So the very first architectural decision — before any ranking logic existed — was building a unification layer that maps every content type into a shared shape (type, author, title, cover, category, engagement counters, timestamp) before any scoring happens. Everything downstream depends on that layer being consistent. Generation 1: a weighted ranker living inside a single SQL query The first real version of the algorithm — internally we called it the hybrid model — had a modest goal: get away from a purely chronological feed without building anything resembl

Nice video of the Arctic bobtail squid. As usual, you can also use this squid post to talk about the security stories in the news that I haven’t covered. Blog moderation policy.

Pruebas una estrategia, o un modelo, sobre datos históricos. El backtest da un número bonito. Y luego, en real, no aparece. Casi siempre es una de estas dos ilusiones — y las dos se descartan con muy poco código. Empaqueté las dos correcciones como librería: honest-eval , Python puro, sin dependencias. Salieron de un bot de trading, pero el rigor no tiene nada de específico al trading. Ilusión 1: el modelo vio el futuro Partir los datos con el clásico train_test_split aleatorio es correcto para datos independientes. En una serie temporal es un desastre silencioso: mete muestras de mañana en el conjunto de entrenamiento, y el modelo "predice" en el test cosas que en producción todavía no habrían pasado. La métrica sale inflada, y confías en un edge que no existe. El test honesto es siempre el futuro : el tramo más reciente en el tiempo. from honest_eval import temporal_split train_idx , test_idx = temporal_split ( timestamps , test_frac = 0.20 , embargo = 24 ) X_tr , X_te = X [ train_idx ], X [ test_idx ] Devuelve índices, así lo aplicas a numpy, pandas o listas por igual. El embargo cierra una fuga más sutil: si tu etiqueta mira h pasos adelante, una muestra de entrenamiento a menos de h del corte ya conoce parte del resultado del test. embargo=h descarta ese borde. La métrica baja — pero por fin es la real out-of-sample . Ilusión 2: la variante ganó por suerte Tienes varias variantes y quieres la mejor. Eliges la de mayor media. Error: con pocas muestras, eso premia la varianza, no la ventaja . La variante más ruidosa suele quedar arriba por azar. Dos correcciones, ambas dentro de select_best_variant : Aparear. Mide variante y baseline sobre el mismo ensayo y trabaja con δ = variante − baseline . La varianza común del ensayo se cancela en la resta, y te quedas con la señal. Exigir cota inferior de confianza > 0. Gradúa una variante solo si media − z·SE > 0 : "incluso siendo pesimista dentro del margen de confianza, sigue por encima del baseline". from honest_eval i

Last month I spent more than an hour cutting a release across six Go microservice repos. Tag log, wait for CI. Update sdk's go.mod to point at the new log SHA, push, wait for CI. Repeat for utils. Then do api, cli, and worker in parallel - except I forgot to bump cli's dependency and the build broke at 11pm. That was the last manual release I did. This is the story of automating that entire workflow with Jenkins + Python + GitLab, then realizing the multi-repo architecture was the real problem, and collapsing everything into a Go monorepo that's 15x faster at cutting releases. The full setup runs on my laptop. You can fork it and try it yourself. Table of Contents The Six Modules The Stack Phase 1: Multi-Repo Automation Phase 2: The Monorepo Pivot The Unified CI Pipeline Real Numbers Caveats and Gotchas Try It Yourself The Six Modules The project simulates a real production system with six Go modules that have strict dependency ordering: Module Role Tag Scheme Depends On log Logger (leaf, no deps) v0.x.0 - sdk API client v0.x.0 log utils Shared utilities v0.x.0 log, sdk api/backend Backend APP-x.y.z log, utils cli CLI cli-x.y.z log, sdk worker Background v0.x.0 log, utils The first three modules are sequential - sdk can't tag until log is tagged, utils can't tag until sdk is tagged. The last three are terminal - they can process in parallel once the sequential chain is done. Every module lives on three long-lived branches: develop → release → master . A release means moving code through all three, in all six repos, in the right order. That's the problem. Do it manually and you're juggling 6 repos × 3 branches × dependency ordering. One forgotten go mod tidy and you're debugging at midnight. The Stack Everything runs on a MacBook. No cloud CI, no SaaS - just local tools wired together. MacBook GitLab.com +-------------------+ ngrok tunnel +------------------------+ | Jenkins LTS | <===============> | Webhooks (push / MR) | | (brew service) | | Commit status API | | :

Some of the biggest names on Google's AI team got new jobs this week. In some cases, including for legendary Googler Jeff Dean, those jobs are no longer at Google. Given that Google's models seem to be behind the best of what's coming out of anthropic and OpenAI, is this a sign of Google in […]

The Quest Begins (The "Why") Honestly, I still remember the first time I tried to build an autocomplete widget for a side‑project. I had a list of 200 k product names, a simple filter that ran on every keystroke, and the UI felt like wading through molasses. Each keypress triggered a full scan of the list, and with a few users typing at once the browser would start to lag. I was stuck in a loop that felt like the infamous “boss fight” where you keep hitting the same pattern over and over, hoping for a different outcome. I kept asking myself: There has to be a smarter way. Why am I re‑checking the same prefixes again and again? If ten users type “tea”, why do I walk through the whole dictionary ten separate times? That question turned into a mini‑quest, and the treasure at the end was the trie data structure. The Revelation (The Insight) Look, the magic of a trie isn’t that it’s some exotic tree; it’s that it stores words by their shared prefixes . Imagine you have the words “cat”, “car”, “cart”, and “dog”. In a trie you’d have a root node, then a c branch that splits into a → t (for “cat”) and a → r → t (for “cart”), while “dog” lives on its own d → o → g path. Every common prefix is stored once , and you can walk down the tree following the characters of a query to land exactly at the node that represents all words with that prefix. Why does this give us O(L + K) time for autocomplete, where L is the length of the prefix and K is the number of results? Walking the trie follows the prefix character‑by‑character → O(L). From that node we just need to collect all words in its subtree. If we keep a list of words at each node (or run a DFS), we touch each result once → O(K). No extra work for words that don’t share the prefix. Contrast that with the naive filter approach: O(N × L) where N is the total dictionary size. For a large N, the trie is a game‑changer—it’s like switching from swinging a blunt sword to wielding a lightsaber that cuts through the prefix forest in

The Trump administration has now convinced developers to abandon 12 offshore wind leases. The latest will cost taxpayers $1.2 billion.

I have been building machin for a while — a Go-flavored, type-inferred language that compiles through C to a single native binary. It has grown a lot recently, and I wanted to answer the obvious question honestly: does it beat Rust and Zig at anything? It does, at two things, decisively. But the first thing I found was not a win. It was my own benchmark quietly lying to me, and the number it was lying about was the best one I had. The benchmark was measuring the order I ran things in machin's repo has had a bench/native-speed suite for months: four compute kernels — recursive fib, a mandelbrot, a sieve, a big integer loop — written in machin, Rust and Zig, producing byte-identical output, so the timing compares the same computation three ways. The published result claimed machin won the integer loop by 20-25% . That claim also shipped inside machin guide , which is what every coding agent reads to learn the language. When I re-ran it, the margin was gone. Not shrunk — gone. So I read the harness instead of the output: for kernel in kernels : for lang in [ machin , rust , zig ]: for _ in range ( 5 ): # all 5 machin, THEN all 5 rust, THEN all 5 zig time ( binary ) It ran every sample of one language before starting the next. On a laptop that heats up and down-clocks during a three-second kernel, that does not measure the languages. It measures who had the misfortune of running last . Zig always went last. Zig always looked slowest. The fix is four lines — interleave the rounds, rotate who starts each one. Here is what my headline number did: intsum 10^9 before (blocked) after (interleaved) machin 2832 ms 3079.7 ms rust 3764 ms 3223.8 ms zig 3556 ms 3189.7 ms "machin +20-25%" machin +3% = a TIE A 20-25% win became a tie. I deleted the claim from the README and from machin guide . The harness now also refuses to declare a winner inside a 3% band, because the worst run-to-run spread I measured was 41% of the min sample. Calling winners inside that is how benchmarks start

"Most people have completely forgotten how chaotic it really was."

Through data brokers, ICE is buying the information you provided to open a credit card.

The Scenario Every Developer Recognizes It is mid-2026, and you have a feature to ship. You open ChatGPT or Claude, type something like "build me a function that parses webhook payloads and routes them to the right handler," and wait. The model returns something plausible. You paste it in, run it, and it almost works. So you prompt again: "fix the edge case where the payload is missing the event key." Another round. Then another. Forty-five minutes later, you have code that functions, but you also have a conversation thread that looks like a debugging session rather than a build session. You never actually described what you were building. You just started building it. This is the default mode for most developers using AI coding assistants in 2026, and it is expensive. According to McKinsey's State of AI in 2024 report ( source ), organizations that adopt structured design and planning approaches before implementing AI tools report higher success rates and better integration outcomes compared to those using ad-hoc implementation strategies. The pattern holds at the individual developer level too. Jumping straight into prompting skips the step that makes prompting useful: knowing precisely what you want before you ask for it. The fix is not a better model. It is a different sequence. What Design-First Actually Means in Practice Design-first means producing a written artifact that describes your system before you write a single prompt asking an AI to build it. Not a full technical document. A tight, structured description of inputs, outputs, constraints, and edge cases. Think of it as the brief you would hand to a contractor before they start work. The contractor analogy is useful because it reframes the relationship: you are not collaborating with the model in real time, you are commissioning it with a clear scope. Here is what that looks like concretely. Instead of opening Google Gemini and typing "help me build a webhook router," you spend ten minutes writing this

If you've ever wondered whether that wild Pokemon you just caught has competitive potential, you've probably heard the term IVs (Individual Values) thrown around. IVs are the hidden genetics of every Pokemon — the 0–31 numbers baked into your Pokemon at birth that determine how strong it can ultimately become. But here's the thing: the game never tells you what your IVs are. You have to reverse-engineer them. In this post, I'll walk you through exactly how IV calculators work under the hood — from the official stat formula, to the nature modifier trick, to why you often get a range instead of a single number. Live Tool: Try the calculator at randompokemongenerator.me/iv-calculator — free, no sign-up required, supports Gen III through Gen IX. What Are IVs, Exactly? Individual Values are six hidden integers between 0 and 31 , one for each stat (HP, Attack, Defense, Sp. Atk, Sp. Def, Speed). They represent the genetic potential of a Pokemon and are permanently set when the Pokemon is encountered or hatched — they can never be changed by leveling up or any in-game action. A stat with 31 IVs reaches its maximum possible value at level 100. A stat with 0 IVs starts at its theoretical minimum. In competitive play, players typically hunt for Pokemon with at least 3–4 perfect (31) IVs , with some strategies deliberately using 0 IVs in Defense or Speed for tactical advantages. The IV system as we know it today started in Generation III (Ruby/Sapphire/Emerald). Gen I–II used a predecessor called DVs (Determinant Values) , which only covered four stats and worked differently — so if you're playing on Virtual Console or Gen I/II, this calculator won't apply. The Stat Formula (Gen III+) The foundation of everything is the official stat calculation formula introduced in Generation III and still used today: For HP: HP = floor(((2 × BaseStat + IV + floor(EV / 4)) × Level) / 100) + Level + 10 For all other stats: Stat = floor((floor(((2 × BaseStat + IV + floor(EV / 4)) × Level) / 100

Google is set to host its next live Made by Google hardware launch event on August 12th, and the company says in a new video that comedian Trevor Noah will be hosting the show. The video indicates that the event will feature other celebrities and influencers as well, including Call Her Daddy host Alex Cooper […]

Groups of hackers are breaking into large U.S. financial firms to steal sensitive data and extort victims, Google’s security researchers report.