标签：#ev

Cloudflare has introduced Cloudflare Computer, a new open-source runtime designed to give AI agents something closer to a real "computer" instead of just ephemeral containers. It leverages Cloudflare isolates for fast serverless execution, making agents cheaper, faster, and more scalable, according to the company. By Sergio De Simone

I used to spend hours rewriting AI-drafted video scripts that sounded nothing like me. At best, I might finish one or two that were just okay, but most of the work was still on my shoulders. Now, with my new system, I can get 15 scripts done in one session. They match my writing style, my voice, and my company's knowledge. Instead of full rewrites, each script just needs a quick review. I built the system in about a week, and the difference showed up before the week was out. The first project was an internal video series to teach people about the software factory and LaunchDarkly, and I barely had to edit those scripts. For the first time, the AI handles most of the work. My recent projects have mostly involved agentic software delivery. The software factory is where all of it was heading, and my company spent months preparing to help customers build their own. We had internal material, public documentation, and ongoing conversations I wanted to add clarity to. So my challenge wasn't just learning, it was learning while producing content at the same time. Essentially, devrel. By the end, I had built what I like to call my own personal content factory. At first, it was separate from the software factory it describes, but over time, the line between them blurred. My approach is to let AI handle the bulk of the work, while I step in for the important decisions. This is the only way I've found to make AI content sound like me. The AI creates the drafts, and I step in at three key points: checking the voice, the facts, and the overall feel. The quality ceiling is set before the first draft. I didn't just ask for a video script about a topic. Instead, I gave Claude access to every source I had—public docs, internal notes, and product requirements. It used research agents to read everything at once and came back with clear, organized notes I could use. I set two important rules for this step. First, every part of the research was labeled as either public-safe or internal-o

There’s something magical about using instant cameras that smartphones can’t match. You can capture a moment, print it out, and then give the photo as a gift or hold onto it. Image quality won’t be all that good, but imperfections are part of the allure. We tested instant camera models from popular brands and landed […]

Change an SVG's color by editing fill and stroke , either as attributes or through CSS. Simple in theory. In practice there are three places a color can be declared in the same file, they follow the normal CSS cascade, and if you don't know that, "I changed the fill and nothing happened" turns into a twenty-minute debugging session. Here's the part of SVG color handling that usually doesn't get spelled out. fill and stroke are separate properties Every shape has an inside ( fill ) and an outline ( stroke ), set independently: <circle cx= "50" cy= "50" r= "40" fill= "#3366ff" stroke= "#000" stroke-width= "2" /> Unset fill defaults to black. Unset stroke defaults to none. If an icon is pure fill with no stroke at all (most converted icon-font SVGs are), editing stroke-width is never going to do anything visible, and that's usually the first dead end people hit. The cascade is the actual bug source A color can come from three places, and they don't have equal priority: A presentation attribute: <path fill="red" /> An inline style attribute: <path style="fill: red;" /> A <style> block or external stylesheet: path { fill: red; } Normal CSS specificity applies: style attribute beats stylesheet, stylesheet beats presentation attribute. Edit the fill="red" attribute directly, and if a <style> block elsewhere in the same file also targets that path, your edit is overridden and nothing changes on screen. No error, no warning, it just loses. If a color edit isn't sticking, grep the file for <style before assuming your tool, or your edit, is broken. This one thing accounts for most "the SVG editor is buggy" reports that are actually the cascade working exactly as designed. currentColor: SVG's inheritance trick Set fill="currentColor" and the shape stops carrying its own color and instead inherits whatever color is set to on an ancestor element, the same mechanism that makes text inherit color: <path fill= "currentColor" d= "..." /> .icon { color : #ff0000 ; } <span class= "icon

If you've opened a <path d="..."> string and had no idea what you were looking at, here's the short version: it's a tiny drawing language. A pen moves around a coordinate space, and each letter in the string is an instruction telling it what to do next. TL;DR M / m moves the pen, L / l draws a straight line, C / c and Q / q draw bezier curves, A / a draws an arc, Z / z closes the shape. Uppercase is absolute coordinates, lowercase is relative to the pen's current position. A visual path editor drags the exact same numbers you'd type by hand, it just shows you the curve instead of making you compute it. Reading a path string <path d="M10 10 L90 10 L90 90 Z" /> Broken down: move to (10, 10), draw a line to (90, 10), draw a line to (90, 90), close the path back to the start. That's a right triangle. Every path, no matter how complex, is this same pattern: a command letter followed by however many numbers that command needs, repeated. The command set Command Name What it takes M / m Move to x, y L / l Line to x, y C / c Cubic bezier control1 x/y, control2 x/y, end x/y Q / q Quadratic bezier control x/y, end x/y A / a Arc rx, ry, rotation, large-arc-flag, sweep-flag, end x/y Z / z Close path none C and Q are both bezier curves, the difference is one control point ( Q ) vs two ( C ). Two control points give you more independent influence over each end of the curve; one control point gives you a simpler, more symmetric curve. There are also shorthand continuations ( S / s , T / t ) for chaining smooth curves without repeating a control point, but the six above are what you'll hit constantly. A is the one people avoid writing by hand. Six parameters, two of which are flags (0 or 1) that determine which of four possible arcs you get for the same radii and endpoints. Flip one and you're not slightly off, you're on the opposite side of the ellipse. Absolute vs relative is the part that bites Every command above has an uppercase and lowercase form, and it's not cosmetic: <!-- a

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 | | :

I used to think IPv4 addresses were just random numbers until recently. It blew my mind when I started digging and understanding that they have an anatomy where every number after the dot means something very important. Note: To understand what IP addresses are, please consult this post because I won't be going over them here. IP Addresses: Digital Connectivity What is IPv4 in the First Place? IPv4 is short for Internet Protocol version 4 . As you might have already realized, it's the 4th version of the early test designs during the development of the Internet Protocol in test labs in the 1970s. The first real release, v4, came out in 1981 in a public document called RFC 791. RFC 791: STD 5: Internet Protocol IPv4 is an Internet Protocol that's written in what's called Dotted Decimal Notation (e.g., 172.17.0.3 ), where each portion is separated by a dot, and these portions are called octets. For example, 172 is the first octet and 17 is the second octet (more on this later). Before We Explore What Octets Are, Let's Take a Stroll to the Basics of Binary (Simplified) Have you ever wondered why your computer or phone requires electricity to function? Though electricity can be used as a raw power source for things like fans or speakers, where it's converted into other forms of energy such as movement or sound, it's good to know that electricity can function differently in your computer's RAM or SSD. Inside your computer are billions of extremely tiny transistors. These transistors form circuits that can create and maintain different electrical states, which the computer interprets as 0s and 1s. In simple terms, 0 represents the absence of the electrical state (OFF), while 1 represents its presence (ON). These states, which we represent with 0s and 1s, are called binary digits (or simply bits). 1 bit has the possibility of representing either 0 or 1, which doesn't represent much information, and that's where multi-bits come in. Every additional bit doubles the number of

There is a line in almost every Python web tutorial that nobody explains: uvicorn main:app --host 0.0.0.0 --port 8000 I copied it for weeks without thinking about it. Then I deployed the same application three times — to a local VM, to a production server, and into a container — and the correct value was different every time. Twice it was 0.0.0.0 . Once, in the place that mattered most, it was not. That gap is worth writing about, because the setting itself is trivial and the reasoning behind it is not. What the Flag Actually Controls A server process doesn't "open a port." It creates a socket and binds it to an address. The bind address answers one question: which network interfaces should this socket accept connections from? A machine has more than one interface: lo (loopback) — reachable only from inside the machine ( 127.0.0.1 ). Packets addressed there never reach a physical network card; the kernel loops them straight back. 0.0.0.0 — a wildcard meaning every interface this machine has , including ones added later. So the flag isn't about security or convenience. It's about reachability — and reachability depends entirely on what sits in front of the process. Case 1: The Local VM — 0.0.0.0 I was running the service inside a Multipass VM and wanted to hit it from the browser on my laptop. The laptop is outside the VM, so binding to loopback would have made the service invisible to it. curl inside the VM would work; the browser outside would get connection refused. Decision: wildcard bind. Nothing sits in front of the process, and nothing needs protecting. Case 2: Production — 127.0.0.1 Here I copied the same line at first, and it was wrong. The production box has a public IP. Binding to 0.0.0.0 there means the application is directly exposed to the internet: no TLS, no rate limiting, no authentication. Within hours of provisioning that server, its SSH logs showed hundreds of automated login attempts against usernames like admin and oracle . The same scanners try

Nota: en Lumora construimos libros infantiles personalizados con IA, y escribimos en diez idiomas. Este artículo cuenta un problema de i18n con el que chocamos de frente y cómo lo resolvimos. Lo contamos desde dentro, con nuestro nombre, porque el error nos costó reescribir una campaña entera. Casi todos los equipos tratamos el idioma como si fuera el país. Ponemos es en el selector, sacamos las cadenas a un JSON y damos el problema por cerrado. Funciona hasta el día en que tu producto tiene algo que ver con una fecha. Y entonces descubres que es no es un mercado, son varios calendarios distintos que comparten vocabulario. El día en que el regalo no llega el 25 Teníamos una campaña de regalos escrita en español. Correcta gramaticalmente, revisada, sin errores de traducción. Y aun así estaba mal para una parte grande de quien la leía: En España , el gran momento de regalo infantil no es el 25 de diciembre: son los Reyes Magos, el 6 de enero . Un mensaje que dice "pídelo antes del 24" le está dando a una familia española una fecha límite equivocada por casi dos semanas. En México conviven las dos cosas: Navidad y Reyes, con la rosca del 6 de enero. En Argentina, Chile o Uruguay , la Navidad es el 25 de diciembre… en pleno verano . A 30 grados. Con las vacaciones escolares largas empezando, no terminando. Fíjate en lo incómodo del asunto: los tres casos hablan español. Los tres pasan el mismo test de traducción. Y los tres necesitan un mensaje distinto, una imagen distinta y una fecha límite distinta. El mismo golpe existe en portugués. En Brasil la Navidad también cae en pleno verano, enero es mes de vacaciones, y el día en que de verdad se reinicia la rutina familiar no es el 1 de enero: es la volta às aulas , en febrero. Si tu calendario de contenidos asume "año nuevo, hábitos nuevos" en enero, en Brasil llegas un mes antes de que a nadie le importe. Por qué el código de locale no te salva La respuesta obvia es "usa es-ES y es-AR ". Es correcta y casi nunca es sufic

What makes the Samsung Galaxy Z Fold 8 Ultra so Ultra? Samsung's first foldable with the Ultra name isn't a souped-up version of the regular Fold 8, as you might expect. Where the Z Fold 8 is squat and wide, the Fold 8 Ultra retains the thin outer screen and square inner panel of the […]

Instacart introduced Blueberry, an AI-assisted incident response system that helps on-call engineers investigate production issues faster. It combines AI agents, operational data, and historical incident knowledge to generate grounded root cause hypotheses in Slack. It uses parallel subagents, MCP integrations, and incident history to reduce investigation time while keeping engineers in control. By Leela Kumili

A corrupted PDF is one of the most frustrating file problems. You have important content inside, but the document won't open, opens with garbled text, or shows missing pages. The file might be damaged from a bad download, a converter error, or a storage glitch. Recovering content from a broken PDF doesn't require complex forensic tools. Often, the individual pages are still readable — it's the document's structure (cross-reference tables, object streams) that's damaged. By extracting pages one by one into a fresh PDF, we can bypass the structural corruption. Here's how to build a browser-based PDF repair tool with Vue 3 and pdf-lib . The repair strategy The core insight: PDF structure and page content are somewhat independent . A PDF can have a broken cross-reference table or missing trailer objects, but the actual page content streams may still be perfectly readable. The repair approach: Load the damaged PDF and attempt to read each page For each successfully read page, copy it to a new PDF document Discard unreadable pages (they're lost anyway) Save the new document This is fundamentally different from "fixing" the original PDF. We're extracting what we can and rebuilding from the ground up. The stack Vue 3 with Composition API pdf-lib for PDF reading and page extraction Vite for bundling The core implementation < script setup lang= "ts" > import { ref } from ' vue ' import { PDFDocument } from ' pdf-lib ' const file = ref < File | null > ( null ) const totalPages = ref ( 0 ) const recoveredPages = ref ( 0 ) const repairing = ref ( false ) const result = ref < Uint8Array | null > ( null ) const error = ref < string | null > ( null ) async function repairPdf () { if ( ! file . value ) return repairing . value = true error . value = null try { const arrayBuffer = await file . value . arrayBuffer () const damaged = await PDFDocument . load ( arrayBuffer , { ignoreEncryption : true , updateMetadata : false , }) totalPages . value = damaged . getPageCount () const resu

If you are integrating ZATCA Phase 2 (Saudi Arabia's Fatoora e-invoicing) and you have seen this: { "type" : "ERROR" , "code" : "signed-properties-hashing" , "category" : "CERTIFICATE_ERRORS" , "message" : "Invalid signed properties hashing, SignedProperties with id='xadesSignedProperties'" } ...after your invoice sailed through /compliance/invoices , this post is for you. It is the single most confusing failure mode in the whole integration, and the fix is not what the error suggests. The trap: SignedProperties exists in two byte-shapes The XAdES SignedProperties block is referenced twice in your signed document: ds:Reference URI="#xadesSignedProperties" carries a digest of the block. The block itself is embedded inside ds:Object > xades:QualifyingProperties . The natural assumption is that both refer to the same bytes. They do not. The hashed shape carries namespace declarations and starts at column 0: <xades:SignedProperties xmlns:xades= "http://uri.etsi.org/01903/v1.3.2#" Id= "xadesSignedProperties" > <xades:SignedSignatureProperties> <xades:SigningTime> 2026-08-07T02:14:33 </xades:SigningTime> <xades:SigningCertificate> <xades:Cert> <xades:CertDigest> <ds:DigestMethod xmlns:ds= "http://www.w3.org/2000/09/xmldsig#" Algorithm= "http://www.w3.org/2001/04/xmlenc#sha256" /> The embedded shape carries no namespace declarations (they are inherited from ancestors) and its root element is indented to column 32 : <xades:SignedProperties Id= "xadesSignedProperties" > <xades:SignedSignatureProperties> Embed the hashed shape verbatim - the intuitive thing to do - and the gateway rejects with signed-properties-hashing , even though your indentation "looks right". The second half of the trap: the digest encoding The digest is not the raw SHA-256 bytes in base64. It is base64 of the hex string : const crypto = require ( ' crypto ' ); // hashedShape = the namespaced, column-0 variant above const propsDigest = Buffer . from ( crypto . createHash ( ' sha256 ' ). update ( Buffer .

👋 Hello DEV Community! I'm excited to finally join DEV! I'm a developer, entrepreneur, and lifelong learner who enjoys building practical web solutions with WordPress, PHP, and modern web technologies. Over the past few years I've been working on: 🚀 WordPress plugins and starter websites 💻 Affordable web solutions for individuals and small businesses 📈 Web analytics and digital marketing tools 🌱 Exploring software architecture, clean code, and open-source development I'm also building and experimenting with digital products that solve real-world problems while documenting what I learn along the way. Here you'll find posts about: WordPress development PHP programming Building and launching web products Software engineering lessons Productivity and business insights for developers Occasionally, mathematics and calculus when it connects to programming or analytics I'm looking forward to learning from this amazing community, contributing where I can, and connecting with fellow developers. Thanks for having me! 😊

Artificial intelligence is rapidly changing how engineering teams respond to production incidents, offering the ability to summarize incident channels, analyze unfamiliar code, suggest remediation steps, generate pull requests, and increasingly assist with diagnosis. By Craig Risi

I tested the best webcams across various prices to find the top option. Here’s what I learned.

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

From the omnipresent One Night Only trailer, you'd be forgiven for thinking that the movie glosses over its terrifying and absurd premise - in otherwise normal modern New York, single people can only have sex one night a year - to deliver an entirely traditional romantic comedy. I'm here to report that it does not […]

ElevenLabs now offers a Dubbing Translation API designed to turn multilingual video and audio localization into a single automated workflow. The service accepts a video, audio file, or source URL, lets developers choose one or more target languages, and coordinates transcription, translation, voice generation, speaker identity preservation, and timing alignment before returning dubbed assets. The central change is not simply another voice feature. ElevenLabs is presenting the pipeline as an integrated API operation rather than a set of services developers must orchestrate independently. Its Dubbing Translation API product page says users can dub and translate video in 90+ languages in one call , with translation, voice cloning, and timing synchronization handled server-side. For localization teams, that approach could reduce the application logic required to move from an original asset to versions for multiple language audiences. It does not remove the need for teams to assess translation quality, brand requirements, and appropriate use of cloned voices, but it consolidates the underlying production stages into one API surface. What the unified dubbing workflow does ElevenLabs' dubbing documentation describes an end-to-end sequence comprising transcription, translation, voice generation, and video synchronization. The Dubbing Translation API places those stages behind a single request flow, with the aim of retaining the original speaker's identity and the timing of the source material in the resulting dubbed output. That matters because dubbing is more than text translation. A usable localized video or audio asset needs speech that fits the surrounding media, while the voice and delivery should remain coherent for the intended audience. ElevenLabs describes its synchronization capability in terms of preserving timing and tone, positioning the API for workflows where the final media asset, rather than a translated script alone, is the required output. The documented

ElevenLabs has introduced Dubbing v2 , a rearchitected AI dubbing model built to retain a speaker's emotion, delivery, and timing while translating content across more than 90 languages. The update expands the company’s localization proposition beyond a basic language replacement: its documented controls cover dialect-specific accents, multi-speaker material, and background audio management for more complex video and audio scenes. According to ElevenLabs’ Dubbing v2 announcement , the model is designed for creators, marketers, studios, and broadcasters that need to localize video at production scale. It is integrated with ElevenCreative for one-click video localization and ElevenProductions, the company’s professional localization service. The central goal is preserving the original performance rather than simply generating translated speech. That matters for material in which pacing, vocal emphasis, and emotional delivery are part of the message, including marketing campaigns, creator videos, and professionally produced programming. Dubbing v2 is intended to synchronize translated dialogue with the source speaker’s timing and delivery across its supported languages. What Dubbing v2 changes for localization workflows The most practical additions are the controls documented for ElevenLabs' dubbing workflow. They give teams more ways to shape a dub around the source material and the intended audience, particularly when a project includes regional language variation or a mix of dialogue and sound. Localization need Dubbing v2 capability Documented control or workflow Regional language variation Dialect-specific accent selection target_accent , marked experimental Scenes with several voices Multi-speaker dubbing num_speakers Music, effects, or ambient sound Background audio management foreground_audio_file , background_audio_file , and drop_background_audio End-to-end video localization Integrated production workflows ElevenCreative and ElevenProductions For Spanish-lan