标签：#t

New gaming gear is always promising to shave milliseconds off latency or response time. The improvements usually aren't obvious. But the Glorious GHS InfinitePlay wireless headset, on the other hand, could save you far more than just milliseconds: It has the fastest hot-swap battery system I've ever used, and the headset doesn't die when you […]

Halliday’s G2 glasses can listen to and summarize your workplace meetings—with no video involved.

It is time to bring opposites together, to stir up these stagnant waters, and to restore an industry that is dynamic from within, with multiple connections that were previously impossible or unimaginable. In the previous parts of this series, we explored how PureScript acts as a quiet rewrite of the Web , offering a territory of mathematical precision above JavaScript, and how its universal polymorphism allows it to target wildly different runtimes like Node.js, Erlang's BEAM, Chez Scheme, etc. But hey, what about the elephant in the room? Yes, what is running the majority of the web, today, in the shadow? You know. The eternal one, the big one, the one whose death everyone talks about, but who never truly dies. PHP. While the modern JavaScript ecosystem (among others) races toward edge computing and containerized microservices, around 70% of the web is still powered by PHP . From legacy WordPress sites to affordable shared hosting (cPanel, OVH, GoDaddy, basic LAMP stacks), a massive portion of the internet lives in environments where escaping the PHP runtime is either financially unviable, technically impossible, or simply unnecessary. And yet... For a long time, the PHP ecosystem has been left without a true, strictly typed functional and safe alternative, like TypeScript for Javascript. Even though recent PHP versions introduced fantastic features (short closures, match expressions, Fibers), the frustration that comes with its historical limitations and dynamic typing remains very real for some of its users. Very well! This is why I built phpurs : a brand new compiler backend that transpiles PureScript directly to modern PHP 8.4+ syntax. It is no longer a proof of concept. It is passing the official PureScript test suite , and it is ready for production. Here is how we bring the elegance of purely functional programming to the big web, while keeping deployment as simple as a legacy FTP upload, or anything alike. The easy deployment paradox In modern DevOps, deplo

Building AI Agents That Don't Hallucinate: Structured Workflows, Guardrails, and Per-Step Evaluation How we replaced fragile prompt chains with typed schemas, validation gates, and evaluation at every step — 94% task success vs 60% baseline The Prompt Chain Trap January 2024. We built a "research agent" — 12 prompts chained together: Decompose question → 2. Search planning → 3. Execute searches → 4. Extract facts → 5. Synthesize → 6. Fact-check → 7. Format → ... It worked 60% of the time. The other 40%: Step 3 returned malformed JSON → Step 4 crashed Step 5 hallucinated citations → Step 6 missed it Step 7 output wrong format → Downstream consumer failed No visibility into which step failed Debugging meant reading 12 LLM calls' worth of logs. Adding a step broke three others. The Shift: Agents as Typed Workflows We moved from prompt chains to structured workflows with: Pydantic schemas for every step input/output Guardrails that validate and auto-retry Explicit state machine (not implicit chaining) Evaluation harness per step (not just end-to-end) ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ │ Decompose │──▶│ Search │──▶│ Extract │──▶│ Synthesize │ │ Question │ │ Planning │ │ Facts │ │ Answer │ │ │ │ │ │ │ │ │ │ In: Query │ │ In: Plan │ │ In: Results │ │ In: Facts │ │ Out: SubQ[] │ │ Out: Steps │ │ Out: Fact[] │ │ Out: Answer │ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ │ │ │ │ ▼ ▼ ▼ ▼ [Schema] [Schema] [Schema] [Schema] [Guardrail] [Guardrail] [Guardrail] [Guardrail] [Eval: 0.9] [Eval: 0.85] [Eval: 0.9] [Eval: 0.95] Core Abstractions # agent_eval/schemas.py from pydantic import BaseModel , Field from typing import Literal , Any class DecomposeInput ( BaseModel ): user_query : str context : dict = Field ( default_factory = dict ) class DecomposeOutput ( BaseModel ): sub_questions : list [ str ] = Field ( min_length = 1 , max_length = 5 ) requires_tools : bool reasoning : str class PlanInput ( BaseModel ): sub_questions : list [

A practical roadmap for students who want to build real AWS skills, create an impressive portfolio, and prepare for cloud engineering careers. Introduction Every year, thousands of students begin learning AWS. They watch training videos, collect certificates, complete online courses, and share digital badges on social media. Yet when internship interviews or entry-level cloud engineering opportunities arrive, many struggle to answer a simple question: "What have you actually built on AWS?" The cloud industry rewards practical experience, not passive learning. AWS itself focuses beginner learning on hands-on experience with foundational services such as Amazon S3, Amazon EC2, Amazon VPC, Amazon RDS, and cloud security because these services power most real-world cloud environments. If you're a student aiming for a career in Cloud Engineering, DevOps, Site Reliability Engineering (SRE), Solutions Architecture, or Platform Engineering, this article provides a practical roadmap that can help you become job-ready. Why Students Should Learn AWS Cloud computing has become the backbone of modern technology. Companies of every size use cloud platforms to: Host applications Store data Deploy AI workloads Build scalable systems Reduce infrastructure costs Improve reliability As a result, companies continue to hire professionals with cloud skills across software engineering, cybersecurity, DevOps, networking, and data engineering domains. AWS offers dedicated learning paths, hands-on labs, certification tracks, and career-focused programs specifically designed to help learners develop these skills. The key question is not: "Which AWS service should I memorize?" The better question is: "Can I design, deploy, secure, and troubleshoot cloud solutions?" The 5 AWS Projects Every Student Should Build Instead of completing another course, build these five projects. Project 1: Host a Static Website Using Amazon S3 What You'll Learn Cloud Storage Static Website Hosting Bucket Policies O

Building a Decompiler Pipeline in Rust: Why Fission Separates NIR and HIR Decompiler output often looks simple from the outside. A binary goes in. Pseudocode comes out. But between those two points, a decompiler must recover several different kinds of information: instruction semantics register and memory effects control flow stack variables calling conventions data types expressions loops and conditionals readable source-like structure Trying to represent all of this in one intermediate representation quickly becomes difficult. While building Fission , a reverse-engineering and binary decompilation workspace written primarily in Rust, I decided to separate the decompiler pipeline into two main intermediate representations: NIR , a lower-level representation intended to preserve machine semantics HIR , a higher-level representation intended to express recovered, human-readable program structure This article explains why that separation exists, what each representation owns, and why it makes decompiler development easier to reason about. Correctness and readability want different things A decompiler has at least two responsibilities. First, it must preserve the behavior of the original machine code. Second, it must produce output that a human can understand. Those goals overlap, but they are not identical. Consider a simplified fragment of machine-level behavior: tmp0 = RAX tmp1 = tmp0 + 1 RAX = tmp1 flags = update_flags(tmp0, 1, tmp1) A human reader may prefer to see: rax ++ ; The concise form is easier to read, but it omits details that may still matter elsewhere in the pipeline. The flags update could affect a later conditional branch. The operation width may matter. The source and destination could alias. The operation may have originated from an instruction with additional side effects. If the decompiler converts everything into source-like syntax too early, it becomes easy to discard evidence. If it keeps everything at machine level until the final rendering st

AI agents no longer just answer questions — they book meetings, approve refunds, call APIs, update records, and chain together dozens of small decisions into a single outcome. Most of the time, this works quietly and well. But when something goes wrong, a troubling question surfaces: why did the agent do that? For a large share of deployed systems today, there's no good answer. The reasoning, the data consulted, the tools invoked, and the intermediate steps simply weren't recorded. This is the problem of undocumented agent decisions — and it's becoming one of the central risks of the agentic AI era. What "Undocumented" Actually Means An undocumented decision isn't necessarily a bad one. It's simply one that can't be reconstructed after the fact. In practice, this shows up in a few common ways: Output without reasoning. The system logs what the agent did, but not the chain of thought, tool calls, or data sources that led there. Ephemeral intermediate state. Multi-step agent chains often discard the "scratch work" between steps — the very material that would explain a decision — once the final output is produced. Reviewer blind spots. A human approves a final recommendation without ever seeing the reasoning that produced it, which looks like oversight but isn't meaningful oversight. No durable storage. Logs exist for a few days or weeks, then age out — so when someone asks for the record months later, it's gone. The common thread is a gap between acting and accounting for the action. This Is Becoming Urgent A few forces are converging to make this problem harder to ignore: Agents are doing more, autonomously. As agents move from single-turn assistants to systems that independently call APIs, touch databases, and trigger downstream workflows, the number of undocumented micro-decisions multiplies. A single customer request might now involve a dozen internal steps, each a potential decision point. Incidents are already happening. Surveys of enterprise AI deployments in 2

AI coding tools have gotten very good at one thing: generating code fast. What they haven't gotten good at is discipline. They don't know your architecture. They don't remember that you rejected a pattern last sprint. They don't know which parts of your context window are signal and which are noise. And they have no concept of a development workflow — no phases, no review gates, no verification steps. You describe what you want, they generate, and you hope the output fits. For small tasks this works fine. For anything that touches your real codebase at scale — refactors, new features with cross-cutting concerns, compliance-sensitive changes — the lack of workflow structure creates subtle, expensive problems that compound over time. We've been building Ortho to address two of these problems: workflow discipline (through ASES, a 6-phase AI development methodology) and context discipline (through a 9-component token optimization pipeline). This post explains both. The workflow problem with AI coding tools When a junior engineer joins your team, you don't just hand them a task and say "generate." There's a process: understand the codebase, plan the change, get the architecture reviewed, build it, test it, verify it, get it reviewed. The process exists because individual steps catch different categories of mistakes. AI coding tools collapse all of that into one step. You prompt. You get code. Done. The result isn't always bad code per file. The problem is architectural: the AI has no model of your layer boundaries, so it imports from layers it shouldn't touch. It has no memory of past decisions, so it re-proposes patterns you've already rejected. It has no verification step, so it confidently generates code that looks right but has subtle issues a reviewer would have caught in 30 seconds. The missing piece isn't a smarter model. It's a workflow. ASES: A 6-phase workflow for AI-assisted development ASES (v1.2) is the methodology built into Ortho's orchestration layer. It

A request starts at 10:00:00.900 and finishes 200 ms later. Your latency log says -800 ms . That sounds impossible until the machine corrects its clock between the two reads. Then ordinary code turns an adjustable wall clock into a broken stopwatch. The subtraction that works—until it does not This is probably hiding in one of your metrics helpers: async function timed ( operation ) { const startedAt = Date . now (); const result = await operation (); return { result , durationMs : Date . now () - startedAt , }; } Most of the time, this reports a sensible number. That is what makes it dangerous. Date.now() answers a calendar question: how many milliseconds have passed since the Unix epoch according to this machine? The answer must remain comparable with logs, users, and other computers, so synchronization software is allowed to correct it. A correction can change its rate, jump it forward, or move it backward. If either correction lands between the two calls, the subtraction measures the clock adjustment as if it were work. A timestamp is a coordinate. A duration is a distance. They need different instruments. Your computer already has two kinds of clock Operating systems expose several clocks, but two mental buckets cover most application code: Question Clock JavaScript example “When did this happen?” Wall clock Date.now() , new Date() “How long did this take?” Monotonic clock performance.now() A wall clock follows civil time. It has an epoch and can be serialized as an ISO timestamp. A monotonic clock starts at an arbitrary origin and promises that later readings will not be lower than earlier readings. wall: 1000 ── 1001 ── 0998 ── 0999 clock correction monotonic: 40 ───── 41 ───── 42 ───── 43 On Linux, CLOCK_REALTIME is settable wall time . CLOCK_MONOTONIC cannot be set and does not make discontinuous jumps backward, although gradual frequency adjustment can affect its rate. Linux even has CLOCK_BOOTTIME for a monotonic counter that includes suspended time. Thre

Open the network tab on almost any web app, trigger a failed request, and you'll usually find one of two things staring back at the user: a raw stack trace, or a message so generic it might as well say "something happened." Neither one helps. Both exist for the same reason they were written by developers, for developers, and never translated for the person actually using the product. The Error Message Nobody Designed Most UI elements go through some level of design scrutiny. Buttons get spacing decisions. Forms get validation states. But error messages? They're usually whatever string got thrown at the moment something broke, copy-pasted straight from a try/catch block into a toast notification. "Error 500: Internal Server Error." "Failed to fetch." "Unexpected token in JSON at position 4." These are diagnostic breadcrumbs for engineers debugging a system. To a user trying to submit a form or complete a purchase, they're just noise confirmation that something went wrong, with zero indication of what to do next. This is where good web app design services earn their keep not in the buttons and layouts everyone notices, but in the failure states nobody plans for until users start complaining. Why This Keeps Happening It's not that teams don't care. It's that error handling sits at the intersection of two disciplines that rarely talk to each other at the moment. Backend logic throws whatever exception the code produces. The front end just needs something to display so the app doesn't silently freeze. Nobody's job, at that moment, is to ask: "what should the user actually understand right now?" The result is a UI layer that's polished everywhere except the one place users encounter when things go wrong which, ironically, is exactly when clear communication matters most. What a Good Error Message Actually Does A well-designed error message does three things a raw exception never does. It tells the user what happened, in plain language not "Error: NetworkException," but "W

You ask an LLM for JSON. You get this: json { "name": "test", "valid": True, "items": [1, 2, markdown Three problems in one response: markdown fences the LLM wasn't supposed to add, True instead of true (Python literal), and the response was cut off mid-array. JSON.parse() throws on all three. A linter tells you what's wrong. But you're left fixing it manually. I kept hitting this wall so often that I built a dedicated repair pipeline: AI JSONMedic . Why existing tools don't cut it JSONLint / JSONFormatter — great for valid-ish JSON with one missing comma. Not built for LLM failure modes: they flag errors but don't repair them. jsonrepair (npm) — solid library, handles many cases. AI JSONMedic actually uses it as a last-resort fallback. But it doesn't tell you what it changed, and doesn't handle all the LLM-specific cases we needed. The 14 failure modes we target Through building this, we catalogued how LLMs specifically break JSON: Markdown fences — ` json wrapping the output Trailing commas — [1, 2, 3,] (the model "runs out" of items but adds one more comma) Python literals — True , False , None instead of true , false , null Single quotes — {'key': 'value'} instead of double quotes Smart quotes — "key" (curly quotes from copy-paste) Unclosed brackets — truncated at max_tokens mid-array or mid-object Unclosed strings — "value without closing Concatenated objects — {"a":1}{"b":2} when streaming produces multiple chunks NDJSON — newline-delimited JSON that needs wrapping Python-style comments — # this is a comment inside JSON JavaScript-style comments — // inline or /* block */ Escaped backslashes — \\n instead of \n Duplicate keys — same key appearing twice (ambiguous — we warn, not silently pick) BOM / encoding issues — UTF-8 BOM at start of response What makes the repair pipeline different Each pass targets one failure mode. The order matters — strip fences first, then normalize quotes, then fix commas, then close truncated structures. Each change is tracked. The

Managing payload cms lexical tables in a content-heavy site means enabling EXPERIMENTAL_TableFeature — but the real trap is the markdown import that strips tables without warning. We lost a whole batch of production blog posts to this exact hole before we found the fix. Here’s why it happens and the step-by-step configuration that keeps your tables intact. The Silent Table Eater: Payload CMS Lexical Tables and Markdown Conversion The default markdown-to-Lexical conversion helper completely ignores your editor’s feature list. So even when you’ve added the table feature to your editor config, every GFM table in imported markdown is silently dropped. Here’s the code that ate our data: import { editorConfigFactory , defaultFeatures } from ' @payloadcms/richtext-lexical ' // ❌ This uses a plain config that doesn’t know about tables const mdConverter = editorConfigFactory . default ({ features : defaultFeatures , }) const lexicalData = mdConverter . parse ( ' # Hello \n\n | A | B | \n |---|---| \n | 1 | 2 | ' ) // result: { root: … } — no table node anywhere The problem: editorConfigFactory.default builds a conversion pipeline from a static feature set, not from your actual editor config. Any experimental or custom feature you’ve wired into the editor simply isn’t there during markdown parsing. Fix It: Wire EXPERIMENTAL_TableFeature Into the Conversion Config Switch to editorConfigFactory.fromFeatures , which actually reads the feature array you provide. Include the table feature alongside the defaults, and the markdown converter will start producing proper Lexical table nodes. import { editorConfigFactory , defaultFeatures , EXPERIMENTAL_TableFeature , } from ' @payloadcms/richtext-lexical ' const mdConverter = editorConfigFactory . fromFeatures ({ features : [... defaultFeatures , EXPERIMENTAL_TableFeature ()], }) Takeaway: You must add EXPERIMENTAL_TableFeature() to both your editor’s features array and to every markdown conversion config. Missing one side silently eat

When you wire up a payload cms live preview , you’re not just plumbing a URL—you’re building a contract between the admin panel and your Next.js App Router. The goal is keystrokes-ago fidelity: editors click Preview, land on your front end, and see exactly what’s in the draft, even when the public site is cached to the hilt. Our implementation at techpotions settled on one preview route, one shared secret, and one shared URL builder. Here’s every decision that made it work. One /next/preview route for the entire site The admin panel’s preview button doesn’t need to know about your page structure. It calls a single /next/preview route with a secret query param and a slug search param that points to the document being previewed. // app/(payload)/next/preview/route.ts import { draftMode } from ' next/headers ' import { redirect } from ' next/navigation ' export async function GET ( request : Request ) { const { searchParams } = new URL ( request . url ) const secret = searchParams . get ( ' secret ' ) const slug = searchParams . get ( ' slug ' ) if ( secret !== process . env . PREVIEW_SECRET ) { return new Response ( ' Invalid token ' , { status : 401 }) } const draft = await draftMode () draft . enable () redirect ( slug ?? ' / ' ) } That’s the entire route. No collection-specific logic, no second-guessing which page type is involved. The redirect lands on the actual page, which reads draftMode().isEnabled and fetches accordingly. This is the pattern the Payload CMS preview documentation expects: a function that resolves to a string with additional URL parameters pointing to your app. The preview-URL builder lives in one shared lib Here’s where most implementations drift apart. The admin config, the preview route, and each page component all need to agree on how a preview URL is constructed. Store that logic in one place—a single getPreviewUrl utility imported everywhere—or you’ll be chasing 404s in production when someone renames a collection slug. // lib/getPreviewU

This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. China’s AI models have Trump’s AI world at war with itself Last weekend, several current and former advisers to President Donald Trump on AI publicly lobbed insults at the country’s leading…

These WIRED-tested wearables reduce your reliance on a phone while keeping you connected.

GitLab has introduced a new approach to Green DevOps, demonstrating how software engineering teams can measure the carbon emissions generated by their CI/CD pipelines. By Craig Risi

Orchestrate agents at scale while reducing cost Discussion | Link

It’s a lot : According to information obtained by The Tech , MIT is spending over $3 million on more than 500 AI surveillance cameras in academic buildings, residence halls, and outdoor areas along Memorial Drive. Installation of the new cameras, along with the wiring and infrastructure that will support them, began November 2025 and will likely continue until September 2026. Technical specifications for the cameras suggest that they will be capable of collecting real-time face and object classification data, including detection of motion, loitering, crowds, face masks, and camera tampering. Individuals can also be automatically classified on the basis of clothing color, gender, and age, up to a distance of 35 feet (11 meters) from the camera. According to a statement from MIT spokesperson Kimberly Allen, any collected data is “retained up to 30 days,” unless an exception is granted...

These locks, lights, and other smart home upgrades let you add automation without messing up your home’s vibe.

Michigan has been at the epicenter of the wildfire smoke and explosive diarrhea outbreaks. The progressive Democrat talks to WIRED about how he would tackle the forces behind them if he’s elected.