Benjamin Lesné

860 posts

Benjamin Lesné

Benjamin Lesné

@LesneBenjamin

Web Developer | mainly with React | also on bluesky 👉 https://t.co/1QaBqGvLaL

France Katılım Eylül 2023
700 Takip Edilen50 Takipçiler
Benjamin Lesné retweetledi
Sadique Ali
Sadique Ali@apnahive·
TypeScript 6.0 shipped March 23, 2026. It's the last version of the TypeScript compiler written in TypeScript. Everything after it — TypeScript 7.0 — is rebuilt in Go. 🐳 The numbers: → VS Code (1.5M lines): 77 seconds → 7.5 seconds → Sentry codebase: 133 seconds → 16 seconds → Memory usage: cut in half That's not a small improvement. That's a different category of tool. But first — TypeScript 6.0 is the bridge release you need to get through. Here's what changed: 🔧 9 new defaults: strict ON, ESM as default, bundler module resolution, all files treated as modules ❌ Breaking changes: ES5 target removed, AMD/UMD deprecated, --outFile gone, @types no longer auto-discovered ✅ New features: - Temporal API types (proper date handling, finally) - RegExp.escape() and Promise.try() types - Map.getOrInsert() typed correctly - #/ subpath imports - Improved method type inference 💡 The Go compiler (tsgo) is already available: npm install @typescript/tsgo tsgo --project tsconfig.json Try it. The compile time difference is immediate and shocking. I wrote the full guide for what changed, what broke, how to upgrade, and what TypeScript 7.0 actually means for your CI/CD and monorepo workflows. Is your team already planning the 6.0 upgrade? 👇 #TypeScript #JavaScript #WebDev #Frontend #NodeJS #100DaysOfBlogging medium.com/p/typescript-6…
English
2
4
58
7.9K
Benjamin Lesné retweetledi
Joan Larroumec
Joan Larroumec@larroumecj·
Je vous propose un graphique qui je crois n'avait encore jamais été fait : les flux monétaires complets Français => sphère publique => Français. En effet, la sphère publique est une grande machine à redistribuer. On prend aux Français de plein de manières différentes (TVA, IR, amendes, etc.), ça passe par des administrations diverses complexes et ça retombe dans le poche d'autres Français de façons diverses (transferts, services publiques, infrastructures, etc.) Ce graph retrace les flux, et permet de voir les transferts par niveau de richesse et par génération. Cela m'a forcé à faire des trucs rigolos, comme décomposer qui paye vraiment l'impôt sur les sociétés. J'arrive à la conclusion que seulement 15% de l'IS est payé par les plus riches des actionnaires, et qu'en proportion de leurs revenus, les plus pauvres payent plus pour l'IS. Pourquoi ? Parce qu'une partie du coût de l'IS est portée sur des salaires plus faibles et des prix à la consommation plus élevés. Ce n'est forcément pas parfait, et je prends tous vos commentaires et corrections. Je pense néanmoins que les ordres de grandeur sont corrects. Vous trouverez les versions détaillées et interactives ici : #depenses" target="_blank" rel="nofollow noopener">francetdb.com/#depenses (Par ailleurs j'ai ajouté partout sur le site les sources et les formules)
Joan Larroumec tweet mediaJoan Larroumec tweet mediaJoan Larroumec tweet mediaJoan Larroumec tweet media
Français
99
256
1.2K
89.1K
Benjamin Lesné retweetledi
MDN Web Docs
MDN Web Docs@MozDevNet·
Did you know about Clear-Site-Data header? 👀 One HTTP header to clear cookies, storage, or cache for your site. Perfect for logout flows. Learn more 👇 developer.mozilla.org/en-US/docs/Web…
MDN Web Docs tweet media
English
9
92
862
44K
Benjamin Lesné retweetledi
Resend
Resend@resend·
Today, we're introducing a Logs API with 2 new endpoints. 🔸 Equip your agent with better debugging tools 🔸 Build custom dashboards, reports, and monitoring tools 🔸 Integrate with your existing infrastructure
English
1
3
44
3K
Benjamin Lesné retweetledi
alistair
alistair@alistaiir·
We're launching the Monitor tool in today's Claude Code release. Claude spawns a background process and each stdout line streams into the conversation, without blocking the thread. e.g. "Use the monitor tool and `kubectl logs -f | grep ..` to listen for errors, make a pr to fix any crashes" This is both more reliable and token efficient than polling within the agent loop.
English
58
93
1.1K
126.4K
Noah Zweben
Noah Zweben@noahzweben·
Thrilled to announce the Monitor tool which lets Claude create background scripts that wake the agent up when needed. Big token saver and great way to move away from polling in the agent loop Claude can now: * Follow logs for errors * Poll PRs via script * and more!
English
224
458
6K
1.1M
Benjamin Lesné retweetledi
Kun Chen
Kun Chen@kunchenguid·
alright agent nerds, if you care about your tokens and usage limits, pay attention to the tools you give to your agents. i built a benchmark that compared various browser tools for agents, and here's an example of their massive difference in cost and latency doing the same task
English
75
157
2.3K
522.8K
Benjamin Lesné retweetledi
Kai
Kai@hqmank·
Claude Code rate limits driving you crazy? Here’s a simple fix.🎯 Go to the Claude website → Code → Scheduled → create a scheduled task → choose Custom and set: 0 8,13,18,23 * * * This locks the 5-hour windows to start at 8 AM, giving you clean time blocks: 8–13 / 13–18 / 18–23 Each window can fully utilize the token quota. Best part? No need to keep a machine running cron. #claude
Kai tweet media
English
26
62
936
112.7K
Benjamin Lesné retweetledi
David K 🎹
David K 🎹@DavidKPiano·
Quick React tip: don't subscribe to data that you don't render; it causes unnecessary rerenders. If it's only used in event handlers, read it there. AI coding agents tend to get this wrong.
David K 🎹 tweet media
English
26
29
664
58.8K
Benjamin Lesné retweetledi
Adam Wathan
Adam Wathan@adamwathan·
Quick ui.sh demo — generating multiple design ideas to choose from, no matter what tech stack you use:
English
186
136
2.8K
258.5K
Benjamin Lesné retweetledi
Vaishnavi
Vaishnavi@_vmlops·
MICROSOFT BUILT A TOOL THAT CONVERTS LITERALLY ANYTHING INTO CLEAN MARKDOWN FOR YOUR LLM pdfs. word docs. excel. powerpoint. audio. youtube urls one pip install and your AI pipeline stops choking on raw files forever no custom parsers. no broken layouts. no garbled text. just clean, structured markdown your LLM can actually read github.com/microsoft/mark…
English
77
515
4.9K
759.7K
Benjamin Lesné retweetledi
Charly Wargnier
Charly Wargnier@DataChaz·
🚨 Extracting data from PDFs just got solved. Someone open-sourced a tool that turns PDFs into Markdown at 100 pages a second 🤯 It’s called OpenDataLoader. It runs flawlessly on CPU and decodes tables, complex layouts, and nested structures like an absolute pro. Best part? 100% free and open-source. Grab the repo link in the 🧵↓
Charly Wargnier tweet media
English
22
214
1.2K
83.1K
Benjamin Lesné retweetledi
Seb ⚛️ ThisWeekInReact.com
Seb ⚛️ ThisWeekInReact.com@sebastienlorber·
New React CVE just dropped 😆 DOS vulnerability in Server Functions Patched versions already out: - 19.0.5 - 19.1.6 - 19.2.5
Seb ⚛️ ThisWeekInReact.com tweet media
English
20
58
497
65.2K
Benjamin Lesné retweetledi
Matt Pocock
Matt Pocock@mattpocockuk·
This is essentially the same as a homebrew workflow I've used for a while: 1. Create a /prototype-ui skill 2. Give it instructions to create multiple radically different designs 3. Give it a picker component to toggle between them Bingo
Adam Wathan@adamwathan

Quick ui.sh demo — generating multiple design ideas to choose from, no matter what tech stack you use:

English
28
24
761
80.1K
Benjamin Lesné retweetledi
AI Engineer
AI Engineer@aiDotEngineer·
🎥🔴 LIVESTREAM ANNOUNCEMENT: Keynotes + @openclaw / Personal Agents track 🔔 Hit the bell on our YT channel 👉 youtube.com/watch?v=O_IMsE… Starts promptly April 9, 9:00am UK / 4:00am ET Featuring talks from: @cramforce - CTO at Vercel @RaiaHadsell - VP of Research at Google Deepmind @_lopopolo - Member of Technical Staff at OpenAI @thekitze - Founder at Sizzy @swyx - co-founder at AI Engineer, Editor at Latent Space @GergelyOrosz - Author at The Pragmatic Engineer @mattpocockuk - Creator, AI Hero @threepointone - Founder, PartyKit, Principal Systems Engineer at Cloudflare @steipete - Creator, OpenClaw @vincent_koc - Lead AI Research Engineer at Comet, Maintainer at OpenClaw @velvet_shark - Founder at VelvetShark @somalley108 - Principal Software Engineer at RedHat, Maintainer at OpenClaw @nickytonline - Developer Advocate at Pomerium @thorwebdev - Dev Relations Engineer at Google Deepmind @fawiatrowski - Founder & CEO at Viktor @onusoz - OpenClaw maintainer + the one and only @philhawksworth -- devrel @deno_land -- as our MC for the day!
YouTube video
YouTube
English
13
18
112
71.9K
Benjamin Lesné retweetledi
Om Patel
Om Patel@om_patel5·
THIS GUY AUDITED 926 CLAUDE CODE SESSIONS AND FOUND MOST OF THE TOKEN WASTE WAS ON HIS SIDE everyone is blaming anthropic for the limits, so he decided to actually look at the data 858 sessions, 18,903 turns, and $1,619 estimated spend across 33 days here's what he found: 1\ one default setting was burning 14,000 tokens per turn Claude Code loads the full JSON schema for every tool into context at session start. whether you use them or not. 20,000 tokens of tool definitions sitting there on every single turn. the fix: one line in your settings.json "ENABLE_TOOL_SEARCH": "true" context dropped from 45K to 20K instantly. across 858 sessions that one setting was wasting an estimated 264 million tokens 2\ cache expiry is the single biggest waste 54% of his turns came after a 5+ minute idle gap. every one of those turns re-processed the entire conversation at full price which caused a 10x cost jump you go grab coffee. come back 5 minutes later. type your next message. everything rebuilds from scratch. the context didn't change. you didn't change. the cache just expired. 12.3 million tokens wasted on idle gaps alone 3\ 42 skills loaded. 19 of them used twice or less across 858 sessions. every one of those skill schemas sat in context on every turn eating tokens for nothing. 4\ 1,122 redundant file reads where the same file was read 3+ times one session read the same file 33 times. he ALSO built a full token auditor dashboard that shows you exactly where your waste is coming from 19 charts, opens in your browser, free AND open source
English
97
154
1.9K
287.6K