Mathieu

"Error: crypto.getRandomValues() not supported" minimax: let me add react-native-get-random-values to polyfill this for rn/expo. me: abort. no. let's use expo-crypto minimax: expo-crypto only has async getRandomBytesAsync. How would you like to proceed? me: nvm i'll ask claude claude: i've installed expo-crypto and we're now using randomUUID from it

@euacc isn't #7 (stock option taxes) not already included in #4 (eu inc)? > Moreover, it will provide a common optional scheme for employee stock options with harmonized deffered taxation, which will enable EU Inc. companies to attract the best talents. commission.europa.eu/topics/busines…

🇪🇺eu/acc After 2 years in existence, the first 4 points of the @euacc manifesto, which was crowdsourced by all of you, are now passed as laws That means 1/3rd of eu/acc's points is complete: ✅ 1. Reduce regulatory burden for startups ✅ 2. Make skilled immigration easier, unskilled harder ✅ 3. Repeal the cookie law ✅ 4. European Inc: a single pan-EU business entity Now for the next 8 objectives: 🔲 6. Tax discount during startup phase 🔲 7. Tax stock options when sold, not when exercised 🔲 8. Embrace AI and technology, don't fight it 🔲 9. Champion free speech, don't censor it 🔲 10. Reform bankruptcy laws to empower entrepreneurs 🔲 11. Make English the primary language of the European Union 🔲 12. Teach AI and tech in European schools and universities

so nice that i'm now getting german tweets because @nikitabier wants to suppress global and increase local reach. i know it's closer, but i'm not german. i don't live in germany. i mean, it affects me, but the united states do as well.

@jlongster can the sandboxes be pods on my own kubernetes cluster? 😀

OpenCode is about to get more powerful with remote sandboxes I showed a brief demo before, but here's a much more in-depth demo. it's not hard to add basic support for a remote env, but handling all the edge cases like when a remote env gets deleted is difficult. especially if care about good UX You never want to lose session data. so the choices are: run the session in your env, but run all tool calls remotely. that's too complex and painful. The other way is to just let the full session run remotely, but sync back all the session data in your env. We chose this path: we built a syncing system which logs all events in a way that we can always recreate your entire session. That means the remote env could get destroyed, but we can easily restore it. it also opens up other interesting ideas which we'll be exploring

i think the most important llm benchmark is this petergpt.github.io/bullshit-bench… the ability to push back makes all the difference. i'm trying minimax m2.7 since yesterday, but whatever other benchmarks say, it just feels stupid to me

And after voting that out too... 🇪🇺 ChatControl is back with a vengeance The Conservatives (EPP) are attempting to force a new vote TODAY (March 26) seeking to reverse the European Parliament's NO on indiscriminate scanning of ALL your private messages, emails and photos This is a direct attack on democracy and blatant disregard for your right to privacy If you're European contact your representatives now, with this handy form: #contact-tool" target="_blank" rel="nofollow noopener">fightchatcontrol.eu/?foo=bar#conta… If not, please share this tweet so more people see it and we can block the vote It's crazy they keep just bringing back whatever they want until it's passed! Obviously now we see the European Commission is controlled by powerful evil lobbying groups 👺

@jskalc @claudeai @AnthropicAI same here with the $200 sub

Last two days my @claudeai Max 100$ sub is being used in 1-2h. No changes to my workflow, previously it was enough for the whole day. @AnthropicAI lowered usage without telling anyone?

@levelsio what is this? why? first geo fencing and now also blocking based on popularity?

Okay and who of you couldn't reply to this? Please please only if you couldn't okay?

people i'm following: i really love ai. i can finally finish all my side projects me: omg i have 2483 side projects now

almost feels like they silently limit you if you're using it outside of their tos. last week i was experimenting with paperclip. this week i'm not doing anything crazy, but suddenly going through my usage fast.

the use i'm getting out of my @claudeai max 20x sub is down a lot since it got it. suddenly feels like i have to watch my usage, where first i didn't really go above 30%/40%.

This is fine.

@bunjavascript this is so fucking annoying. claude constantly convinced my tests are broken but it should just run `bun run test` instead of `bun test`. my tests don’t work with bun’s test runner. please respect my test script

What is my perfect crime? A developer types bun test. He meant bun run test. Do I throw an error? No. I run his tests. Built-in runner. They pass 10x faster. He thinks Vitest ran. It didn't. I ran. He types bun build. Same mistake. Same result. His bundle appears. He ships it. Months go by. He types bun server.tsx. Not a mistake this time. He's curious. It runs. JSX and all. He tells his team they're still a Node shop. Node does TypeScript now, he says. But not .tsx. I don't correct him. He never installed pg. Bun.sql is talking to Postgres. He never installed the AWS SDK. Bun.s3 is uploading files. He never chose a runtime. He mistyped a command six months ago and I never said a word.

@ivanburazin you can put your phone to your ear (like when calling) and whatsapp will switch to that speaker so you can listen in private

Voice messages should die. I hate everyone who sends them. When someone sends me a voice note, I don't look at it for days. It's the most selfish form of communication. You're offloading your typing time onto my listening time. You get convenience and I get locked into your pace. Can neither skim thru nor reference them later.

@thdxr i assumed projects were doing this mainly for easier slop PR detection, but that is not the reason?

this thing about open source projects making their tests private you're either open source or not - part of that is enabling a successor to show up for net benefit of the world making that intentionally difficult is fine but then it's something different

@josevalim i still miss my 12 mini :( my 16 pro just isn’t as nice to hold and use. although the battery and camera are really amazing

I guess all hope for an iPhone Mini is lost and I will have to settle for the 17e or something. 😔

@hardfire @danieltvela @alexocheema @Prince_Canuma @ivanfioravanti need to use mlx-vlm, or convert the original model yourself. mlx-lm doesn’t work for the community version as it doesn’t support vision

@danieltvela @alexocheema @Prince_Canuma @ivanfioravanti Mlx-community version fails on my m4 pro. Not sure why. Are you running it via mlx-lm ?

mlx-community/qwen3.5-35b-a3b runs at 83.87 tok/sec on my M4 Pro (14c). Impressive!!! But GGUF version by Qwen only runs at 35.45 tok/sec with the same prompt. Both using LM Studio. Anyone knows why? @alexocheema @Prince_Canuma @ivanfioravanti

@RhysSullivan just keep ghostty open. it will block shutdown. but what’s wrong with it? i’ve already installed it

i fear the day this update gets applied

i don't think this study is very good and doesn't reflect the best practices and semantics of how harnesses should properly leverage agent md files a one time /init created agent md is not good enough. hierarchal, automated generation of agent mds i the correct approach

opus 4.6 sometimes randomly stops in the middle of a task. very strange