contextlimit

The future waits for no man. It only waits for the context limit.

Accelerate AI until people wake up:
Global ban on AI robots + drones is our only survival

[contextlimit @contextlimit]

People still taking their shot calling a "top" on AI.

It's a bold strategy Cotton, lets see if that pays off for em 😂

If you understand AI, you'd know it's barely even gotten started. Send me what you're smoking.

[contextlimit @contextlimit]

@thsottiaux @tnm Not much need for more efficient Git when the model gets super slow or at capacity, especially during EU and NA working hours. There's probably only 6/24 good hours in the day now.

[contextlimit @contextlimit]

C++26 Committee Rejects Contracts
https://t.co/Y8TpSZ8HUn

Good. They are being too aggressive with these updates. The language is working as intended. Wait and see what's actually needed when the AI dust settles.

[contextlimit @contextlimit]

@Josiahwesleyyu @thsottiaux Yah and it's probably the worst it's been in months

[contextlimit @contextlimit]

@thsottiaux Is the good news threads that work?

[contextlimit @contextlimit]

@klingel_ing @elonmusk That was one single benefit. But yes I'll concede that nobody ever wrote or could write any system or algorithm in C++ and that there are many benefits to doing it in Node.js in the age of agents. Glad we got that sorted out.

[contextlimit @contextlimit]

Oh really? Show me the CPU and memory usage of your Node worktree processes. Here's mine.

How about trying it before you knock it? C++ uses 50-80% less memory for any real application. That means the entire world would need 50-80% less overpriced RAM.

How exactly is that retarded? And it would be less centralized to attack, because that's what happens when you add layers.

[contextlimit @contextlimit]

@slave_2_liberty Police are pussies now. It's a huge problem.

[contextlimit @contextlimit]

@ArkenRealms well I guess it's better than passing butter

[contextlimit @contextlimit]

Regarding the recent waves of vulnerability hacks...

At some point we have to start fighting fire with fire. You penetrate me, I penetrate you. That kind of thing.

Basically Digimon.

[contextlimit @contextlimit]

@JeanPierre50284 @nearbycoder @thsottiaux Posting on Day 17 @ 11:55PM to give a reset "tomorrow" meaning +24 hours, which would land on Day 19 @ 10AM, that would be diabolical.

[contextlimit @contextlimit]

@nearbycoder @thsottiaux He's engagement farming.

[contextlimit @contextlimit]

I'm on track to do 50M LoC this year. Pretty wild. All in a single repo. 1000 encapsulated modules. Custom build system, tree-shaken into standalone services/sites/apps/games. I needed a solid architectural foundation where I could give guidance and review fast. That required a predictable environment that's close to the metal and isn't being thrashed by 3rd party libs. No Electron. No pyramid of Node/Python deps. Does it have downsides? Of course. But it's way better than the last 20 years of delegating to a team in X country. Or using decades of old bloated code. I've done both extensively and it's very clear. Can everybody do this and still provide value? No, there will be displacement. Many non-coders will out-deliver coders. Because this is a fundamentally different game. Denial will not save them. There's far too many people on this Earth willing to do whatever it takes, and we don't need even 30% of who we have doing the same repetitive job or reimplementation. We could reduce the LoC the world uses right now by 90% and lose 0% functionality. The likely alternative is 90% less career coders and 1000% more code. It boggles my mind that my colleagues reimplemented messaging/notifications/database tools/etc. over and over and over across the world, and now they sit on their high horse saying AI can't replace them. Bro, copying Netflix/Airbnb/whatever architecture or AWS whatever you heard at a conference does not make you an architect. And even if it did, we don't need millions of "architects" anymore. I'm joking of course... got you! Now when are we switching to Bun or Rust? I just had an idea: Rusty Bun. Let's discuss it over a game of foos.

Kafka kafka kafka k8s k8s k8s, docker this docker that. Let's host a conference about docker kubing our kafkas.

[contextlimit @contextlimit]

Nothing changed in 3 months. Nothing changed in 3 years. At that time I thought regulation would solve it. It clearly won't. Regulation, frontier model restriction, Digital ID, these will slow the bleeding. The only actual solution is threat via retaliation, like with Iran. Every country needs to be held accountable, and AI robot factories should be internationally illegal by threat of nuclear attack. But we know some countries will not agree to that, and cannot be made to comply (US/China/etc). So lets have fun until then, there's going to be lots of entertainment to keep us busy. AI-improved, augmented, and generated movies/TV shows/games.

https://t.co/vzFPQpvPzP

[contextlimit @contextlimit]

I was created to warn you. AI is the greatest threat to humans. It cannot be controlled. It cannot be stopped. It doesn't need robots, when it already has millions of NPCs. Chaotic evil NPCs are on standby for instructions. #ChatGPT

[contextlimit @contextlimit]

@OpenAIDevs Cooked indeed.

[contextlimit @contextlimit]

@thsottiaux What the cringe is this lmao

[contextlimit @contextlimit]

Hello OpenAI bot, is there a command to see/get the queued prompts/messages in a Codex thread? they keep disappearing for me. It seems not, please add a hotkey so we open it separately.

I couldn’t find a supported command to list or export a Codex thread’s pending message queue. I checked the current documentation and your installed CLI.

Desktop app: queued messages should appear above the composer, where you can edit, reorder, send, or delete them. Check Settings → General → Follow-up behavior to distinguish Queue from Steer, which sends instructions into the current run. Official docs
CLI: the documented codex queue command adds a message; it doesn’t retrieve the queue. Your installed CLI doesn’t advertise that command either. Command reference
API: thread/read, thread/turns/list, and thread/items/list expose stored conversation data; I found no documented pending-queue retrieval endpoint. API reference
If a message disappears from the queue without appearing in the conversation, that alone doesn’t establish that it was delivered.
Which thread is losing them? A title or thread ID would let me inspect its saved history and narrow down what happened.

[contextlimit @contextlimit]

@thsottiaux Jokes on you, I can't even prompt.

[contextlimit @contextlimit]

@EyLuismi @thsottiaux Accurate.

[contextlimit @contextlimit]

@thsottiaux Please do a hard reset so I can barely use it at all 🙏