CGamesPlay

Throw a loop/island in the maze and see how your kid reacts.

I think it would be nice if the effect happened on hover as well, maybe to a lesser degree. Also, some of the controls don't appear to have any jelly in them, like the progress bar, toast, and so on. Finally, the performance on that showcase page isn't the best, and it feels like you're hijacking the scrolling somehow. But overall, it's cute. Would fit nicely as a game's UI.

The product of software engineering (or computer science) is clarity and understanding.

Hard disagree. Software engineering is about satisfying requirements while optimizing costs. Computer science is the part where clarity and understanding are the goal.

"A normal DM is one pi session by default." A single message, or a single message thread? Both are pretty extreme trade-offs to make, honestly. I spent a bunch of time looking at how the various Claws do this, and I settled on an idle timeout causing the next message to start a new session, but I also built /continue, /new, and a /resume that shows recent sessions with a telegram keyboard.

There is contenteditable, which is what all these (Wordgard, ProseMirror) are fundamentally built on. The rest is just the UI, and interop with systems that don't desire arbitrary HTML as input.

I'm trying to come to terms with this issue in my own interactions with open source as well. Where I'm at currently: since code is cheap and analysis is expensive, it can be more beneficial to a project's maintainer to get a precise, well-researched report of the issue than a PR. This is an inversion of most of my open source life, where opening an issue was asking for free work while giving a PR implied more reciprocal effort was given. I'll typically just end with "PR available upon request", unless the project has a no-LLM policy.

Midway through https://www.chinatalk.media/p/how-to-buy-cheap-claude-tokens... which was the article from the top-level comment.

I mean, which lawyer caste do you respect? Is that one is cool with stealing credit cards to buy Claude subscriptions?

3. At an Italian airport: Constantly stealing bags, opening them to pick out MacBooks and credit cards, a credit card manufacturer-who sells stolen "black" credit card info to transfer stations— is racking his brains to save you money.

I just started referring to those coworkers as Claude directly. Not like, at the start of the message, but after the first line? Nobody’s noticed yet.

If they find another success half as good as Claude Code it will all be worthwhile. (Monetarily good, not like, quality good)

Isn't the first section no-longer accurate for several years? I understood that, while we serialize the end of turn markers in a text format like `</think>`, internally they are a dedicated token that cannot be forged (a user message containing `</think>` would encode to a different sequence of tokens). Am I mistaken about this?

Obviously, this doesn't really affect the results of the paper, but it feels like it's the obvious first-line of defense: at least the model has a solid fence between the different roles.

Can somebody help me understand the Quantization Analysis? It says "dynamic 4-bit UD-Q4_K_XL and dynamic 5-bit UD-Q5_K_XL are generally lossless" while showing a top-1% token agreement on the chart of 97.5%. Not what I would consider "generally lossless". Is this implying that some post-processing is going to account for the 2.5% loss? Beam search?

Perhaps most people in the future won't bother. They'll just let agents do it all.

But can you imagine the look on some young teen’s face when they train their own GPT on their local computer for the first time?

It doesn't seem to actually differentiate at all. If I move my iPhone around on a plane (table), the dots move. If I hold in in my hand and move left/right, the dots move. If I hold it in my hand and raise/lower my arm, the dots don't move. That's actually just an integration of acceleration/gyro, and possibly combined with a simple model of how a phone is held (e.g. assume rotations happen from a point 30cm away from the bottom of the phone).

Don't forget to count your EBS and egress fees when you make your AWS spreadsheet.

I doubt it has an "age of sail" mod, but there's this one: https://store.steampowered.com/app/794860/eSail_Sailing_Simu...

Nice! I am still looking for the best AI integration for my setup. Currently I don't have any interaction between Forgejo and my coding agent. I experimented with a Forgejo Actions runner, but the problem that I had was there's not a great way to manage the context there: you get what's in the issue or PR, but it gets muddy once you have multiple rounds and/or discussion moves from the issue to the PR.

Is there a widely-used open modern chat network? Specifically, I'm fine with the feature set of IRC, but I want durable messages and a mobile client.

Speaking as someone who hasn't run their own bouncer in 10+ years.

I think its killer use case is actually embedded in non-web places. Tree Sitter parsers require arbitrary programs to be able to parse arbitrary languages. WebAssembly is a natural way to achieve that: write your parser in any language, compile to WebAssembly, use that result in any supported editor. You get sandboxed execution and arbitrary compute.

It has to compete with more domain-adapted use cases though. Does WASM make more sense than eBPF for packet filtering? It doesn't seem to make more sense than JavaScript for making websites. Maybe it makes more sense for deploying edge services (which IIUC is the main use case for WASI).

I don't think this is an east-coast/west-coast thing, but I think people all over the USA use the word "park" to mean anything on the scale of corner playground to national wilderness area.

Enshrine it in actual law. Theodore Roosevelt donated some of his land when the national parks became law, and that’s held up reasonably well. There’s no such thing as a guarantee, but it is pretty decent precedence.

The announcement elucidated this, and it's IMO worse than this. They don't downgrade to a cheaper model ([edit] for certain classes of offense they suspect you of). They sabotage the model's outputs in other, undisclosed, ways (specifically, "prompt modification, steering vectors, or parameter-efficient fine-tuning"). So, for example, they might load in a steering vector that just forgets the API to PyTorch. But it isn't just "we redirected you to a cheaper model!"

It's unstated, but I'm willing to assume that only the root package.json is consulted to decide if these scripts are allowed. Otherwise, yes, this would not actually change anything.

Our biggest perf/resource gain is dynamic memory, which reduces memory usage a lot by releasing unused memory back to macOS. Nothing else supports this, including Containerization.

Wow, missed this when reviewing OrbStack. I assumed that you just used Containerization and therefore would have the same limitation.

If you can express that operation in Terraform, then Coder would let you do that. First problems I can think of are connectivity from the Coder provisioner to your local machine (Tailscale? Local?), and migrating disk images if you want to actually switch a workspace between environments (local provisioner could do this, but no matter what it’ll be slow and janky).

Not quite, it’s still a VM. And while it supports virtio balloon for growing RAM, it doesn’t yet support releasing that RAM back to the host. And there isn’t a convenient way to shrink the sparse disk images as they grow yet, either.

After Terry Bisson's "They're Made Out of Meat".

https://www.eastoftheweb.com/short-stories/UBooks/TheyMade.s...

I'm using https://coder.com for all my development containers. I've got mine hooked up to a k8s cluster, but anything that you can provision with Terraform can be used (e.g. docker containers).

That's a terrible distinction to make on a topic about how the coding agent gained root inadvertently.

I'll accept that it shouldn't be default, but just because your web app runs in rootless docker does not mean that root docker has no place. There are several limitations: https://docs.docker.com/engine/security/rootless/troubleshoo...