spankalee

Seriously. The CEO is opening posting white supremacist content like it's Stormfront. If you don't support that, you should get out.

Do you have someone who can babysit and review what the LLM does? Otherwise, I'm not sure we're at the point where you can just tell an agent to go off and build something and it does it _correctly_.

IME, you'll just get demoware if you don't have the time and attention to detail to really manage the process.

OT, but why in the world would you have your blog posts pop up in a little modal dialog in the middle of the screen and force readers to scroll more than they have to?

If it's using an LLM it'll make stuff up... about people and sex trafficking.

We're talking about servers here - the article specifically said that one of the requirements was no special _server_ software, and a web server almost certainly has zip (or tar) installed. These gwtar files don't work without a server apparently either.

I really don't understand why a zip file isn't a good solution here. Just because is requires "special" zip software on the server?

Wow, this looks potentially amazing!

I've been look for a way to use Git for smaller, high volume document storage - think something like Google docs where every doc is a repo and every change is a commit - but also, where the vast majority of docs age out and are effectively archived and not used, but still need to be available.

This looks like it technically, I just wonder how well the pricing scales for that case of docs that might never be read again...

Since a human can also be a "plagiarism machine" (it's a potential copyright violation for both me and an LLM alike to create images of Mickey Mouse for commercial uses) it'll matter exactly what the output is, won't it?

> the ones who are having a magic robot write a thousand lines of code that almost, but not quite, does something sensible

Gee, I wonder which "side" you're on?

It's not true that all AI generated code looks like it does the right thing but doesn't, or that all that human written code does the right thing.

The code itself matters here. So given code that works, is tested, and implements the features you need, what does it matter if it was completely written by a human, an LLM, or some combination?

Do you also have a problem with LLM-driven code completion? Or with LLM code reviews? LLM assisted tests?

Not really. One version of this might look like implementing agents and tools in WASM and running generated code in WASM, and gluing together many restricted fine-grained WASM components in a way that's safe but allows from high-level work. WASM provides the sandboxing, and you have a lot of sandboxes.

I think at least a few teams are working on information flow control systems for orchestrating secured agents with minimal permissions. It's a critical area to address if we really want agents out there doing arbitrary useful stuff for us, safely.

This is very, very wrong, IMO. We need more sandboxes and more granular sandboxes.

A VM is too coarse grained and doesn't know how to deal with sensitive data in a structured and secure way. Everything's just in the same big box.

You don't want to give a a single agent access to your email, calendar, bank, and the internet, but you may want to give an agent access to your calendar and not the general internet; another access to your credit card but nothing else; and then be able to glue them together securely to buy plane tickets.

SBA loans and other state funding...

> The topology shouldn't be hardcoded; it should emerge from the task's entropy

What does this even mean?

I'd be curious about situations where human drivers have caused the deaths of people who didn't sign the driver's TOS.

I'm not a C programmer - having coded in high-level languages only for the past 20 years - but I've been doing a lot of WASM recently, and eagerly looking forward to the stack switching proposal so I don't have to implement an asincify-type transform for an async/await feature.

If it's true that a C program doesn't have control of the stack, what does that mean for supporting the stack switching in Wastrel? Can you not reify the stack and replace it with another from a suspended async function? Do you need some kind of userland stack for all stacks once you support WASM stack switching?

Wireit does both DAG and content-addressing. It figerprints the inputs and outputs of dependencies. And you run scripts externally with plain `npm run` commands. It's really beautiful.

Wireit really is that. The script declares dependencies and input, Wireit caches based on the direct inputs and dependency outputs.

This happens with all agents I've used and package.json files for npm. Instead of using `npm i foo` the agent string-edits package.json and hallucinates some version to install. Usually it's a kind of ok version, but it's not how I would like this to work.

It's worse with renaming things in code. I've yet to see an agent be able to use refactoring tools (if they even exist in VS Code) instead of brute-forcing renames with string replacement or sed. Agents use edit -> build -> read errors -> repeat, instead of using a reliable tool, and it burns a lot more GPU...

Parallel and sequential, especially at the command level, are really the wrong abstractions for running scripts. If you have multiple packages, each with builds, there's a high chance you have dependencies and multiple packages depending on common ones.

What you really want is a way for scripts to describe their dependencies, and then the runner figures out what order to run them in, and cache scripts that don't need to be run because their inputs didn't change.

Wireit[1] is an npm script runner that adds that incrementally on top of package.json. I can't manage an npm monorepo without it now.

Deno started integrating the idea directly into their built-in script runner. I think this is an important enough feature that more runtimes should follow Deno's lead.

[1]: https://github.com/google/wireit

I'm working on what I hope is an AI-first language now, but I'm taking the opposite approach: something like Swift/DartTypeScript with plenty of high level constructs that compactly describe intent.

I'm focusing on very high-quality feedback from the compiler, and sandboxing via WASM to be able to safely iterate without human intervention - which Hoot has as well.

Large swaths of many European cities are like this. Copenhagen has a huge pedestrianized shopping area and it's amazing.

Good thing that this article is not arguing for the elimination of parking!

There are almost no Walmarts in urban centers.

Solar panels do not solve the problem of parking lots being community dead zones. You can put solar panels on anything - it'd be better if it were housing, a store, a pub, etc. than a parking lot.

Changing street parking prices is a lot easier than changing buildings built with previous parking requirements.

I'd say change the requirements first, then if there's a surge in street parking demand there will be natural pressure to raise prices.

I used to work for a competitor of NI that went out of business years ago. NI's polish, attention to detail, and focus on UX was always pretty extraordinary, I could just tell at the time they were going to crush us even when they were new. This is sad news :(

There is actually a huge difference between checking in all of your dependencies and checking in a lock-file. Some people work with hundreds of repositories on their local machine and checking in dependencies would lead to massive bloat. It really only works if you primarily work in a single monorepo.

Yeah, npm never has "version lock" where it can't figure out a valid solution to the version constraints.

This is mostly good, but version lock does encourage packages to accept wide ranges of dependencies, and to update their dependency ranges frequently, instead of just sitting there on old versions.

They had a lot more than metadata on Enrique Tarrio.