Hacker Newsnew | past | comments | ask | show | jobs | submit | igorbark's commentslogin

the main thing i use it for that seems like a pain with other coding agents is my sandboxing workflow: i have a custom extension which allows the pi process to run on my laptop, keeping all my transcripts in one searchable place, while delegating all the bash and file system commands to a VM.

i haven't gotten around to it yet, but i'd also like to change how `Bash` functions on a basic level (and subagents), where rather than the agent picking a fixed timeout, it just gets notified with exponential backoff about commands that aren't done and then gets a turn to decide what to do with it.

having said that, i find it annoying and not empowering that basic things like subagents and web search aren't built in. the ideal to me seems to be an agent with polished extensions for all the common use cases that can be disabled if you do want to rewrite them. but i put up with the pain because i really want my pet feature ¯\_(ツ)_/¯


im not sure that there is a plug n play set up that will work for everyone, because as with any security boundary, each layer of hardening has a usability trade-off. i definitely feel you about the uncertainty of it all, how do you actually know everything is tight?

personally, i think either a VM or microVM is the way to go. these things are actually designed as security boundaries, as opposed to containers. and as compared to bubblewrap, you can just give the agent a whole FS to work with and run it in yolo mode, whereas with bubblewrap you have to manually bootstrap the availability of each individual dev tool and make sure its config dirs and package caches and etc are mounted in a secure way and still will probably hit perm errors all the time. and there's just way less isolation.

also, something that has limited support in harnesses but IMO would make a lot of sense is running the harness process in the host, but having all the tool calls and file system interactions delegated to the VM. that way you keep all your session data and auth keys on the main machine where it can never get into context. otoh it makes your harness part of the security boundary, so that's the trade-off.

there's also all the usability questions around how to actually get data in/out of the VM. i have a script which can push local git repos into the VM and then pull from them as a remote, so the VM can't initiate any connection with the host doesn't need to hold git credentials. but ig for someone who wants their agent to push straight to GitHub that's a waste of effort.

options i've tried or seen for the VM itself: - qemu + libvirt: takes some doing to wrangle it together, but very battle tested and configurable - crun-vm is a PoC of higher level integration layer between podman and qemu, which is a really cool way to go about it. seems maybe abandoned but i just think it's neat and very existing tools/standards oriented rather than starting a new project and brand so i mention - libkrun is a newer entrant, and several ppl have built wrappers around it: - microsandbox - smolvm (posted/discussed on here recently) - krunvm

this is all Linux oriented, it's all i know.


this logic does not hold up if the reason that labor is more expensive than parts is that the labor involved in creating those parts has been outsourced to a "shithole country"


If anything the causality is exactly the opposite. The labor cost will go up (empirically provable) in such "shithole countries" once work is outsourced to them, improving their livelihood.


if we're going to plug random puzzle games on this thread, IMO the most underdiscussed puzzle game of all time is Recursed, the only game i've played which properly explores recursion as a mechanic, and (MINOR SPOILER) the only game i've played which detects when you have created logical contradictions, and for each contradiction you achieve rewards you with a secret bonus level.


Recursed is really fantastic. There's puzzles in there that took a few days for me to solve


If we're talking about recursion, Patrick's Parabox is another stellar pick.


i definitely agree that it would've been nice to have images in the book as it was hard to get a sense of exactly how well Boquila was mimicking neighbouring plants!

but in reference to the linked article, i will say that the researchers interviewed in the book (and i got that sense for Zoe as well) were in agreement with you that the research didn't support a vision-based mechanism. but everyone agrees that the imitation is going on. the researchers in the book suggest a gene transfer-based mechanism instead! (mentioned briefly in your linked article)


not sure if you will still see this 7 days later, but the claim "we work more hours than medieval peasants did" jumped out at me so i looked into a bit and am curious if you have more thoughts on it!

i found this lively criticism of the video on reddit: https://old.reddit.com/r/badhistory/comments/16y233q/histori....

my brief takeaway was that the claim might be true if "work" means "working for an employer for wages", but not if "work" includes "necessary labor for shelter, food, clothing, survival".

but it's an interesting thought though so i'm curious if you have other related resources to dig into.


i enjoy that i cannot tell whether this is written by an AI, or by a human pretending to be an AI. my guess is human pretender!


I too have had to use Next but didn't want to feel locked into Vercel.

this is the biggest effort I'm aware to run Next with a full-ish feature set outside of Vercel: https://opennext.js.org/. Supports AWS, Cloudflare, and Netlify. You can also run Next as a normal node webserver. I've only used the Cloudflare integration and it was a bit janky but worked (and seems to be entering 1.0 soon so may be less janky).

AFAIK this is completely unsupported by the Next team, but would love to be proven wrong!


(I'm on the Next.js team) We're working with Cloudflare on an official adapter.

https://github.com/vercel/next.js/discussions/77740


that's great to hear, thank you for the update :)


Next works great as a vanilla node app. What features are you missing from Vercel's platform out of curiosity?


the biggest headache i had in particular was different ways of handling environment variables, but the different adapters at OpenNext have had a rolling list of caveats/unsupported features for as long as i've been following the project so i didn't want to outright say "full". hopefully the effort on Next's side to build a standardized adapter API will help with this!


Ah, fair. That is an area we struggled with. We run next in kubernetes (GKE) and really wanted to be able to promote the Docker/OCI image up through our environments (ie: not bake environment variables into the build artifact). We settled on having them dumped in a meta tag in the root document, and then have some helper functions to access those. We still have to use process.env for any secrets though, which makes them inaccessible to the frontend code (but that's sort of the point).


you don't need to send 100kb+ of JS over the wire to build a static site in react: for example https://vike.dev supports static HTML-only output for a site built with React.

as for "why React", speaking just for myself it's really nice to just have one tool that can do everything (static HTML-only, static with JS, SPA, SSR) and not have to context switch or potentially even have to split my site into two projects just because I want to hop between one or the other approach. and React has the biggest mindshare and ecosystem.


I think you just proved the point by introducing yet another frontend framework to learn.

And you absolutely don't want one tool to do everything. HTML/CSS is native and understanding it is a requirement for React. It also doesn't require Node and a build step.


I think all software engineers in the world who know HTML/CSS (like who doesn't?) beg to differ

Really funny how some devs think they know the secrets of engineering simplicity and everyone else is a fool for not knowing what they know (HTML/CSS).


Read this thread. There are some comments about using React components for everything and never touching HTML tags and CSS. And they probably call themselves senior frontend engineers.


If not React it's just another abstraction (including whatever you come up with) that is arbitrary and shitty in a different way. Front-end/mobile is so boring and unimportant that it absolutely makes sense to just pick a thing and use it everywhere, and save your brain capacity for interesting problem solving. I say this as someone who's been doing front end for 10 years.


If frontend is boring and unimportant, what is exciting and important? saving records in a database? UX is what makes money and it is done through frontend.


Client side is only interesting in complex generic reactive programming problems, state machines, domain modeling in the client, local first patterns. The part about browser or mobile platform intricacies is very lame.


Coding and algorithms are just some of the tools available for frontend and software engineering. UX is another layer that most software engineers can't grasp, it involves art, sociology, politics. Calling it boring and unimportant when you are not good at it is a misunderstanding of software's true purpose.


No one even mentioned agorithm. If you are really user focus as you claim to be, you should love cross platform, focus on delivering values and not be a simp for these "native" tribal knowledge. Talking about misunderstanding software's true purpose.


From talking shit about frontend to love cross platform it looks like you want to hate something but can't formulate well what's the problem.


Obviously I love to reduce front-end to generic problem solving and abstract the platform aways which contradicts directly with your amateurish "looks ma how minimal and artisan my html & css are".


> Obviously I love to reduce front-end to generic problem solving and abstract the platform aways

And I'm sure you do it as well as other delusional backend-only coders


> HTML/CSS is native and understanding it is a requirement for React.

It's not necessary and the ROI is poor IMO, especially for CSS. Better to just use React for everything and not worry about the implementation details; yes those details will sometimes help you debug, just like knowing machine code will occasionally help you debug your compiled code, but it's not something to put a lot of your learning budget into.

> It also doesn't require Node and a build step.

The point is assuming you already have node and a build step. You already know React. So just use it for everything. Everyone who thinks they can "just write HTML/CSS" ends up introducing Hugo or Gatsby or whatnot (or, worse, writing their own "simple" Makefiles and shell scripts) and gradually adding more and more features until it's just as complex as React. Just use React.


> It's not necessary and the ROI is poor IMO, especially for CSS. Better to just use React for everything and not worry about the implementation details; yes those details will sometimes help you debug, just like knowing machine code will occasionally help you debug your compiled code, but it's not something to put a lot of your learning budget into.

...what? What are you writing in React if not HTML tags and CSS classes? How can you even write anything in React without that? React doesn't do anything CSS specific either, so I don't understand how you'd even style anything if HTML and CSS were not a "necessity."


> What are you writing in React if not HTML tags and CSS classes?

UIs made of React components. I try to avoid having to know or care about the underlying implementation details.


> I try to avoid having to know or care about the underlying implementation details.

This feels quite ignorant. I'm not sure, in some ways I understand it, for example, when running software on the JVM, I typically don't think about how it's implemented, so I share some of that ignorance.

At the same time, specializing to just knowing one technology (e.g. React) and not caring about up or down the stack can also be limiting - both for any career moves (Vue? Angular? working with how the resources are packaged and deployed?) as well as just debugging things in more detail (since one can feasibly imagine cases where you need to look at both the output HTML/CSS/JS to understand why certain things are happening, not just a React plugin in DevTools).

You don't have to learn things that are of no value right now, but you probably shouldn't go out of your way to not learn things. Staying curious is generally pretty nice!


I mean, sure, I do actually know a bunch of the details and it's occasionally useful for debugging. But I look for ways to work at the React level if possible (e.g. using the React dev tools to debug the component hierarchy rather than using the raw browser dev tools where possible). IME it's better to fully specialise in a small set of technologies than to dabble across a wider range.


And what are in those components? Divs with a className property I assume? Or are you literally only using components other people have made in UI libraries with no changes of your own?


I generally use preexisting components, there's not a lot of value add in creating my own date picker or dropdown box or whatever. Obviously I put my own content in, and compose together existing components to make bigger new components if the component I want doesn't exist (e.g. if I couldn't find an address entry component I might make one out of individual text entry components). I think some of the components I use might actually be or correspond to "basic" HTML tags, but that doesn't make any difference at the point of use.

If a component doesn't have built-in styling support (i.e. configurable via React properties) then I'll use Tailwind so I can at least ignore all the selectors and cascading nonsense of CSS. (I remain hopeful that the world will eventually see that inline styling is the way to go; with Tailwind I meet them halfway)


Okay, well you do realize you're not the average programmer right? I'd even say this is junior level if you are not really creating your own code and components, so it's not accurate to say that HTML and CSS are not "necessary" unless you're doing what seems to be very basic work. And Tailwind is still CSS, again, so without knowing what things like flex actually do, you can't use Tailwind either. I'm beginning to suspect you may just use AI to build much of your code for you.


If your company has a good component library, you can really stay in that circle for 95% of the time. If you need that 5%, usually there's a library unless you're doing something really bespoke like canvas/webgl.

And AI really can do a lot of work nowadays. I met a founder who built their entire mvp with v0.dev. 5 pages, multiple flows, modals with photo taking/uploading and working color picker. They only needed to add the api-calling logic.


> I'd even say this is junior level if you are not really creating your own code and components

Must be a case of the midwit meme, or https://cspages.ucalgary.ca/~robin/class/449/Evolution.htm . Intermediate programmers satisfy their ego by writing everything themselves; a real experienced professional knows that custom code or a complex technology stack is a liability, and delivers the functionality the business needs in the way that will impose the smallest maintenance burden.

> it's not accurate to say that HTML and CSS are not "necessary" unless you're doing what seems to be very basic work.

Disagree. You need to be able to lay things out, compose components, and hook up behaviour. But you don't need to use the lowest level languages to do that, any more than you need to write all your programs in machine code. You need to understand the DOM, but that doesn't mean you need to express it in HTML, and similarly for layout.

> Tailwind is still CSS, again, so without knowing what things like flex actually do, you can't use Tailwind either.

You need to understand layout and styling properties, sure. But you can skip the things that make CSS CSS - selectors, cascading, and the like - which the industry is ever so gradually starting to recognise for the failures they are.

> I'm beginning to suspect you may just use AI to build much of your code for you.

LOL. Is this projection, or just reaching for any possible way to smear me? The AI fanboys are happy to mix a dozen technologies in their project because the AI "understands" all of them equally well - mix React and CSS and HTMX and what have you, why not? I'm actually genuinely curious about your thought process here (if you had one), can you talk through what lead you to this bizarre idea?


> you don't need to send 100kb+ of JS over the wire to build a static site in react

It is telling that the blog of the fw creator ships 500kb of JS/CSS/HTML to display text on a screen.

> it's really nice to just have one tool that can do everything (static HTML-only, static with JS, SPA, SSR)

WebComponents (+lit-html)


Pretty much none of that is blocking first paint. Disable JS and see that my site works just fine. I have interactive examples on some pages but all the extra stuff is just “nice-to-have”.

Also I’m not a framework creator.


Yep. 411 KB of JS (126 KB minified). 525 KB of fonts! 40KB of HTML. 1.1MB (704 KB minified) total for approximately 4.3 KB of actual text.

That's a crap:content ratio of 256:1


Disable JS and see that my site loads just fine. Fonts are nice-to-have. Interactive examples (on other pages) are nice-to-have. None of this is render-blocking.


It's still half a megabyte with JS turned off. For page of text. With no images.


Wow this is a great metric to use.


probably adding a confirmation message the first time the alias is used for each command would be good, it would be nice to know when i'm invoking git and when i'm invoking a third party binary regardless of any exploit attempts!


If malicious code ends up in your $PATH you have much bigger problems than git having a seamless plugin architecture.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: