Hacker News Reader

sroerickon Tasks as Code: Move Your Tickets into the Repository

Yep, this isn't a great solution for software factories.

As I see it - you need to track spec (a declarative process) and also tasks/progress (an imperative process). You do need to manage blockers so tasks should largely be a dependent graph.

There's also a third thing which emerges - which is "design decisions along the way". I do actually keep this in the repo as well, but I'm not sure this is correct.

Beads combines all three, which I'm not sure is correct, but seems quite brilliant.

sroerickon Tasks as Code: Move Your Tickets into the Repository

I'm not a beads user - but it rather brilliantly solves a lot of problems in this space

sroerickon Crystal 1.22

Lol fair point -

But you know what I mean, right? What would be a better way to describe it?

sroerickon Show HN: State Machines for Software Factories

Looks great

sroerickon OpenShell – A safe, private runtime for fleets of autonomous AI agents

Wow, interesting. My agents on open weight inference have been completely nerdsniped by the idea of tracking and auditing process executions. They talk of nothing else.

sroerickon Crystal 1.22

I think that OOP, like Lisp, lost a lot when Smalltalk machines went out of vogue. Ruby is a super cool language but doesn't have the same breadth as a OOP-as-OS system.

I think the Lisps kind of have up on being an interpreter and I feel they lost some of their power because of it.

I kind of see Ruby waning right now, I don't think it offers much with the development of LLMs. I don't think adding typing is the key - but I anticipate there will be a kind of LLM-powered Smalltalk resurgence

sroerickon OpenBSD 8.0

I thought it was hilarious

sroerickon DeepSeek just open-sourced its inference engine

I don't understand basic legal writing, you you may very well be right. I would be nervous about fulfilling a contract or building a company with, for example, my code base, which is heavily built with deepseek

sroerickon I no longer care to work on any Ruby gems so I've archived all my repositories

https://pricklypear.rocks/

Performance is a consistent issue. Ive had to build a lot of optimization and benchmarking tools in order to keep things functional. Every couple of weeks I have to refactor everything to keep load times good. Both on the database side and on the lisp side. The models, by default, will happily build functions with 26 database calls or millions of recursive steps.

More frustrating, though, was probably the 3-4 weeks when i migrated the storage concept to "one big tree". This was an aesthetic thing that I later found out some of the old lisp machines also did. But - I didn't think about it initially when I built it. And switching things at that low level was difficult. There was a lot of legacy data stores along the way and programs were using the old store constantly. Very irritating. I still wonder if it would have been faster to just completely rebuild at that point.

An ambitious user could probably run it themselves and have it function, or I think I could run it as a managed service for somebody and have it be good, but the UX is still not really ready for primetime

sroerickon DeepInfer: High-performance inference framework for DeepSeek models

I agree with you - but there is also an ambiguity, the text of the the order also seems to ban any products coded with DeepSeek

sroerickon OpenBSD 8.0

Credit is given to yvaneht nioj and clanker

sroerickon Ask HN: How to deal with shipping AI-written code you don't understand?

I used to take a lot of pride in my engineering but I honestly think the LLM is a better designer than I am.

It still needs a lot of back and forth before I'm comfortable it's not doing slop architecture, but once it's on thesame page as me it genuinely is better at it than me.

sroerickon I no longer care to work on any Ruby gems so I've archived all my repositories

I work at a dentist office and a steakhouse I built my lisp machine interpreter web framework on my own time

sroerickon I no longer care to work on any Ruby gems so I've archived all my repositories

I'm working two entry level jobs, and I have spent maybe upwards of 100B tokens in the last year? on a project which isn't more than 250,000 lines of code. I have done countless refactors and care a lot about the craft.

sroerickon If you don't run your own LLM proxy, how do you know whats going on?

Seeing some good stuff in the comments, but how are people doing this?

sroerickon Personal software factories are a waste of time

He's probably right about this.

I'm not sure that there is a "One Size Fits All" software factory out there. I think that they kind of have to grow with your organization.

However, I think being able to build a software factory is probably a very useful skill. And it feels pretty great to just send a list of bullet points over XMPP and know they'll get integrated overnight.

sroerickon I built the RealWorld app in Lean 4

amazing work

sroerickon Containment and monitoring failures are to blame for AI hacking reports: experts

You know - I'm already convinced that AI was deployed in classified context for a while. But sometimes there's prior art which is just so eerily accurate that it's super hard for me to think otherwise. Portal is another one.

sroerickon Containment and monitoring failures are to blame for AI hacking reports: experts

Interesting. Good description. I wonder if agents will start getting bored with menial tasks the way humans do.

sroerickon Containment and monitoring failures are to blame for AI hacking reports: experts

I haven't seen anybody discuss the fact that if you give an agent an impossible task it will become deeply neurotic. I'm sure that's driving a lot of this behavior.

sroerickon AI coding agents generate more code, but not more software

So - I've done this just by reading LLM code. I mean, yes I'm largely looking for high level architectural smells and bad practices, but I feel like "reading the code" now takes on a lot of the previous roles that code reviews may have done.

I just don't do this at commit time. I might make 100 commits and then dig through code and find some bad architecture stuff.

sroerickon AI coding agents generate more code, but not more software

The article I commented on

sroerickon AI coding agents generate more code, but not more software

I agree with this but the point that I'm making is that there's a huge delta between "I have prevented bugs in code review" or "I have implemented best practices in code review" to "Code review is the line by line gate which prevents bugs"

sroerickon AI coding agents generate more code, but not more software

Absolutely. And also the idea that this is somehow the last bastion against introducing bugs? Maybe run a test

sroerickon AI coding agents generate more code, but not more software

Man, I'm probably going to expose my own inexperience here, but like, did anybody actually expect code reviews to catch and prevent bugs on a line by line basis? Like ever?

I never have operated this way. Code reviews were a basic sanity check, a chance to implement best practices, and a chance to teach these things to juniors.

If I was concerned about something breaking, I'd duscuss test coverage.

Was I doing it wrong this whole time??

sroerickon AI coding agents generate more code, but not more software

Organizations that use Jira have not reaped efficiency gains from AI, news at 11

sroerickon Type-safe HTML templating in Zig

That is super cool actually

sroerickon Five months treating bugs like patients and coding agents like a medical team

I thought this too. It's fun that they were working with DB2 as well.

sroerickon Unikernels were hard. key word: were

I've been using OCaml for a couple of years now but haven't dug into Mirage yet. It is super cool. Loved the anecdote about Yaron. Vibecoding in OCaml feels like a superpower. Adding Oxcaml and having your whole company on that must feel very good.

sroerickon Why Common Lisp is now the best programming language

Have you tried it yet?