The argument is maintainer bandwidth/fatigue.

From the closed issue:

> Closing this out for the time being, as noted previously (I'm currently leaning towards designing some review refinements and shelving the stacked concept design for the moment) and to clarify that it isn't current and active work.

https://codeberg.org/forgejo/design/pulls/48

Which is totally understandable.

Managing a healthy highly-popular open-source codebase requires effort to not bloat it.

Which brings me back to wanting good APIs for native Kubernetes CI runners and time-limited PATs for agentic coding.

I can vibe that in a day. But it sure as heck won't be aligned with the future of all Forgejo users.

Thanks for squinting and finding sense in how I said it. It's been 12 years since I took an econ course.

What I like about (the idea of) ForgeFed is that it lets existing forges speak to each other.

In practice I probably just need Forgejo and GitLab to be able to speak to each other.

I believe the future of GitHub, for me, is to solve two problems:

  - Discoverability for public open-source projects
  - Backup since self-hosting is fragile long-term

GitHub repos sit stale for decades.

For day-to-day reliance, my self-hosted Forgejo and CI runners have better uptime.

Only pet peeve with Forgejo:

  - It's a highly active project, RFCs, tons of PRs and issues.
  - Becoming a daily user, I want to extend it, and in its beautiful simplicity, it's not highly extensible.
  - So to avoid maintaining a fork of a very active project, extending it in unison is a social commitment.

I'd like to see more hosted Forgejo solutions pop up; it's very low-resource cost.

The aspiration is to distribute stdx in a way similar to libc and never rely on crates.io, bypassing supply-chain problems altogether.

What I do to compensate:

  - make it my duty to own every change, i.e. cognitively debt-free:
  - write summaries on every new thing I do (blog post, memo to colleague)
  - contribute documentation to the open-source projects I rely on
  - practice for CKA/CKAD certificates which require pre-LLM muscle memory
  - build interactive learning material for what I’m trying to learn
  - work with things that LLMs don’t yet trivially solve
  - repeat or reconstruct my recipes to perfect workflows, 

That's very imprecise.

There are multiple optima on the price/demand elasticity curve.

At the two ends: Few expensive items and many inexpensive items, Hetzner has consistently operated at low margins and massive volume.

Which means they charge customers as little as they can to get as many customers as possible.

You'll see that evident on any price comparison chart prior to these adjustments.

They're probably still cheaper than everyone else in spite of 3x'ing the price of some products.

I’d like to gently push back on that. ;-D

Terraform, when committed to git, provides organisational memory. But less so uniformity, since all providers are different (and you should expect different things when applying). No tracing besides git. And tfstate is hard to share between developers, unlike kube state.

Kubernetes is more the same across providers. And it manages drift after something is applied, which is not a direct argument of OP, but a strong reason over other IAC.

And yes, I also enjoy how well deploying works. And how things generally fit together. Liking the networking complexity less so.

Like when games detect aimbots, they don't ban people, but put them in an aimbot bracket, so everyone you play with is a cheater.

Provide a captcha that is essentially harder for a human to solve, but trivial for either a human or an AI, and transparently separate them into two communities.

As someone who has done and is doing Nix as part of my job: you just have to carve it out of reality.

I start with non-invasive stuff like a shell.nix. The farthest I got is dendritic kubenix. :-D

https://simonshine.dk/articles/three-levels-of-nix/

I'm not sure I could click an App Store link on your page; either I couldn't, or I could and the button wasn't visible enough.

Only to Sim City.

Feel free to submit bug reports. I actively use it on NixOS, Ubuntu WSL and nix-darwin.

Reading AI-generated text for hours every day, it's obvious to me.

I take care to make my messages easily readable. I don't care if they're AI-made, as long as they're short.

I'm a very verbose person, and if I don't make an effort at being concise, I'm just as annoying as the average AI.

Being flooded with AI text every day has made me appreciate brevity because I'm exposed to so little of it.

With half a dozen people who don't read or listen to half of what the others do, slop + cognitive drift is a bad cocktail.

It's just not as big of a problem on my own projects, because the ideas that get fed to the slop-machine are not that different from one day to the next.

---

> For human code review requests, I always review my AI-generated code first.

For human code review requests, I always review ANY code I submit first.

This is partly because it's the agreed-upon culture where I work now.

And partly because the codebase is not robust enough for slop.

I have hobby projects where this does not apply. I spend half of my time in those projects building hard guardrails.

---

> Keeping AI generated content clearly labeled and demonstrating human effort helps show consideration for teammates

I actually like the shamelessness, because it's honest.

So often this year when I ask "why did you do X?" pointing at a line, my colleague doesn't know.

Because they didn't really write that line, and they didn't really internalise the choices made.

When my colleague sends me a text dump from Claude, I know that my role is just being a sub-agent.

Demonstrating human effort: I'd like to see more of it.

One way is to spend more time owning "cognitive debt" as part of the daily cycle.

You could say Google has two purposes of funding Firefox:

One is to inject search engine and ad distribution dominance into every known browser.

The other is to prop up Firefox as an alternative browser to avoid monopoly lawsuits.

They're not sponsoring Firefox out of goodness.

Which is why I don't like to be charitable with the interpretation.

It's Talos Linux on Hetzner VMs where all resources are managed via kubenix, and all CI runners are running NixOS.

So... extremely affordable and extremely performant, but very complicated.

Drop me an email if you want the writeup when I get around to making it.

Until then, here's a Forgejo Actions that compiles its own CI runner image:

https://git.shine.town/infra/runners/src/branch/main/.forgej...

No Docker involved in the build process, it's all Nix.

Unfortunately the CI runner itself is still Docker-in-Docker because:

https://codeberg.org/forgejo/discussions/issues/66#issuecomm...

  - Directing people to Google Search means Firefox users get exposed to ads
  - The money given to Firefox was made selling ads
  - Google is an ad company