I’ve been slowly working on a set of decoupled services that could replace some aspects of GitHub.

https://pr.pico.sh/ — a pastebin supercharged for git collaboration.

https://pgit.pico.sh/ — static site generator for git repos.

Both are still WIP but I think they are pretty handy

If you want low effort high value then get a synology 2 bay. If you want full control over the host OS then run Debian/arch with zfs

This was a really great read. The numbers don’t lie: 7 companies each independently invested 100 bn in AI and have seen little revenue in return (relatively speaking). When overlaying that on top of its impact on the us stock market as a whole, once the veil has been lifted that this is not leading to super intelligence, a bubble will burst and the entire economy will feel it.

That doesn’t mean “nothing to see here, move on.” It means that AI isn’t the bow-wave of “impending superintelligence.” Nor is it going to deliver “humanlike intelligence.”

It’s a grab-bag of useful (sometimes very useful) tools that can sometimes make workers’ lives better, when workers get to decide how and when they’re used.

This is the part that is overlooked when discussing anti-ai hype: these tools are very useful, but they appear only useful to the skilled laborer wielding them, instead of the investor claim that they will be replaced.

I went down a similar path as you. The entire proxmox community argues making it an appliance with nothing extra installed on the host. But the second you need to share data — like a nas — the tooling is a huge pain. I couldn’t reliably find a solution that felt right.

So my solution was to make my nas a zfs pool on my host. Bind mounting works for CTs but not VMs which is an annoying feature asymmetry. So I decided to also install an nfs server that exposed my nas.

I know that’s not what you want but just wanted to share what I did.

The feature asymmetry between CTs and VMs basically made CTs not part of my orchestration.

Here’s my homelab journey: https://bower.sh/homelab

Basically, containers and GPU is annoying to deal with, GPU pass through to a VM is even more annoying. Most modern hobbyist GPUs also do not support splitting your GPU. At the end of the day, it’s a bunch of tinkering which is valuable if that’s your goal. I learned what I wanted, now I’m back to arch running everything with systemd and quadlet

It is being rewritten using swift

This is absolutely not true in my org. LLM use is ramping up where every SWE is using it to some capacity