splitbrain

I build something similar as a claude skill but less impressive and complete than yours. https://github.com/splitbrain/ndemo

I came to the same conclusion when I built my NAS. Just Nix for the system, zfs for raid and docker compose for any service I might want to run. https://www.splitbrain.org/blog/2025-08/03-diy_nas_on_nixos

> it’s giving a lecture about the problems you’re pondering in the shower; it’s thinking out loud about the ways in which your project doesn’t work at all.

I'd love to do this. But I have learned that my brain does not work this way. The moment I explain my shower thought project to anyone, I immediately lose all interest in actually building it. I don't know why.

If I want to succeed building a new thing, I can not talk to anyone before I have actually built the first fully working version of it.

Is that available somewhere? Maybe as a blog post on how you set this up?

A password hasher PWA (JavaScript) https://www.splitbrain.org/blog/2026-01/02-password_hasher_p...

A GUI for the notmuch email indexer (PHP) https://www.splitbrain.org/blog/2026-01/09-gmail_backup_notm...

An AppImage installer (Go) https://www.splitbrain.org/blog/2026-01/20-appy_an_appimage_...

All three just this month. Each of them took me several hours, because they're not vibe coded but carefully reviewed and refined until I was reasonably happy with the resulting code. Could I have written them myself? Sure. Would I ever have started without claude? Hell no.

  Learn how it works. -> This page is incomplete

A pity. It's obvious what the "latest" would be, but what is the best? How is that decided?

Shameless plug: for randomly discovering IndieBlogs check out https://indieblog.page/

All kinds of personal blogs are welcome at https://indieblog.page no matter your fame on Hackernews.

Shameless plug as always when the topic comes up: submit your blog to https://indieblog.page to be discovered. subscribe to its RSS feed or mastodon account to discover indie blogs one random post at a time.

Shameless plug: Submit your blog to https://indieblog.page and you'll get the occasional random reader who might even become a RSS subscriber.

Offtopic: several of the embedded Bluesky posts at the end of the article show "The author of the quoted post has requested their posts not be displayed on external sites." Seems not to phase the PC Gamer "journalists".

I wish there were any SBCs under 10 bucks. Remember the promise of the RPi Zero being 5$? That promise never came true.

Unfortunately it seems not to have any wireless connectivity.

I was in the same boat and built my own commenting system. It has an importer for disqus (so you don't lose your old comments) and also imports Mastodon replies.

https://www.splitbrain.org/blog/2025-03/26-meh_another_comme...

A few years ago there was an Ask HN thread asking people here to show off their personal sites (https://news.ycombinator.com/item?id=30934529). Reading that thread gave me the idea to build indieblog.page (previously discussed at https://news.ycombinator.com/item?id=31002171). The list of blogs grew to nearly 5000 in the last years.

Alex Schroeder suggested to add a search engine on top of people's feeds as a way to search the Indieweb without adding even more crawlers to our collective infrastructure. I think it's a pretty cool way to quickly see what others have written on any given topic.

I am using the original RM1 nearly daily for the last 8 years or so as my primary note taking device. I bought it used because it was ridiculously expensive new. I was grandfathered in when they introduced the subscription. I really love the device, but I would never buy it with connectivity locked behind a subscription.

No, the traffic is not caused by client requests (like when your chat gpt session does a search and checks some sources). They are caused by training runs. The difference is that AI companies are not storing the data they scrape. They let the model ingest the data, then throw it away. When they train the next model, they scrape the entire Internet again. At least that's how I understand it.

I built a NAS off them. They're in a ZFS mirror and important stuff is backed up offsite.

I recently bought two SSDs from Chinese brands I had never heard before. My thinking being that they were reasonably priced (a bit cheaper than name brands but not by much) and that probably no-one would fake no-name brands. Reviews seemed to be mostly genuine as well. ¯\_(ツ)_/¯ so far they seem to work fine, we'll see.

No Guybrush Threepwood joke in here yet?

> Embedding was a nightmare - any inherited CSS from parent pages (margin, padding, line-height) would shift alignment.

This seems to be the perfect use case for a web component and its shadow DOM. Instead of using a div.editor, this component could wrap around a textarea and it would progressively enhance the textarea experience.

> No accounts, no tracking, no ads

Your cookie banner says otherwise:

Analytics: Microsoft Clarity for usage insights

Error Monitoring: Sentry for bug detection and fixes

This sounds interesting. How big is your library of content? And did you do this for yourself or the whole family? If the latter, how does your family interact with it? Is it automatically running on a central TV? Or on their personal devices?

I think OP sums it up pretty well. If there isn't an existing module or one that does exactly what you want, your only way forward is to program your way out of it. That's just not something I want from my daily driver of a distribution. It doesn't help that this programming happens in a weird (to me) functional language.

I only got started with NixOS a couple of weeks ago and I must agree. I would never consider running Nix on my daily Desktop machine (I'm using Arch BTW). It's more work and hassle than it's worth it.

OTOH I just set it up as the base for my new DIY NAS. There it is limited to just the bare minimum of bringing the system up and providing some core services (including Samba). And for this I found Nix' declarative approach quite good. I can easily restore the root system from the backed up config alone.

Everything else will be handled by Docker compose stacks outside the Nix eco system (stored on the RAID).

https://www.splitbrain.org/blog/2025-08/03-diy_nas_on_nixos

Yeah, and if this can also automatically send mails, we not only have prompt injection but also data exfiltration.

I built a site to give you a starting point for discovery: https://indieblog.page

done

sources are on github, blog URLs can be downloaded.

I wouldn't want to overwhelm the results with a single platform. also I do check each submission manually so I want to keep it manageable.

both are answered on the faq page in detail: https://indieblog.page/faq

quick answer: php+sqlite and a hackernews thread. nowadays mostly submissions.