Posted by david927 11/9/2025
Ask HN: What Are You Working On? (Nov 2025)
https://github.com/Fingel/gelly
I've been enjoying rebuilding my music collection from both old hard drives and ripping old CDs. Jellyfin is great but I wanted a native application focused on music, not video. Thus Gelly. It's been really fun to work on.
And the other goal is minimal dependencies. The only bootstrapping stage is a very very small core in Common Lisp + FSet but could also be replaced with other languages, and then using that subset to bootstrap the rest.
There is absolutely zero claim to be highly performant, it is more of an educational experience.
All of it is done via literate programming in org-mode. So far it's working pretty well, but will have to see how that approach works if the project grows.
But as the first stage bootstrap in CL is mostly done at this point, I have to hot-reload anyway.
At some point it might be nice to have my own REPL running in Emacs, but that is a worry for way later when I actually get something usable. For now this is purely for personal entertainment.
You just need to pay for a fixed monthly upfront cost rather than PAYG, giving small developers a good save of their money.
In other words, this is similar to self hosting with K0S/K3S/OpenShift, except you don't have to own servers to begin with, in other words, it is a little similar to serverless K8S.
Well, all you those you can actually do with a VPS today, heck why do I have to do it if EKS/GKE/LKE/OKE/DOKS exists? That's because it takes a lot of time to properly setup VPC/EBS/S3/EC2, you need to pay an insane amount of premium and overheads to those while an ordinary user just don't want to hassle too much.
I want to undercut the big clouds by saving people's money and time. I have had enough of seeing a ludicrous EKS billing. I just want K8S to be the control panel of everything.
Deploy, run and scale later, simple as that
Right now I'm hosting my own test cluster under my bed so I can't show it.
You don't have to manage CNI, CSI, Linux kernel, etcd. You just need Kubernetes app development knowledge and that's basically it.
Now I'm still thinking about how to get live migration and failover working, so it is going to take a painful while...Kata Container doesn't support it out of the box but Cloud Hypervisor does
I tried to submit it as a startup project last year but the feedback isn't great, I want to have something polished first before making it public
The main question is which P2P overlay network to use, if any: the prior incarnation used Pinecone (a variant of Yggdrasil), whereas this time we're pondering keeping it simpler and more scalable and using Matrix itself as the backbone to connect together smallish local P2P meshes - so by default you try to route via Matrix, but failing that you look on your LAN or BLE to see if you can talk directly to whoever you're addressing. Time will tell if this works :)
Edit: another option on our todo list to look into more is Iroh (https://github.com/n0-computer/iroh)
If you don't want to be banned, you're welcome to email hn@ycombinator.com and give us reason to believe that you'll follow the rules in the future. They're here: https://news.ycombinator.com/newsguidelines.html.
Kinda like this (https://github.com/open-telemetry/opentelemetry-injector) but with support of having multiple service name for different services. This includes tomcat, normal systemd services and also services running inside docker containers.
EDIT: I am popping my cherry with this comment on HN. Been a lurker since past 2-3 years.
FURS does for Forth, what headers do for C, namely provide all the embedded configuration information inside a Cortex-M MCU, for the up to 100 inbuilt peripherals.
Without this data, neither C nor Forth (or any other language) have any clue about how to use the peripherals.
FURS does this by intercepting the Forth user source as it's uploaded to the on-chip compiler and transforming it into language the MCU inherently understands.
The Forth user source code is not altered in any way.
I've used the Fossil DCVS for the entire FURS project so that all the flowcharts, pictures, code, user doc, trouble-ticket, wiki ... everything is contained in the ONE FILE, under 5MB.
This one file gives you a web server so all you need is a browser to easily view all the above from the main menu.
Howto: https://sourceforge.net/projects/mecrisp-stellaris-folkdoc/f...
The idea is to take boring components: PostgreSQL, Bifrost (LLM gateway), Open WebUI, LanceDB, Agentgateway (MCP and OpenAPI gateway) and deploy them in Fly.io. One Fly.io "org" per user. The closest equivalent is blaxel.ai, but it caters for AI SaaS startups, not individual customers.
The combination of the fact that Fly secrets are visible only from within the apps, distroless containers, and transparent data encryption for PostgreSQL assures that the service (Pocketdata) provider cannot access their data, only the infrastructure provider (Fly.io) theoretically can, but practically speaking, this gives an extremely high degree of privacy assurance.
The latest update on the project: https://engineeringideas.substack.com/p/tasklet-is-the-o1-mo...
I got all the components, tested it on a breadboard, learned to solder and now I'm working on the 3d Print to enclose everything.
I actually just did a test run to see if my current 3d design would fit my PICO board, and it fit, but not that secure yet.
Im a developer but never worked this close to metal, so I've been so happy with how it's been going so far, making me real proud of myself.