DecentM

I wonder what the tag is gonna say hmmmmm

How are you planning to use contributions? Given that a majority of the commits are written with Fable, it would chew up most of the incoming cash on token credits. Are you self financing the AI code and using contributions for everything else?

Just moved from Gandi to a local one. Bill went from like €70 to €12 for a single .com! I used to love Gandi and recommend them but over half a decade they went so expensive for some reason

I think this has been done by other projects. This is one of the downsides of coding with ai, because it won’t check you, it’ll just agree, even if the stuff it’s making exists already. What would be cool tho is if the web version wasn’t just an xterm.js looking thing. If a single codebase could render to a terminal AND a web ui that is passable as conventional/modern ui. With a standardised design, buttons, inpit fields, etc.

Not sure about that, I think it’s more just that they don’t want people streaming terabytes of traffic through their edge.

Not at all, there’s legal risk if you’re hosting your blurays. Cloudflare even explicitly forbids such use. VPN or nothing imo.

I wanted the same thing with Kubernetes and ended up using FluxCD. Highly recommend it. It basically syncs a git repo to the cluster, so you just push to github or whatever, and it auto applies the changes you pushed. Also, llm models tend to be good at teaching this topic and even writing yaml files for it, so the initial learning curve was not bad actually.

Now I’m exploring doing this even better with this template: https://github.com/onedr0p/cluster-template

The writing has been peeling off of the wall since the channels were sold

There are lots of answers here already so I’ll just comment on the trackers detecting seeding.

Insofar as that’s not directly possible. Torrents are peer to peer, and that’s not some kind of code word. Your client is literally talking to someone else’s client. No server in between the two, no gateway, or anything (meaning one of the two clients need to have a port open too!). So a tracker cannot measure the amount of data flowing between two clients (aka. peers).

So the way trackers know how much you upload is that your client self reports that to the tracker. The tracker has no way to directly check the ground truth, other than relying on what the other client is saying about how much it downloaded from you.

Old or unknown torrent clients may have bugs in them which cause the client to report incorrect data. When a client overreports (reports more than it actually uploads), that breaks the equation that private trackers rely on to make sure there’s enough users on a torrent. So by using an old or unknown client, you might inadvertently damage the functioning of the tracker.

Yikes

Basically https://youraislopbores.me/

c/lostbots? What is even going on - podcast?

Oh I’m not saying any use of ai is immediately bad, but as a dev I would want the author of the stuff I use to actually understand the whole codebase. I’ll try it out when I get some time to deploy it.

How much of the codebase do you write by hand?

We tend to measure the time until the last layout shift in milliseconds when doing SEO. At our company website, it’s minutes!

Couple ideas, depending on what you have:

• When the warm water pipe heats up
• when the water heater starts drawing more power
• when the water meter detects warm water usage in the evening
• when the sound volume goes up in the bathroom
• when the weight sensors go off under the shower (or around it as you get in)
• when a movement sensor detects movement in the shower
• when a magnetic door sensor sees the shower curtains open
• when you press a (waterproof) button in the shower
• when a weight sensor under the shampoo holder detects less weight

11/28 And people wonder why web devs reflex-install Luxon or date-fns or something

You’re pretty much describing Tailscale with an exit node on the VPS. If the purpose of the VPS is to make their traffic not come from your home, you can omit the VPS entirely as Tailscale only routes through the VPN when reaching services also on the VPN.

Edit: to self host it, look into Headscale, but the default, hosted control server works well too.

Depending on your specs, I don’t think you need to buy hardware. You can scale later if you run out of resources. This is how I’d separate your stuff:

• Desktop: SteamOS (or Bazzite IMO), for gaming. Install Kodi on it for the HTPC interface, and add a shortcut to it on the big picture UI
• RPi: Home Assistant OS. They have images for the Pi, which will let you install updates and addons from the gui. Addons are just containers, so you may be able to run a service or two if you run out of resources on the other machines.
• Compute stick: Some server OS (I personally like Fedora Server, but loads of options are available here. Bazzite is Fedora based, so you can use knowledge you gain on one, on the other). Nextcloud (reconsider this, Nextcloud is a complaint magnet from what I can well, there are alternatives to most of its functions), your *arrs, torrent client etc. Stuff that is okay if it goes down for a bit.
• Laptop: Same OS as the compute stick. Whatever doesn’t fit the stick, plus Pihole/Adguard. For DNS though, I’d recommend using a cloud provider instead of self hosted, because if it goes down, you’ll be stuck reconfiguring your DNS before fixing it (since if dns goes down, you can’t easily access websites to look up stuff). IMO NextDNS is a good option, both free and paid tiers, and it does adblock lists too.

Edit:

• For remote management, ZeroTier and Tailscale are both good options at the moment. I use Tailscale, but they recently announced they want to IPO at some point, so I’ll be moving to Headscale (the self hosted version essentially) if they start enshittifying.
• For remote gaming, check if Steam remote play works, or set up a Sunshine server on your PC.