setVeryLoud(true); @ isVeryLoud @lemmy.ca

He / They

---

Read more

Posts

41

Comments

1,575

Joined

2 yr. ago

I've been looking at K3s deployed on FCOS, but I have no clue how I'm supposed to use Terraform to deploy FCOS.

My understanding is that FCOS is supposed to be ephemeral and re-deployed every so often, which would imply the use of a hypervisor like Proxmox on the host, but Proxmox does not play well with Terraform.

I also considered OpenStack, but it's way over my head. I have a very simple single-node Kubernetes setup to deploy using GitOps, and nothing seems to fit the bill.

The piss stain colour palette confirms it.

Both, it's a dual-purpose item!

Where my bitchat

In case of armed road rage, remember that your car is a weapon. It's self-defense if a gun is drawn on you.

Ubuntu Core, based on Snaps, is very much not ready for prime time IMO. It's kind of a mess outside of server use.

Look instead at Fedora Silverblue, Vanilla OS, and for the bleeding edge of immutable systems, GNOME OS.

KDE is about to launch their analogue to GNOME OS relatively shortly, named "Project Banana". These two are not exactly distros as they do not distribute the kernel, they are simply platforms that layer a bunch of images together to create a stable, reproducible system. There's also OpenSuSE Aeon, but I don't like its style of immutability as it's immutable by rootfs lock-out rather than immutable by image.

As for advice, learn how to use Distrobox / Toolbx containers. If you're a developer, this is where you will be working.

Immutable Linux is still young, and a lot of software isn't written with it in mind, so expect some growing pains.

I'm running an immutable distro at the moment (GNOME OS), and I felt no loss of performance due to Flatpaks. Snaps, on the other hand, do have a perceivably longer launch time.

Given that it's an immutable distro, everything I need needs to be either a Flatpak, a Snap, an Appimage or an extracted tarball, otherwise it runs in a container. The advantage of this system is stability and making the host incorruptible, as well as the ability to very easily roll back updates or failed systemd-sysext layers.

Not everything can run in a Flatpak at the moment, but we're hoping the evolution in Flatpak, XDG portals as well as encouraging developers to use the available XDG portals can make this a possibility someday. Namely, IDEs don't run that well in a Flatpak, but GNOME Builder has proven that it's 100% possible with the currently available XDG portals as well as connecting your IDE or editor to a container.

I've actually been discussing the idea of Flatpaks offering "terminal aliases", similar to what Snaps do, with some people involved in Flatpak. It's something that could happen in the future, but for now, you can totally create an alias to run a Flatpak from a single word, it's just a PITA.

Flatpaks aim to be a middle ground between dependency hell and "let's pull in the universe" bloat.

Applications packaged as Flatpaks can reference runtimes to share "bases" with other applications, and then provide their own libraries if they need anything bespoke on top of that.

I think they mean the guy you're describing, that sounds like a red flag. Sex is a conversation, not an assumption of likes and desires.

What can we do to pressure the government into updating its travel advisory to the states?

Germany and the UK literally have a higher degree of caution listed, which is kinda insane.

Fitting username

I don't have a degree, so I never peaked 🥳

"leopards ate your face? deserved."

"leopards ate my face, I want my vote back!!!"

Get back to me when it federates with Friendica, but I'm 0% interested in this side of social media, personally.

What an odd car!

That sounds so pleasant! Here, you go to Rona / Lowe's, you ask them a question and you're met with an "iunnodude". Maybe home hardware is comparable.

Story time?

To be clear, I'm perfectly ok with ethically trained (open-source weights and data set) generative AI being used locally on a small scale, I think generative AI is a double-edged technology like anything else.

It should never be the end product, but simply a tool.

In this picture here, you can see the skeleton is weird and other images and text is a bit wonky, these elements should have been touched up by a human. This is what I consider slop, raw AI output has this look and feel to it that makes it immediately identifiable, it is up to the artist to touch it up and adjust colours. Again, it should never be the final product. Something as simple as text should probably have been created normally.

I'm against Meta, Google, OpenAI, Anthropic, MidJourney, etc.'s use of generative AI for the reasons stated above, but small scale genAI on your local device? Go for it.

I find genAI imagery extremely uncanny and creepy, and I can't condone the usage of a system whose creators yearn for a day where companies won't have to pay human creators anymore and can simply funnel their funds directly into the pockets of giant corporations instead.

Additionally, commercial-scale generative AI is already destroying the environment in communities across the world due to its power use.

Also plagiarism: https://spectrum.ieee.org/amp/midjourney-copyright-2666872100

Commercial image generation models are trained on real people's copyrighted works with no regard for paying them for using said work, putting more money in large corporations' pockets.

It's not something I can accept nor condone, and I will continue to shame people for facilitating the transfer of wealth and destruction of our environment.

I thought we could do better on Lemmy than to lean on large, polluting technology companies for content, those same companies who have military deals to create civilian-killing robots.