want someone to prove his LLM can be as insightful and accurate as paid one.

The full DeepSeek model is available for download, and should generate about the same quality answers as the official one, with the bonus of less censorship. I pretty trivially got it to talk about the Tiananmen Square, and they can’t even ban me for it.

That said, that’s rarely the point. It’s usually because you can, a cost saving measure, sometimes you plainly just don’t need a good model, sometimes you want privacy, sometimes you need privacy at the cost of quality.

If your business is shoving customer reviews into a model, you really don’t need the best model for it to tell you how angry the customer is.

Personally I just do it for fun and because I can. Sometimes you just do things for no other reason than because you can.

Another reason to use ~/.local is you can do things like

./configure --prefix=$HOME/.local
make -j$(ncpu)
make install

And then you get your .local/bin, .local/share, .local/include, .local/lib and such, just like /usr but scoped to your user.

and it should mostly just work as well.

That was 7 years ago, and he seems to have distanced himself from that past. He’s kind of retired from the whole gaming channel thing and does more family life things.

People can grow a lot in 7 years, I sure did.

I really like the positive vibe and “here’s what you can do with Linux, for funsies” instead of the usual “here’s all the problems I had and I switched back”.

No “it’s perfect”, no “it runs all my games”, just “I tried it and had a blast setting it all up”. He’s legit enjoying it and sharing those feelings is powerful.

It’s shaping up to be pretty good at least. It’s pretty good for being in alpha state still.

The main thing it needs to beat for me is Kwin’s excellent Wayland support. Everything just works.

The per-screen workspaces are appealing though.

For what it’s worth, I experience none of that. My laptop is absolutely rock solid with KDE, it’s like a MacBook you pull it out of your backpack and it’s ready to go before I’m even done opening the screen.

My desktop is currently just over 5 days of continuous uptime (no sleep). I’ve crashed more often because of ZFS than KDE.

Both are ArchLinux. I also have a friend on Bazzite that doesn’t have issues with KDE either, and it runs great in my VM.

Those all sound like possible graphics driver issues.

In that specific context I was still thinking about how you need to run mysql_upgrade after an update, not the regular post upgrade scripts. And Arch does keep those relatively simple. As I said, Arch won’t restart your database for you, and also won’t run mysql_upgrade because it also doesn’t preconfigure a user for itself to do that. And it also doesn’t initialize /var/lib/mysql for you either upon installation. Arch only does maintenance tasks like rebuild your font cache, create system users, reload systemd. And if those scripts fail, it just moves on, it’s your job to read the log and fix it. It doesn’t fail the package installation, it just tells you to go figure it out yourself.

Debian distros will bounce your database and run the upgrade script for you, and if you use unattended upgrades it’ll even randomly bounce in the middle of the night because it pull a critical security update that probably don’t apply to you anyway. It’ll bail out mid dist-upgrade and leave you completely fucked, because it couldn’t restart a fucking database. It’s infuriating, I’ve even managed to get apt to be incapable of deleting a package (or reinstalling it)/because it wanted to run a pre-remove script that I had corrupted in a crash. Apt completely hosed, dpkg completely hosed, it was a pain in the ass.

With the Arch philosophy I still need to fix my database, but at least the rest of my system gets updated perfectly and I can still use pacman to install the tools I need to fix the damn database. I have all those issues with Debian because apt tries to do way too fucking much for its own good.

The Arch philosophy works. I can have that automated, if I asked for it and set up a hook for it.

Pacman just does a lot less work than apt, which keeps things simpler and more straightforward.

Pacman is as close as it gets to just untar’ing the package to your system. It does have some install scripts but they do the bare minimum needed.

Comparatively, Debian does a whole lot more under the hood. It’s got a whole configuration management thing that generates config files and stuff, which is all stuff that can go wrong especially if you overwrote it. Debian just assumes apt can log into your MySQL database for example, to update your tables after updating MySQL. If any of it goes wrong, the package is considered to have failed to install and you get stuck in a weird dependency hell. Pacman does nothing and assumes nothing, its only job is to put the files in the right place. If you want it to start, you start it. If you want to run post-upgrade, you got to do it yourself.

Thus you can yank an Arch system 5 years into the future and if your configs are still valid or default, it just works. It’s technically doable with apt too but just so much more fragile. My Debian updates always fail because NGINX isn’t happy, Apache isn’t happy, MySQL isn’t happy, and that just results in apt getting real unhappy and stuck. And AFAIK there’s no easy way to gaslight it into thinking the package installed fine either.

Yeah that’s a pretty good point. As a technical user that seems solid but for the average user that makes sense.

Isn’t owning the domain proof enough already?

Nobody else could possibly use max-p.me as their handle, and proving control of the domain is plenty for security sensitive things like LetsEncrypt.

Anyone you’d care to mark verified already brought their own domain.

You can’t really easily locate where the last version of the file is located on an append-only media without writing the index in a footer somewhere, and even then if you’re trying to pull an older version you’d still need to traverse the whole media.

That said, you use ZFS, so you can literally just zfs send it. ZFS will already know everything that needs to be known, so it’ll be a perfect incremental. But you’d definitely need to restore the entire dataset to pull anything out of it, reapply every incremental one by one, and if just one is unreadable the whole pool is unrecoverable, but so would the tar incrementals. But it’ll be as perfect and efficient as possible, as ZFS knows the exact change set it needs to bundle up. It’s unidirectional, so that’s why you can just zfs send into a file and burn it to a CD.

Since ZFS can easily tell you the difference between two snapshots, it also wouldn’t be too hard to make a Python script that writes the full new version of changed files and catalogs what file and what version is on which disc, for a more random access pattern.

But really for Blurays I think I’d just do it the old fashioned way and classify it to fit on a disc and label it with what’s on it, and if I update it make a v2 of it on the next disc.

Both use Linux under the hood. You can even install LineageOS on some TVs.

The only reason AndroidTV is bullshit is the manufacturers because casual users want shit like Netflix and Prime preinstalled. Google TV in particular comes with a lot of crap and the ads, which believe it or not some users take as a feature.

But that’s not inherent to Android TV as an OS, it’s exactly like Android phones and manufacturers preloading a bunch of crap to make an extra buck. If your run AOSP you get none of that crap, and it’s fully open-source.

I would distrust my carrier well before I distrust the encryption. Even when roaming, your Internet is tunnelled through your carrier using an internal VPN. It even works in China, that’s a fairly common way to get around their firewall.

I also wanted to put an emphasis on how working with virtual disks is very much the same as real ones. Same well known utilities to copy partitions work perfectly fine. Same cgdisk/parted and dd dance as you otherwise would.

Technically if you install the arch-install-scripts package on your host, you can even install ArchLinux into a VM exactly as if you were in archiso with the comfort of your desktop environment and browser. Straight up pacstrap it directly into the virtual disk.

Even crazier is, NBD (Network Block Device) is generic so it’s not even limited to disk images. You can forward a whole ass drive from another computer over WiFi and do what you need on it, even pass it to a VM boot it up.

With enough fuckery you could even wrap the partition in a fake partition table and boot the VM off the actual partition and make it bootable by both the host and the VM at the same time.

Yeah, that’s enough to not have it exposed directly. I understand why they did it that way but very good to know, thanks!

What you’re trying to do is called a P2V (Physical to Virtual). You want to directly copy the partition as going through a file share via Linux will definitely strip some metadata Windows wants on those files.

First, make a disk image that’s big enough to hold the whole partition and 1-2 GB extra for the ESP:

qemu-img create -f qcow2 YourDiskImageName.qcow2 300G

Then you can make the image behave like a real disk using qemu-nbd:

sudo modprobe nbd
sudo qemu-nbd -c /dev/nbd0 YourDiskImageName.qcow2

At this point, the disk image behaves like any other disk at /dev/nbd0.

From there create a partition table, you can use cgdisk or parted or even the GUI GParted will work on it.

And finally, copy the partition over with dd:

sudo dd if=/dev/sdb3 of=/dev/nbd0p2 bs=4M status=progress

You can also copy the ESP/boot partition as well so the bootloader works.

Finally once you’re done with the disk image, unload it:

sudo qemu-nbd -d /dev/nbd0

I keep hearing claims that it’s not secure enough to be exposed on the Internet, but I can’t seem to find anything about unauthenticated vulnerabilities. It’s got a fair amount of CVEs but they all seem to affect when you’re an already authenticated user, mainly to XSS an admin as a regular user or the likes.

It’s written in C#, and publicly all you can do is pretty much attempt to log in, this feels like it should be pretty sane compared to some other PHP crap I run.

Do you have any examples of previous exploits or anything else to be concerned about?

deleted by creator

I think it counts. You always have the option of taking your data with you and go elsewhere which is one of the main points of self-hosting, being in control of your data. If they jack up the prices or whatever, you just pack up, you never have to pay or else.

Also hosting an email server at home would be an absolute nightmare, took me 10+ years to get that IP rep and I’m holding on to it as long as I can.

I have a mix of it: private services run at home, public ones run on a bare metal server I rent. I still get the full benefits of having my own NextCloud and all. Ultimately even at home, I’d still be renting an Internet connection, unless you have a local only server.

It’s literally been working just fine for like a decade? Even for NVIDIA users that’s kind of a stretch.

Maybe if you share more details about your issues and your setup we can help fix it.