For anyone who is confused: This is exploiting an old soundness bug in the Rust compiler that is still present. The GitHub issue page has this comment from maintainers:

we already had a crate published on crates.io before which used this bug to transmute in safe code, see #25860 (comment).

this issue is a priority to fix for the types team and has been so for years now. there is a reason for why it is not yet fixed. fixing it relies on where-bounds on binders which are blocked on the next-generation trait solver. we are actively working on this and cannot fix the unsoundness before it’s done.

To be honest, you can say the same about any large cloud provider. What happens if AWS, or Azure, or Google Cloud go down, or become terrible?

I believe you are correct; if the unsafe code can cause undefined behavior if input data is not following a specific contract, then the entire function should be labeled unsafe so the caller knows that.

The other option is to check to make sure the contract is valid, and return an error or panic if it is not. That function would be sound, as no inputs cause undefined behavior.

Check out Ollama and its extensions for VSCode; might save you some money paying for other services if your computer can run models locally.

Unfortunately, I don’t have experience with mangohud. Does Legacy work without it? And does mangohud work with other games?

Just going to ask this just in case: have you tried doing a full update and reboot? If you updated and have not rebooted, sometimes drivers get messed up.

The link they use is working for me; what is the code you are using to fetch the data?

Also, dbg!() is a very useful macro for inspecting state. Might help see what is going on.

That link seems broken (the date is wrong). This worked for me:

https://this-week-in-rust.org/blog/2023/11/08/this-week-in-rust-520/

I haven’t taken it myself, but “The Last Algorithms Course You’ll Need” is free and is written by The Primeagen. He works at Netflix and runs a programming-focused YouTube channel, and as far as I can tell is very knowledgeable and level-headed.

Exactly; the idea is familiarity, not efficiency. To be fair, this argument doesn’t make sense for all situations, so it is possible I misunderstood what the original post was talking about specifically.

I think the behavior could actually make sense with real physics, as the vehicle might be designed to mimic what the driver expects rather than real physics. For example, my car often shuts off the engine when I am not accelerating because it is a hybrid. So, if I don’t press the gas pedal, it wouldn’t really make sense for it to move. However, it is designed to artificially engage the engine when none of the pedals are pressed to more closely mimic the behavior of non-hybrid cars.

If most pilots are used to the behavior if a vehicle in atmosphere, a space ship might be designed to mimic that behavior (through weak reverse thrusters or something else) to make it easier for pilots to get used to.

Just listing some things that I needed to do for working remotely on a personal device: have an antivirus installed, make sure Windows firewall is enabled, enable automatic updates, screensaver or lockscreen configured for 15 minutes of inactivity, and use a strong password (and a good password manager).

However, you can configure GRUB to use an encrypted boot partition, and even have detached encryption headers. It does take a bit more work, and you should make sure you know what you are doing. (e.g. losing a detached header basically means your data is all lost)

https://wiki.archlinux.org/title/Dm-crypt/Encrypting_an_entire_system#Encrypted_boot_partition_(GRUB)

Just installed, and now I’m wondering why I’ve never found this before. Its great - open source, well-designed, and pretty full-featured

For my (admittedly nonprofessional) use cases, I have found nothing that Krita, Gimp, or Inkscape could not handle. Honestly, I think the UI is equal or better as well.

I didn’t even know that, that’s pretty cool. I’ll have to try it later

The repository links to a list of examples, but the best way to learn is just to mess around with it. It is fairly intuitive to work with (especially if you have used another node-based UI before, like Blender).

The UI also has the ability to import/export your current setup (what I call a workflow) as a json file. If I get some time, I might share some of mine with a pastebin link or something.

The usefulness of ComfyUI is not just making one simple image. It is the ability to completely customize how that image is created.

For example, I have a workflow that generates a half-resolution preview image, then upscales the latent and puts it through two more sampling nodes. All three of the nodes have a different prompt input, with the focus slowly shifting to style instead of content.

I have also created a custom upscaling workflow, where the image is upscaled with normal upscaling, then re-encoded and put through just a few sampling steps, the re-encoded with a tiled VAE decoder (to save my VRAM). It creates much better results (more detail and control) than a direct ERSGAN upscale, and can even be put through ERSGAN afterward to get a super large image.

Wow, that makes so much more sense. It would be nice if the UI made that behavior clear, as just returning 404 errors generally means “this does not exist”, not “I might need to take a second to look for this”.