I’m using Silverblue and it also symlinks to /var/mnt. I don’t think it does that on traditional distros, like Fedora 40 Workstation.

Not aware of any correct pictures, but I can tell you what’s wrong with this one

• /usr: explaining it as “Unix System Resources” is a bit vague
• /bin: /bin is usually a symlink to /usr/bin
• /sbin: /sbin is usually a symlink to /usr/sbin, distros like Fedora are also looking into merging sbin into bin
• /opt: many, I’d say most, “add-on applications” put themselves in bin
• /media: /media is usually a symlink to /run/media, also weird to mention CD-ROMs when flash drives and other forms of storage get mounted here by default
• /mnt: i would disagree about the temporary part, as I mentioned before, stuff like flash drives are usually mounted in /run/media by default
• /root: the root user is usually not enabled on home systems
• /lib: /lib is usually a symlink to /usr/lib

I would also like the mention that the FHS standard wasn’t designed to be elegant, well thought out system. It mainly documents how the filesystem has been traditionally laid out. I forget which folder(s), but once a new folder has been made just because the main hard drive in a developer’s system filled up so they created a new folder named something different on a secondary hard drive.

I don’t get why this sort of picture always gets posted and upvoted when it’s wrong for most distros nowadays.

My drive was brand new when the issue started. I don’t think SMART showed anything wrong with it, apart maybe from the improper shutdowns count.

Not sure if it was Linux only, I never had Windows installed on that drive.

There’s nothing technical stopping Google from sending the prompt text (and maybe generated results) back to their servers. Only political/social backlash for worsened privacy.

I experienced that failed run shutdown binary a lot, the issue was that the OS I installed the drive on was defective. In use, the entire filesystem would become read only, the OS would freak out, and shutting down would fail with that message.

LLMs are expensive to run, so locally running them saves Google money.

Are the default policies good though? There’s some collaboration between Fedora and Tumbleweed for SELinux policies, I imagine even more now since Tumbleweed plans to move to SELinux in the near future and derivatives like Aeon are already using SELinux.

Tried it on Gnome, didn’t look the greatest. The numbers in the time were really close to the colon in the top panel. Very well could just be a Gnome issue though, the way it handles fonts is weird.

The sidebar is present, but no tabs just yet. The only way I was able to activate the sidebar was changing the chatbot from settings, not sure how to toggle it without doing that.

There is a way for just your home folder to be encrypted, Linux Mint has it as an option.

Fedora Silverblue is great, it’s my daily driver.

I can sleep “sleep”. All system components are still powered on at this stage, so it uses the most power. But at the same time it’s the quickest to get back into your system. All that’s really happening with sleep is that the screen turns off.

Then you have suspend. Laptops often first go to sleep but then suspend after a long period of inactivity to save battery.

Then you have hibernation. I don’t think this is used that often nowadays.

That’s true for hibernation, but not suspending. Hibernation stores everything in RAM onto the disk then shuts off the PC; to resume the system, you need to unlock the disk to access that data. Suspending doesn’t turn off the computer, it keeps the CPU and RAM active.

On my Fedora system, I can hit the suspend button and get back into the OS without needing to type my encryption password, only my user password.

With an encrypted disk, you only need to enter the encryption password when you shutdown or restart. Suspending and sleep lock screen don’t need your encryption password.

Snake case.

• Starts with a lowercase, good for shell autocompletion
• No spaces, so no worrying about spaces in shell commands
• ‘_’ is better than ‘-’ because it shows the spaces between words more clearly

On the kernel side, there are disagreements between long term C maintainers (who may not know Rust or may actively dislike it) and the new Rust community trying to build in Rust support. To make the Rust parts work, there needs to be good communication and cooperation between them to ensure that the Rust stuff doesn’t break.

On the Debian side, they have strict policies that conflict with how Rust development works. Rust has a dependency system called Cargo which hosts dependencies for Rust projects. This is different from C, C++ where there really isn’t a centralized build system or dependency hoster, you actually install a lot of dependencies for these languages from your distro’s repos. So if your Rust app is built against up to date libraries in Cargo, it’s going to be difficult to package those apps in Debian when they ship stable, out of date libraries since Debian’s policies don’t like the idea of using outside dependencies from Cargo.

Triple buffering is only active if the GPU isn’t keeping up with double buffering. So it will mainly only be active for lower powered devices, like older integrated GPUs.

Yup. I like their just in December approach too. I have a problem with distrohopping so I’m often re-setting up my system. Every time I do, Thunderbird pops up donation prompts both in the app and in my browser. I get why they do it, but it’s annoying when that happens. KDE’s approach avoids this pitfall.

Most package managers do not touch your home directory, so they will not delete user data. That needs to be done manually.

Snap and flatpak are exceptions, with an optional argument they will also delete the app’s folder (~/snap/appName for snap, ~/.var/app/flatpakID for flatpak).