SavvyWolf

Skipped to the “ugly” part of the article and I kind of agree with the language being hard?

I think a bigger problem is that it’s hard to find “best practices” because information is just scattered everywhere and search engines are terrible.

Like, the language itself is fairly simple and the tutorial is good. But it’s a struggle when it comes to doing things like “how do I change the source of a package”, “how do I compose two modules together” and “how do I add a repo to a flake so it’s visible in my config”. Most of this information comes from random discourse threads where the responder assumes you have a working knowledge of the part of the codebase they’re taking about.

It’s relatively trivial, you just need to write a kernel module. You’d just need/want to make it gpl so everything it does is fully audited and transparent. That’s not a problem, is it? Right?

From a technical standpoint, you could argue that someone could create a fork of the kernel that spoofs the interface that the anticheat uses to make it ignore things. You can, of course, also do something similar in Windows, but security theatre never let practicality get in the way.

Daily backups. Then you can have as much wild ambition as you like. Disk failures do not care for your permissions bits anyway.

Practically though, one thing I find that’s a good habit to get into is to use rmdir on directories that you know should be empty instead of rm -rf. If you’ve made a mistake and try to delete the wrong folder, it’ll error out.

Ah. If this is a new build that suggests a hardware issue. Can you try running with only one stick of RAM and/or xmp disabled to see if that fixes things?

What do you mean by “green screen of death”? Is there any text or something you can transcribe or screenshot for us?

I meant in the Passwords and Keys app below, sorry should have specified!

If you want to keep autologin enabled, you can also just set the password to nothing (that is, when you’re setting the password, just press enter without typing anything) to disable the password.

Saves having to mess with the .desktop files.

Does it mention something about “your login key was not unlocked when you logged in”?

If so, open up the application called “Passwords and Keys” (or “seahorse”), right click on “Login” > Change password and set it to match your Linux user account.

> Claims to support xdg.

> Stores config in ~/.matchctl.conf

Yet another dating app that gives me trust issues. ;_;

MIT license, for those wondering.

https://en.m.wikipedia.org/wiki/Magic_SysRq_key

Alt+SysRq+C, although your distro may have it disabled by default.

A fair warning though, safety is relative and crashing the kernel can be destructive. Make sure you have backups when breaking things.

Ubuntu back in the Gnome 2 days.

For my main desktop I use Mint because it just works, widely supported and Cinnamon is good (sadly no Wayland yet. ;_;). I also use Home-manager for my configuration because it allows me to easily just specify my config as a set of files I can check into git.

For my server, I use NixOS, because having all my configuration in a few text files is very nice to get an overview of what my server is doing.

It’s the directory that needs to be writable to delete files, not the file itself.

Although the immutable bit (if that’s what you’re talking about - I thought you meant unsetting the write bit) might change that, I’m not sure.

The home directory would need to be immutable, not bashrc.

I don’t think that actually works; the attacker could just remove .bashrc and create a new file with the same name.

In an ideal world, a search engine will point to this thread, where the answer is the topvoted comment.

With the death of Stackoverflow and Reddit, hopefully Lemmy can fill the void of an information archive. :P

What improvements are you thinking of? I can see that reasoning with something like the Linux kernel where there’s a lot of complex and integrated code, but ultimately individual coreutils commands are really simple. There’s very little you can do to extend something like ls… And if you do, you can just make your own superls command and not have to deal with any licensing restrictions.

With regards to AGPL vs GPL, none of the coreutils programs have network connectivity, so I’m not sure what the network requirement actually adds?

getting rid of the gpl is the motivation behind e.g. companies sponsoring clang/llvm so hard right now.

Is it? As I understand it, LLVM is much easier to work with than GCC, especially given their LLVM IR and passes frameworks.

here, take my stuff and don’t contribute anything back, that’s totally fine

I mean, yeah? They are probably fine with that and think that software should be distributed without restrictions. You may not agree with it, but it’s their choice. Not really stealing if they give it away willingly.

I cannot fathom any FOSS project not using the AGPL anymore.

I mean, most of them that want to use a GPL-like license use the GPL or LGPL rather than the AGPL. :P

why are developers even agreeing to this?

Are they? Last I checked this wasn’t as much of a plan as much of it was just a developer thinking out loud. And even if it was a real plan, developers should continue doing what they should be doing anyway: Write their scripts without any GNU/uutils/whatever-microsoft-calls-their-evil-uutils-fork extensions. Then their scripts could run across all platforms, including GNU, uutils, FreeBSD and BusyBox.

At any rate, if Microsoft really wanted to make their own coreutils fork (if they haven’t already), they’re not really that complicated tools. They could devote like maybe a year of engineering time and get it pretty much compatible.