sysV is the init system linux distributions used before systemd, openrc, upstart, runit, smf etc. It’s pretty much the old daddy and comes from Linux unix roots. Even MacOS used it before they made their own called launchd.
S6 sounds like a update to it since the capital V in sysv stands for the Roman numeral 5.
I didn’t spot that, thank you.
Great, a subscription based mail program. Because that’s clearly what people want and need, paying rent for the software on their machines.
Using ptyxis even on KDE, it’s neat. Very clean and some interesting integration with distrobox, definitely recommend.
Huh, didn’t even know my country had a sovereign tech fund. Looking more into it … yeah. It gets money from the federal government but it is in no way run or even associated with it. Looks like a GmbH is behind it, which is a for profit company in Germany. It has a volume of 17 million €.
Also its name is literally sovereign tech fund, even in German, I.e. that’s not a translation, that’s its literal name. I wouldn’t say it’s sketchy, the people behind it definitely look legit, but it certainly doesn’t quite meet the lofty associations the name suggests.
I get visits multiple times a month because a Eula or something shows up on phones or tablets asking them to accept whatever. You know, the stuff nobody reads and just clicks accept on. Old people sometimes have trouble discerning where messages come from, anything from a pop up to a disclaimer for an update is “my phone asking me something”.
Chrome OS is still corporate BS. They will manage to confuse people with legalese. I have elder neighbours come to me confused that were literal pages into that BS. I always tell them they can just accept everything google/microsoft/apple asks of them, but that’s the problem, they can’t tell where it’s coming from. To them it’s just a legal contract showing up when they wanted to read their mails, it’s scary and they rather want me to check everything is ok.
Aurora is better for them. No legalese pop ups, fully automatic updates(no “click to accept”, “when do you want to schedule the update” or even an info that an update happened). I just tell them to make sure they turn off their PC at night and they will boot into the update next morning, never being the wiser. It just works.
Careful with gnome, it works for people that are unfamiliar with gnome, if you use extensions. Then gnome updates, the extensions break and you have to come over do maintenance, maybe use a different extensions which will look slightly different, throwing them off.
And yes I’m talking from experience. It’s bad because there is no easy fix for a broken extension, you essentially have to wait till the author fixes it. It was dock to dash in my example.
For old people you need point and click and the thing they have to click on should be permanently visible. They often do not understand the logical separations with virtual desktops and things showing up on screen only sometimes can be confusing. It’s easier if they have an area that’s always the same.
Installing things on fedora atomic spins is hardly tinkering more than necessary. You either layer the package, install it in a distrobox, use something like homebrew that installs packages outside of /usr, use app images, nix package manager, docker/podman or a flatpak.
These things exist for a reason, because they complement many distros that would be otherwise lacking. They can add a new app to a stable Debian, a stable dev environment to a bleeding edge arch, an isolated environment to use a untrusted app. If you use Linux these days you should be aware of these distribution agnostic options, or you will have issues understanding what is even going on and limit yourself unnecessarily.
I remember so odd not so transparent stuff not working well when I tried it years ago, like throttling of cpu/gpu, fan control, sleep modes, stuff like that. Provided your hardware is fully supported though I think they provide an excellent experience.
Maybe I should try to rebase my home server to freebsd …
How is HDR on the BSDs? On Linux it works nicely with plasma6 and wayland.
Eh, the way you phrased that I think it’s either fedora or opensuse. The up-to-date criteria basically knocks out everything with a fixed release cycle besides fedora which is pretty bleeding edge since they update certain things like kernel between releases.
Some criteria are non-sensical though imho. Ease of use? Speed? They are all the same, sure pacman works faster than zypper but it’s not like I’m waiting for either, they work in the background while I do stuff. As for ease of use … kde is kde, terminal is terminal. I think you would have to branch into the realm of the BSDs to get real differences there.
Debian is really solid, prefer it over any of its derivatives. You being unable to install something in kinoite is just lack of research on your part, ofc you’re going to have issues with a distro if you don’t know how to perform the most basic stuff. Stay far away from nix if kinoite gave you issues, with nix 90% of your pre-existing Linux knowledge will not only be useless but actively harmful.
Reading between the lines I think opensuse tumbleweed might work for you. Stable, powerful installer, very up to date and most of your pre-existing knowledge should transfer. Fedora is nice but you mentioned the magic word production, I don’t like fast cyclers in production, major version updates are a hassle at the best of times.
Ublue provides a really solid base template that makes it very easy to do stuff like this, all the heavy lifting of a distro is done by fedora, ublue adds the codecs and drivers and build system for updates.
Wouldn’t be unreasonable to make your own distro with it honestly.
Nice overview, don’t want to contest anything you said, just add my 2 cents to it.
The truth is the lines have gotten awfully blurry the past decade. It’s not just about FHS(basically a standard how the file system is laid out, where binaries go etc) getting more or less phased out(there are dozens of places where binaries can end up these days for example) but also some deeper changes of how we run software on these distributions.
Frankly arch and cachyos(which is a arch variant, yes it has optimised packages, but so does opensuse, it’s just a decision to leave behind compatibility with older hardware, not some inherent magic) belong into the standard Linux distro bin for me, they do nothing special or noteworthy beyond being competently implemented. They are not really different from Debian, fedora or Ubuntu, you can install those just the same manual way you do arch. Yes arch is rolling and others are release based but opensuse shows how there isn’t that much difference between the models, they run both on the same package base, as did Debian with its sid repo since forever.
Then we have gentoo and yes it is a bit special even today. The idea behind it is that you compile your own packages instead of using a binary repo. But why? The answer to that is that when you compile a package from source you have a vast influence on the resulting binary, for example by giving instruction to the compiler, that’s how cachyos gets its optimised binaries. But another even larger influence is by using configuration options built into the package by its developers. For example to disable or enable certain parts of the code. What gentoo did was collect and categorise the most common of these options into what became known as use flags, a system configuration that affects every single package built on that system. If you add the -dvd use flag it will strip dvd support from any package that has it. Or maybe you don’t have a printer -cups will remove cups support from all packages. This doesn’t just not install cups, it removes the very support of cups from packages that would otherwise interact/look for it in some way. This has obvious security advantages and is where the notion of gentoo being a lean system comes from, you’re stripping out entire functions of code from your binaries. If there is a bug in a certain OpenSSL mode that’s included in all binaries shipped by other distros, but you have deactivated all modes besides the one you intend to use, you are not affected by the bug. The idea behind gentoo is a kind of customisation that goes beyond the package layer, you’re no longer just choosing your individual packages but also the options of these packages.
As for the others, immutable, declaratives, cow or a/b root distros… that’s where the lines are getting blurry. The declarative like Nixos are very different in their implementation, but then again, you can use the nix package manager on other distros and we have been using docker containers set up declaratively via compose files for years by now. Likewise the immutable seem very alien, until you realise that they are only divided from their normal counterparts by a very thin line, important yes, but thin nonetheless. There is a reason these new distros get spearheaded by the old guard, a fedora workstation distro is very similar to a fedora silverblue immutable and from opensuse tumbleweed it’s a very close step to opensuse microOS. It’s mostly different default packages and some config options with an added package or two. Sure they seem very different, but just because you bolt the hood of your car closed doesn’t fundamentally change it does it?
These days I’d say gentoo is for learning. Not just about Linux but about interacting with source code and learning about the individual software you choose to install. The optimisations frankly matter less these days, sure you can optimise for speed or size of the binaries but are you going to be able to tell on that 12 core machine with 2 TB of nvme storage? No, not really. Security through a lean system might be nice, but there are specialised distros that already do so and you can run software in their own namespaces, control them via SElinux, put them in jails, bubblewrap them, containerise or VM them, hell even flatpak them, all probably more effective ways of archiving security.
There is literally a gentoo-install: https://github.com/oddlama/gentoo-install
Eh, plenty of dos machines still used in banks and industry. It’s both scary and impressive. I have worked on cnc machines only a few years back that were from the 80s I think. The data transfer between the computer and the machine used a band of paper that had holes punched into it by a printer like device physically attached to the computer.
Eh, building anything modern on a system that old would be painful I bet.
Maybe you could use https://github.com/openssh/openssh-portable since that’s meant to be portable. I’d certainly would give it a try if I didn’t want to bother trying to upgrade that system. Then again, trying to upgrade it through the releases to a modern Debian might be fun too.
If you are worried that the NSA might be reading your email maybe it’ll be better for society if you don’t update … just saying.
The amount of dos systems I have seen powering critical infrastructure in banks and hospitals is quite frankly nightmare fuel.
Nobody is going to rewrite the entire kernel in rust. Parts of it are still written in assembler. It’s well over 30 million lines of code, 60% of it drivers. You can’t just go and rewrite that in a different language, hell it doesn’t even compile on the wrong C compiler version. You would need access to the hardware and run tests for every module you change at least or risk breaking stuff in production.
C programmers will always be around since they are necessary to keep the old code running on newer hardware. There are thousands of companies relying on the Linux and BSD kernels, for example every network router, switch etc.
I have nothing against rust, but there is always a danger of having too many programming languages used in the same project, especially if a error in one language can break something in a module written in the other. That’s just a nasty complication, especially for a time critical project like the kernel.