Oh, that plasma. Yeah, that naming conflict is totally not confusing.
You could switch all your repos to the core Arch ones. I did that by accident once, and it was fine (although, I did switch them back eventually). Maybe it’d add release stability? I’m not really clear how the EOS repos vary off the baseline, except by adding some custom packages.
Inspired by our discussion, I installed snapper on two boxen. I included snap-pac and snapper-support to get system change and grub integration; there’s probably also a utility out there that adds visudo-like snapshot-before-manual-edit of anything in /etc. If not, it’d be an easy script. snapper-gui and btrfs-assistant both look useful. While I’m comfortable with rescue SDs and restic backups, what I’m seeing with Arch’s snapper package is pretty nice, and super easy.
I suppose anything that borks grub is going to be a PITA no matter how immutable your OS, or how fancy your rollback. Or - god forbid - fucks up your BIOS firmware. I have never had that last happen, yet (knock on wood).
Everyone who has ever served in any branch of the US military knows this fact.
In 2020, US gvmnt estimates were 19.4m living vets in the nation. With a population of around 331m, that’s around 6%. If you include the number of non-vets civilians who’s jobs directly support the military, and who therefore are also required to use metric in their jobs, this number will be higher - I couldn’t find population estimates, but it’s at least a couple percent as the DoD alone employs a million civilians. NASA employs around 17k, and anyone working for a company who works with NASA is going to have to use metric. I’d be shocked if the number of Americans who aren’t having to use metric at work, and are therefore aware that it’s a standard here, wasn’t at least 10%.
Nowhere near a majority, but certainly significant; over 20m people warrants “many,” surely?
Ah. K, I think the differrence is that I’m the outlier. Your system has far larger components, with more moving parts, which I think is more common:
On most of my systems, I’m not running any graphical system; they’re all servers. That eliminates a huge amount of stack that can fail. On all but non-servers I run X, which is very stable (in that upgrades almost never impact it) on non-Nvidia GPUs. And of those, all but one run herbstluftwm - Gnome and KDE are both large systems with a lot of moving parts, any of which can break (or be broken) – in your case, it was Plasma, a KDE component. And the last desktop is running Budgie which, while still Gnome, is a lighter one based on the older GTK3. All of these things tend to make for more stable systems.
But, most people are probably running fancier, full desktop software. Larger, more complex, more development, more frequent changes. And, consequently, more prone to cascading packaging breakages, like the Plasma one.
I think if I were using software like that, I’d consider either giving up Arch and using an immutable distro, or using something like snapper or timeshift that allows boot-time system roll-backs.
Are you updating with eos-update, or yay? TBH, I only use yay, or pacman. I don’t imagine it makes any difference, but… IDK. I happened to upgrade and reboot two EOS machine yesterday, with again no issues. Are you running an NVidia card? I’m an Radeon guy, won’t touch NVidia, myself. How about Wayland? I’ve alwayw found Wayland to be super flakey, which keeps me on X.
I dunno. I wonder why you’re having so many issues, while for me EOS has just been Arch with an easier install.
The Honeywell HomeAssistant integration works pretty well, and has been around for a while, but it works through a web API. I’d prefer to have a fully local connection, but I’m not going to replace the entire HVAC control system to get it.
I know I should be smarter about them like you are
TIL being lazy is “smart”.
Nokia 3310!! Those were the days. When you had drive over to your hosting provider (some guy’s garage, who was paying for a T1) so you could sit at your server (a tower you’d built) to fix something that an upgrade had broken. Those experiences with dependency hell put me off Redhat forever.
It always bugs me a bit that all ST handgun-like phasers are “Type 2”, regardless of which design. I get what they’re doing, it’s just confusing. There are more than one TOS Type 2 designs, and obv. TOS Type 2 are radically different from TNG Type 2, of which there were several variations. The toxonomist in me wishes they’d been given at least sub-types; as it is, they’re categorized by (in-universe) date of introduction. But the shows and props department were sort of all over with them, making tweaks between seasons, so it can be rough to talk about without having memory-alpha up in a window.
That sounds stressfull! It’d put me off a distro, too. I had something similar happen in the early days of Gentoo - multiple times. Those trials by fire did teach me a lot, and I’m now consequently far more sanguine about the boot process, and thank god these days we have smart phones as mini-backup computers to search for solutions! Still, we’re in a time when PCs are not as indispensible, and having one down for a couple days can be a minor disaster.
Rolling updates or no, I rarely -Syu on my desktop more than once a week, and most of my machines get that TLC more like monthly. And sometimes I’ll hold out packages that require rebooting, because FTN. It probably contributes to the fact I’ve avoided these types of dramas --statistically.