Hacker News Reader

Comment by Majestic121 | Hacker News Reader

If you have to configure and set up everything yourself, it can not really be called 'just works'.

I'm definitely a fan of setting things up and diving into config myself, but let's be honest : Arch is not a 'Just works' distribution. Ubuntu could be considered one, maybe elementary as well

View on HN

This "configure and set up everything yourself" bit is overstated.

It's not like Ubuntu, or similar batteries-included distros, doesn't require configuration and tweaking to get it your personal liking, anyway. And it's not like you cannot make choices during installation that affect the final state of your system. The difference between the Arch manual installation process and Ubuntu's is simply that you use a CLI and start a few scripts instead of pointing and clicking.

The amount of configuration necessary is exaggerated, probably by people that want to portray Arch as this expert-user distro. Once you've done e.g. "pacman -S gnome gnome-extra sddm && systemctl enable sddm", you can just boot into your graphical desktop and start using it just the same as Ubuntu, no extra configuration steps needed.

And, like I've said, for the nth time, use Arch-Anywhere to get a similar installation to that of Ubuntu.

I've been very impressed by Arch. Yes, it's a rolling release, so you have to update your system now and then, but this is a good thing--install once and run an up-to-date system for years on end. Unlike Ubuntu, which is outdated already on install, and if you want to upgrade, good luck with that process if you've installed a sundry of packages. Eventually, with an Ubuntu system, I've always started suffering from package-related problems, breakage, dependency issues, etc.

I won't detail all the problems I've had with other distros, suffice to say I've had none of those using Arch.

These supposed to "just work" distros haven't lived up to that in my experience.

"Yes, it's a rolling release, so you have to update your system now and then, but this is a good thing--install once and run an up-to-date system for years on end."

In my experience, this also lets the system collect all kinds of cruft over time, which can definitely impact stability and predictability when it comes to updates, unless you're extremely vigilant about housekeeping.

A pacman -Syu a day keeps the doctor away.

As long as you remember to check the website for manual intervention requirements....

Which is exactly my point.

Put off updating for a couple of weeks because you're got better things to do? You may end up borking your system unwittingly.

Not subscribed to the Arch news feed and intimately familiar with the packages installed on your system? You may end up borking your system unwittingly.

Arch is good for dedicated power users, who want to tinker around. But most of us just want to get stuff done, not mess with the OS.

Stop perpetuating this myth that you have to mess around and tinker to keep Arch in working condition. It's one update command, and it's not something you need to do daily. You can put it off for weeks, all that will happen is you'll have a bigger download.

I use Arch exactly because I want to get shit done. My way.

I ran Arch for ~9 years, and there were several cases of OS/boot-breaking updates, requiring me to boot from a USB stick and manually fix things when the updates decided to shit all over the system. Having to manually cross-reference an update list with a list of "oh damn, you need to do these manual steps or you'll break everything" errata from the Arch news feed is not my idea of a good time.

The only reason the Arch forums are so informative, is because people run into stupid issues on a regular basis.

Mint updates itself and gets out of my way. That makes it ideal for actually getting shit done.

YMMV.

Mint packages tend to be quite old, even older than Ubuntu. This is no good if you want to run close to, or, upstream versions, which Arch excels at. Of course, you can install software not using your package manager, but that defeats the point of having centralized package management in the first place and can cause problems.

I used to think I wanted the very latest version of every single application, not matter what.

These days, only Chrome, Spotify, Qbittorrent and a couple of others are the newest versions (through PPAs), everything else is just stock and works perfectly fine.

I'm not even talking about end-user applications. You have different needs as a developer. I need libraries and compilation chains to be up to date.

So you add backports or PPAs for those exact libraries.

Which, in the long run, and sometimes short term, causes problems.

Sorry, but that is exactly my experience. Over and over. Arch breaking every couple of weeks in a time consuming way. Maybe your setup is so minimal that you aren't affected? For me it was "oh my, what's broken now again?". This really hurts, if you're a freelancer and time is your own money and not your employer's.

As I said, it's a different definition of "just works" - obviously no non-technical person is going to run Arch - it's never going to be Ubuntu and it's not trying to be. I wasn't trying to make that claim.

As a power-user, Arch "just works" for me far more than other distros have, because I know the system and it's simple, I've never had a problem where I didn't have a handle on it.

I can affirm that. I don't understand how Arch gets this reputation for breaking during an upgrade. In the four years I've been using Arch on a variety of systems, I've never had anything break in a way that didn't have a trivial, obvious fix, and most of the time pacman points out potential problems during the upgrade. Arch has allowed me to spend way less time on system administration than I used to before I switched. Are people doing "sudo make install" or running binary driver installers as root? Or maybe it's because I switched right after the systemd transition and missed all the fun? Maybe I'm just lucky?

Do you understand the difference between Arch and distros like Debian? When GNOME ships a new version, it almost immediately hits the Pacman repos. Do that for every package and you're going to have integration issues. Your users also get to be the guinea pigs for new releases. Ship buggy packages and sooner or later your users' systems will install a critical package with bugs. This is why Arch gets "this reputation for breaking during an upgrade." And yes, you're just lucky. When I hear Arch users saying "it hasn't broken for me!" I hear "my girlfriend and I don't use protection, and she's never gotten pregnant!"

Right, this is what people always say about Arch. And yet, in practice, on a variety of hardware, I've had nary a problem in the four years I've been using it. Contrast this with big-release distros, where I've literally never had a successful upgrade between major versions. But maybe the key point is in your second sentence. I've never had a problem with Gnome breaking because I have no interest in running a DE, and even on those systems where I've got one installed, I boot to the console and start i3 or sway. If Gnome was broken, I'd never notice.

And this might be the key point: if a package like Gnome is broken, I personally blame the package rather than the OS. I don't see a GUI desktop as integral to the operating system, so when I say that Arch has been trouble-free, I mean that I'm always able to boot to a usable system console and get my work done. So that's my disclaimer, and probably the big reason people shouldn't use Arch: if you rely on a DE, and see the system as broken without one, you're probably going to experience way more breakage than I've reported.

Fair enough, every time I've used Arch I've gotten burned by an update 2-3 months in, ranging from annoyances to an unusable system. I like to think I'm pretty good with Linux having used it on the desktop and sysadmining for a long time, but I put Arch in the same category as Gentoo, cool but too finicky and fragile.

> it almost immediately hits the Pacman repos.

I tried Arch and it's not the case, my examples:

PostgreSQL: https://www.postgresql.org/docs/9.6/static/release-9-6-2.htm... 9.6.2 released 2017-02-09, in Arch repos - https://git.archlinux.org/svntogit/packages.git/log/trunk?h=... - almost a MONTH later. 9.6 released 2016-09-29, in Arch repos - 2016-10-29 (version 9.6.1, they skipped 9.6) - again, a MONTH later.

Had issues with outdated Node.js and some other package I can't recall.

To be honest, up to date packages for me were only available with brew in macOS.

There's a difference between running on the [testing] repos, which do what you say, and the regular default repos, which wait for packages to graduate from [testing] before rolling them out.

Only a small percentage of packages go through [testing]. Typically only when problems are expected. [community] and [extra] require no testing, just that the packager says "good enough for me." [core] is held to higher standards.

It's meant as in what packages go in, what DE you use, which mirrors do you pull in from etc. so that you know what is going on in your system, despite this, it still 'just works', that's the beauty of it.

I would call it 'just works as you spected' contrary to 'just works as someone else spected'. Neither is good or bad, it depends of what are you looking for.

Reply on news.ycombinator.com