On Arch, $DEITY help you if you don't think to check the webpage before installing updates with pacman. Pacman thinks nothing of installing an update that will break your system if you don't do exactly the steps detailed on Arch's front page, nor (last I used it) will it even notify you "hey, you really should do these things first, y/n?".
Debian and apt are considerably less dickish.
True. Knowing that's how they do things, it doesn't feel like a particularly onerous requirement, but there are better ways they could communicate such things. (They do have a mailing list for those changes - it's still definitely not ideal, but probably better than just checking their site.) I don't think of it as a pacman problem so much as a rolling release problem, though I could be way off base. Something with stronger versioning has a well-defined place to make changes like that, and without that it's a bit tricky. I prefer Arch's way to any kind of half-baked automated system, if that would be the alternative.
The average effort this adds to updating is very low, as changes requiring manual intervention are quite rare. Here's
that news page, for reference, and I find that maybe one of those entries per year requires me to do anything before updating.
It does sound like quite a pain for Antergos and Manjaro to handle, and I'm curious how they do it (but apparently not curious enough to go find out).
edit: Maybe Waco and I's luck has to do with Z170+6700K+16GB of RAM + Nvidia graphics + Seasonic PSU. Joking, but I am always a little miffed when the horror stories come out because I have experienced precisely none of them.
mainly on my users' computers.
Not so sure that's a joke. On reasonably powerful SSD-equipped machines, Win10 seems to be doing mostly fine these days. When dealing with Atom-class CPUs and/or spinning rust, I still see it do all kinds of nonsensical things (some update related, more of them not). It almost feels like they handled some not-entirely-critical synchronization issues by going "that other task has had two whole seconds to do its thing, it's gotta be done by now amirite?"
Really? My Win10 desktop has never ceased functioning for a task just because of a background update. What did Windows 10 screw up so badly that Manjaro is better?
Most commonly I see machines stuck updating in infinite loops in which WU hogs CPU time / IOPS / downlink indefinitely (many hours on every boot) without ever actually managing to commit any updates. Depending on the hardware, internet connection, and use case, this can be anything from a moderate annoyance to a complete show-stopper. 1710's update throttling will have made that massively better, but it's still a security hole. Theoretically there are proper in-place fixes for it, but I've never been able to make any of them work for various reasons (like Win10 Home not giving me the permissions needed), and have always side-stepped the issue somehow (manually installing 1703 did the trick when it happened on my main rig).
There was also that one time that nothing I did could get the right graphics driver to install on one system. WU would constantly replace my work with its own incorrect guess, and none of the overrides designed to prevent that did anything. (Settings like those overrides not having any effect is one of the common themes I've noticed with Win10 on weak systems.)