Debian has documentation of similar quality, with the Debian Reference[2] and the Debian Administrator's Handbook[3], which is more up-to-date. Also in DocBook, and all I said about the Fedora documentation applies as well.
This will not be a popular opinion, but I don't think Arch Linux's documentation compares to Debian's and Fedora's.
[1]: https://docs.fedoraproject.org/en-US/Fedora/23/html/System_A... [2]: https://www.debian.org/doc/manuals/debian-reference/ [3]: https://www.debian.org/doc/manuals/debian-handbook/
systemd has extensive documentation, both in manpages and in higher-level documents explaining how all the pieces fit together. I've generally found it far more straightforward to learn about than the components it replaces.
> One of the most egregious problems was forcing the change from text logs to journalctl binary logs.
You can keep a syslog implementation like rsyslog installed, and you'll continue to have text logs too. And journalctl outputs plain text.
people seem to have this analogue of "you can configure away the bad bits" but in the end the bad bits are orthogonal to it's design and 'bypassing' them or hacking around them should not be encouraged- it should be able to be done by default.
Where in my comment did I imply that the journal was "bad"? The journal is awesome. But if you have some workflow designed around syslog, such as a log analyzer or statistics package that you don't want to tweak, text logging still works just fine.
...and the manpages for every tool, every config file, built-in units, and every type of unit, plus the extensive documents linked from https://wiki.freedesktop.org/www/Software/systemd/ , including a 21-part series of articles targeted at system administrators, a dozen more pages targeted at users and administrators about specific topics, several dozen documents for developers, a dozen videos...
There are many criticisms you could reasonably throw at systemd, but "not well documented" certainly doesn't seem like one of them.
On boot, systems waits for two minutes or so for dhcp on an Ethernet interface with no cable, and it's not cancelable if I happen to be at the console
On shutdown, it turns off swap before shutting down daemons; not an issue for my system because I don't expect swapping, but in what world does this make sense?
Closing the lid sleeps the computer without configuration; some people might find this to be a good thing, but in my experience in the last 15 years, Linux only slept on lid closure if i was running software to handle that (acpid and/or something from a desktop environment), if i have apache running on an old laptop with no GUI, i expect it to stay on. (yes, there's a config, but its still annoying)
With sysvinit you'd usually just put it on S99 and be done with it in most cases.