John Hughes <john@???> wrote:

> Yes, the impression I get around here is that this is a religious argument for most of you.
>
> I had hopes for Devuan, but the lack of rational thinking convinces me that it's going nowhere.

There's no lack of rational thinking.
People here don't want to run SystemD, and what's more, we don't want to encourage people to add gratuitous dependencies because "it's OK, there'll always be a bit of SystemD present". While libsystemd0 may appear harmless, it isn't for two reasons :
1) By making it "OK" to have because "it does nothing", it encourages people to assume it's presences and use it - rather than actually checking first or just not using it if not needed.
2) I don't have the skills to check, and keep checking, that libsystemd is in fact "harmless". Just because it doesn't do anything now, doesn't mean that tomorrow someone will find that "doing nothing" is inconvenient* and so some "does more than nothing" code gets added.

* I'm thinking, someone decides they want to use a systemd function, but finds that the call "fails" when systemd isn't installed. So instead of just accepting that "if systemd isn't installed, they'll have to do X another way", they may well suggest that the functions supporting X are moved from the systemd package to libsystemd0 - "it's still OK, it's only a tiny support function". And so it goes on, like boiling a frog, until having libsystemd installed equates to running significant chunks of systemd itself.

Now, you may consider "doesn't want any part of systemd on my system" as religious zeal. It's not, I just don't want stuff that as far as I can tell is primarily designed to reduce reliability (in terms of the stuff I run). Not a single (claimed) "benefit" of SystemD is actually a benefit for my systems, in fact far from it. While I'm not a "programmer", I do know enough about the subject to read between the lines of some of it and see just how bad it is.
And yes, I've seen a function implemented while has just one function to cause data loss - why else would anyone go to the trouble of making an "async" sync call and complain about sync being sync ? It may or may not have been fixed, but the very fact of it getting into the project in the first place simply shows that the project is run/managed by people who (being generous) simply don't have a clue.

I don't want that on my systems.
I *like* text log files.
I *like* shell script init files.
I *like* sequential (deterministic) service startup.
These things have got me out of the brown stuff more than once !

I'm not in the least bothered about shaving a few seconds off the *apparent* boot time given that the hardware alone can take minutes before the OS itself gets to start.

Like most, iff SystemD was "just an init system" as some of it's supporters keep suggesting then no problem. I'd just not install it and carry on. But it isn't an init system - it's a Windows style "blob" of all encompassing stuff that goes against everything I like in Unix/Unix like systems !

> Bye.

Good bye - please don't come back until you've understood.