Lennart Poettering

I am not entirely sure I grok the reasoning here. Why precsiely make this a high-level knob? Can you explain?

Hmm, I am still not sure I grok the need for a separate knob, but ok, others seem to see this differently. Not a biggie. However, can we rename this...

> Hmm, it's not necessarily clear to me. `LogLevel=` and `LogTarget=` and actually backend-neutral knobs, i.e. they configure the logging itself, no matter where the logs end up. But we...

Frankly, we probably should refuse such setups, i.e. socket units are started before the services behind them, but RuntimeDirectory= is about directory ownership, about binding lifecycle of dirs to service...

I'd really prefer if we wrote this in our coding style. Yes, it's a bit different from the kernel's one, but we do care about consistency within our codebase. Sticking...

Hmm, I am not sure I follow? If this is just a new stanza for boot loader type #1 files (i.e. .conf), then how exactly is this beneficial? i mean,...

I am not convinced we should add this. A system that doesn't clean-up during runtime basically means you are letting resources to be used unbouned, and that's not a model...

does dracut invoke "systemd-shutdown" as last step as pid 1, i.e. a 2nd time? this wasn't the original idea, but some initrds do this, and I think it is good...

so, what do these commands show: ```console # busctl get-property org.freedesktop.login1 /org/freedesktop/login1 org.freedesktop.login1.Manager HandleLidSwitch # busctl get-property org.freedesktop.login1 /org/freedesktop/login1 org.freedesktop.login1.Manager HandleLidSwitchDocked # busctl get-property org.freedesktop.login1 /org/freedesktop/login1 org.freedesktop.login1.Manager HandleLidSwitchExternalPower ``` This...

> Just to make it clearer: with the patch applied the fs strict enforcement will be disabled only during the short period of time when PID1 will basically "manager_free() +...