anti-apXos
Surely dbus could be launched through the run script but that is when you delve into the logic why use runit when sysvinit does the same (or similar thing)?
In fact, the bluetooth and connman services do start the system dbus if it wasn’t started already. In my experience, the benefit of runit over sysvinit is mostly just the simplified runlevels, plus on antix it has a nicer GUI service manager. As I understand it, runit’s handling of service supervision is supposed to be more efficient and reliable, too, but I’ve really never run into issues on either init, so I don’t know.
Runit has its merrits, but the other modern inits just do things better every way.
This statement needs some justification. Just because dependency resolution is handled differently if using runit, doesn’t make it worse. Why do you feel that having the init system itself handle it is better than having the service run script handle it. This sounds an awful lot like you’re equating “doing more things” with “doing things better,” by which reasoning we should all prefer systemd.
Handling dependencies in the service script seems to be to have some obvious advantages. One, it’s more flexible: you can define the dependency conditions however you want instead of having to work within the constraints of how the init system defines them. Two, it’s explicit in the service shell script itself, so anyone can just look and see what is being done. Three, it means runit is simpler and less prone bugs.
I don’t doubt that there are probably advantages to having the service manager handle the dependencies itself (though I’m still curious what exactly they are, though). Like with all things it seems to me to be a trade off, not just “better in every way.”
"--"