It’s clear that you’re upset at systemd, but what you said here doesn’t really scan if you actually read the document. Everything you could describe as “lock in” has a conditional statement clearly making it optional. The document barely even covers the extensive motivations for features it discusses, which I would expect if it was really propaganda.
If you implement a daemon in the new style documented here, and skip 5,7,8,9 - all optional, then the outcome is more portable than one following the sysv list (just doing the sysv list correctly for Linux and a BSD is a quite large burden due to differences). If you follow the Unix purist notion that this should all be composed of commands then all of the other sysv tasks could be handled by wrapper programs - you can also best use the new style. That’s mostly what using daemontools feels like, and it works pretty well, but boy is it a mess in the process tree and require a ton of boilerplate.
If you implement a daemon in the new style documented here, and skip 5,7,8,9 - all optional, then the outcome is more portable than one following the sysv list (just doing the sysv list correctly for Linux and a BSD is a quite large burden due to differences). If you follow the Unix purist notion that this should all be composed of commands then all of the other sysv tasks could be handled by wrapper programs - you can also best use the new style. That’s mostly what using daemontools feels like, and it works pretty well, but boy is it a mess in the process tree and require a ton of boilerplate.