Let us all remember that, at least back when it started, the establishment alternative to systemd was a product named after its original operating system, System V UNIX, which is a direct descendent of the original UNIX from AT&T. This sysvinit software used complicated shell scripts to manage daemons. Contrary to some opinions, these shell scripts were not “just working”; they were in fact a constant and major maintenance burden for Linux distributions. When I started on Linux at least, Debian had a suspiciously large fraction of bugs on init script breakages.
All this is to say that the new system, systemd, doesn’t have to be anywhere near perfect to be worth replacing sysvinit.
People argue that systemd is rejecting the “UNIX philosophy” of small tools that do one thing well. I argue that this UNIX philosophy is not some kind of universal good with no tradeoffs. It’s an engineering rule of thumb. There are always tradeoffs.
People argue that systemd is too much like Windows NT. I argue that Windows NT has at least a few good ideas in it. And if one of those ideas solves a problem that Linux has, Linux should use that idea.
Let us all remember that, at least back when it started, the establishment alternative to systemd was a product named after its original operating system, System V UNIX, which is a direct descendent of the original UNIX from AT&T. This sysvinit software used complicated shell scripts to manage daemons. Contrary to some opinions, these shell scripts were not “just working”; they were in fact a constant and major maintenance burden for Linux distributions. When I started on Linux at least, Debian had a suspiciously large fraction of bugs on init script breakages.
All this is to say that the new system, systemd, doesn’t have to be anywhere near perfect to be worth replacing sysvinit.
People argue that systemd is rejecting the “UNIX philosophy” of small tools that do one thing well. I argue that this UNIX philosophy is not some kind of universal good with no tradeoffs. It’s an engineering rule of thumb. There are always tradeoffs.
People argue that systemd is too much like Windows NT. I argue that Windows NT has at least a few good ideas in it. And if one of those ideas solves a problem that Linux has, Linux should use that idea.