Seems like sour grapes from the author of another initialization subsystem (mdev). His argument would seem to apply to any effort to standardize initialization on Linux.
The point he's making is that having One True System is the death of one of the biggest strengths of the Linux ecosystem: diversity. What he's ignoring is that the various distros that are making this change have done so because they think it's the best decision for their users.
No higher power decreed "systemd or death" and forced it on the distros. They've adopted it on its merits, and most have retained the ability to run other init systems if the user wishes.
This is a pinnacle of the Linux ecosystem, not a blemish on its record: the larger community is working towards better interoperability between distros.
What the systemd supporters are ignoring is that if the existing init system was as all-encompassing and heavily intertwined with the rest of the system as systemd is, it would be impossible to replace it with systemd no matter how much of an improvement it is.
One of my favorite phrases from the Refactoring world is "speculative generality". Adding customization points in the present can make adapting in the future easier if the future aligns with your expectations, and harder if it doesn't.
As I commented elsewhere, I'm running systemd with syslog-ng, netctl, and cronie. It plays just fine with them. Do you have examples of pieces I'd want to swap in but can't?
His argument would not apply to an actual standardization effort without the negative attributes explicitly called out in the article. (Non-standard, undocumented moving target, and a community of developers who go out of their way to antagonize competing projects.)
Yeah, that seems to be the point, that initialization shouldn't be standardized or at least should be standardized based on a formal spec. It's his entire criticism.
