@canartuc Not aimed at you but rather the "parasite" conclusion of the other mentioned handle. Should probably have left you out as mention there (new to participating more actively on mastodon)

Absolutely no issue with your original post and reply. It was a good summary 👍

@maat @canartuc It does so in a way that was at least as modular as any of these approaches I've seen and generally better documented as well. The reason that no alternatives to those facilities generally are developed may very well be because the baseline they provide is good enough.

When things start to depend on systemd it's usually because it provides a reasonable high-level interface for something that they had to maintain themselves before.

@uvok To do that robustly you basically need the service to specify what it needs in some rough abstract terms.

There isn't really a standard way of doing that and if you try to formalize something I'd say there is a large chance it would look a lot like the information included in unit files today.

To have a convention of services starting to provide that by default in some known locations would help a lot.

@cve @zekjur I didn't really read it as bashing openrsync but rather the people calling for everyone to change to it.

For many it's probably as you say a fine replacement but this debate has been about a not that widely used feature that broke and arguing for making a switch for fear of their unusual usecase failing sounds like a very backwards strategy.