we're currently working on setting up what is honestly quite a small tool to run on a NixOS server and uh

this is one of those open source tools with a company behind it, the documentation is honestly very professional

but the documentation completely does not speak to the concerns a sysadmin has AT ALL. it pushes very heavily to use Docker (in any of ten different ways) and the non-Docker explanation leans on Debian to have already done the hard parts.

and it's

we don't want to complain about this tool (so we're not telling you which one it is)

but like

we see mass knowledge loss happening here, much like the thing about twenty years ago where mainstream development practices stopped "knowing" how to do GUI development that isn't web-based

like it feels as if the people writing this software do not, themselves, know how to be a sysadmin

and like the web isn't terrible? we dislike this thing that web practices and Docker both do, of encysting complexity rather than engaging with it, which doesn't really work and winds up taking, like, three orders of magnitude more computational resources. but the web isn't terrible, people build careers around it, we wouldn't do it this way but it's ... humanity has made many far worse decisions than the web?

but this is unsustainable

all y'all know that, right?

frameworks that try to simplify things can remove the accidental complexity of a task, sometimes, but never the essential complexity, the stuff that necessarily arises from what the task is in the first place

deploying software requires, like, knowing how operating systems work. fundamentally. you can't hide that.

we use containers when they make sense (rarely. if you want a security boundary use a VM)

but not through Docker, through Nix, which engages with the complexity

we REFUSE to use Docker in both our personal infrastructure and the stuff we're professionally responsible for

investment of time into learning Docker's particulars would just be wasted, it would move our knowledge backwards not forwards