I have read some fascinating threads about the history and implementation of the
shutdown commands (and equivalents, eg. runlevels and systemctl targets) across different versions of Unix(-like) OSes/Linux over the past several decades, and it has been made abundantly clear that the purpose, use cases, and implementation of
shutdown and the relationship between the two has and still does vary from distribution to distribution.
Having said that, I have a very specific question: are there any current/modern practical use cases for (directly) using
halt implementations that shut down the OS to a non-responsive/non-existent state (ie. not even a single-user shell) and leave the computer system itself powered on?
I'm asking because I would imagine most modern hardware would have a software mechanism for powering off, so I can't conceive of a need to halt an OS before manually powering off/restarting a system. (If I am wrong in saying "most", or if there is in fact a benefit to manually powering off/restarting a system with the OS terminated, please share.) I've also read that halt could be used for 'performing maintenance' on a computer system that is still powered on, but I haven't seen this point elaborated on - what kind of 'maintenance' could be done (and would be beneficial to do) on a powered-on computer system that doesn't have an OS (or any other software interface like BIOS) running?
For reference, these are the more useful threads I've read on the
shutdown discussion, though they haven't really discussed modern uses for
halt implementations that shutdown the OS without powering off the system: