[Zethson]

See title.

[falexwolf]

It's been so far a technical internal parameter.

I see that it'd be sometimes useful for CI runs, but I was worried about people accidentally using it.

If we think the danger is low, we should only briefly ponder whether this is an ideal name.

require-empty currently only checks whether all managed storage locations are empty, it ignores the database itself.

[Zethson]

This is actually behavior that I would have expected under a --force parameter. However, the --force parameter currently skips the confirmation which isn't what I would have expected it to do. Having a --confirm/--no-confirm parameter instead of force and force behaving like require-empty sounds more intuitive to me.

[falexwolf]

--confirm/--no-confirm to replace the current --force sounds awesome! That's a much better naming choice.

Having an "actual" --force that enables people to leave data unusably dangling with cryptic ids in cryptic storage locations still sounds like something I'd rather not expose to a user. It's rather small nuisance to manually delete a bunch of files but it's an absolute catastrophe if anyone ever accidentally adds --force, e.g., because they confuse it with --confirm.

[Zethson]

@falexwolf all right.
I changed this PR to now rename --force to --confirm/--no-confirm instead of it's earlier purpose.

[falexwolf]

This is great! We'll just need to either update all notebooks and scripts everywhere to reflect it or this change should be made in a backward-compatible way together with renaming the param within lamindb-setup.

I suggest to make it backward compatible so that we have more time with the renaming (it'd be quite a bit of work).

[falexwolf]

Deleting databases will likely look different in the future. To avoid disruption the flag will remain --force until then.