This makes sense to me.

However, in order to maintain conistency, I'd rename ".filetag_tagtrees" to ".filetags_tagtrees".

Would you modify your pull-request accordingly? I could not find out how to make the change myself ...

Sure, I renamed the files accordingly. I also re-added the options.tagtrees_directory condition to ensure that existing users don't have to create a .filetags_tagtrees file manually, oops.

Allowing edits by maintainers
If checked, users with write access to novoid/filetags can add new commits to your fix/directory-emptiness-check branch.
You can always change this setting later.

I think you'd have to use my fork to make any changes yourself.

I'm writing unit tests for this change, but tests for this are rather flaky. generate_tagtrees always assumes that the directory parameter is the default directory in the test environment. assert_empty_tagfilter_directory skips the check for the default directory based on options.tagtrees_directory, but the tests never set any cli options. generate_tagtrees relies on assert_empty_tagfilter_directory, hence flaky tests.

I'll probably replace the options dependency with a check for the real default directory or smth, not quite sure yet.

I read #42 and wonder whether I should incorporate that into this PR. I'd be weird to create a .filetags-tagtrees file for tagtrees and tagfilters when there's an issue that expects separate files for each in my opinion.