Hey @xiuxiu62 thanks for taking an interest! Unfortunately I've decided to move forward with #67 mainly because your change would break backwards compatibility, therefore necessitating either a major version bump or a deprecation process leading to an eventual major version bump. I do, however, like the simplification of how comparators are treated so I'd be happy to have that as a separate PR. Let me know your thoughts.

Completely understand, no worries. I'm not sure it makes sense to change the comparator implementation, unless there are more client options that can be used to produce a comparator. It got a little hairy dealing with the move semantics of multiple comparators when you have more than one branching source, but with just the SortType and dirs-first bool, the original implementation works great. I'd be happy to work on a separate PR though, if it would be beneficial for the project moving forward.