Cool. I actually think we have very similar understnading of semantics and naming in this case.

+1, we do

I am open to remove it, but I think that should be another issue.

Thanks for the thorough explanation. Personally I like keeping (or making) things simple but you guys can better judge whether its worth the effort or whether there will be use cases which actually require the mechanism in its current form. I'd expect an implementation without the reference counting business and the involved synchronization to be a bit faster, whether that would have any impact in practice is another story. Anyways, good to know the backgrounds of this. Thanks again.