Thanks! One thing I am concerned about is that it might be more efficient to operate on the underlying low and high bits directly and return a new long from the computed bits just once, instead of relying on the high level wrappers that potentially allocate intermediate instances multiple times. Wdyt?

It's a fair point. I'll take a look at it.

Perhaps something more along these lines?
I think I got the logic right. (Been on vacation, so my brain might be running in a lower gear.) I certainly added a couple more test cases.

@dcodeIO When you get a moment, I think I'm ready for more input. Thanks.

Thanks, looks good! Merging, and if there are still issues we can continue from here.

Hi @dcodeIO, are you going to publish a new version containing this commit?