I started down this path at one point but never finished. I have a draft of a versioned, coupled lock implemented elsewhere, but it's relatively untested and I will need to return to it.
I go back and forth on how much existing code can be shared.
Testing for correctness is perhaps the most challenging part here.
As per the https://db.in.tum.de/~leis/papers/artsync.pdf paper, implement a form of the tree using optimistic lock coupling.
I started down this path at one point but never finished. I have a draft of a versioned, coupled lock implemented elsewhere, but it's relatively untested and I will need to return to it.
I go back and forth on how much existing code can be shared.
Testing for correctness is perhaps the most challenging part here.