I don't understand what this whole RCU business achieves. It's trivially completely wrong: the RCU pointer is stuffed into a larger structure that might be deleted anyways, nothing's actually protected. And it seems that you fitted all the metrics into one value that is small enough to be atomic anyways, so is RCU needed at all?

Move to AvalancheState, run the score update in the scheduler queue and use RCU to make it lock free most of the time.

mengandalkan bohlam petunjuk klaim tangga inti kurung buah ayam inti pasien

This is lacking the bigger picture. We can't just break the proof format with every version. What's the format of a proof you are moving toward?

Note that I do not intend to land this. It is meant as an example of a possible solution, but the tradeoffs of the various solutions need to be discussed.
In particular, there is a solution that involves making the ProofId no longer depend on the stakes that would have a bunch of upsides (no need to regenerate a Delegation after updating a proof) but also challenges (if the ProofId does not change, it does not trigger proof relaying).

Add multiprecision/cpp_int.hpp Boost dependency (Lord have mercy)

Fix off-by-one error in overflow_add_int64 and overflow_sub_int64 (doesn't seem to appear on Darwin/clang)

Looks good. This is an important step in preparation for upgrading to webpack 5 and continuing to simplify the dependency tree.

Feedback

Fix functional test

Add to Tokens page

You missed the main one:

Resolving comments

testing depends on core#16975

needs backporting 17984

The test plan is not testing the feature

Please backport 17984 in its own diff, it's difficult to make sure that nothing is missing here

Side note: you may want to add yourself a commit hook or something to automate the @generated mark in the snapshots

The display is not updated in the tokens page

@bot build-master