Pieter Wuille

@Sjors An earlier version of the code had net_processing construct a CBlockLocator normally, hand that to HeadersSyncState at creation time which would remember it, and then the locator returned was...

> I'm not sure how to confirm that the performance decrease o(1) to o(log(n)) is negligible in all the places it's used. As far as I know, they're only constructed...

ACK f0dae7c6fa54d4d00489da02eebfa3e4ef40cb63 (for the parts that weren't authored by me). Reviewed the diff with my previous review.

re-ACK 3add23454624c4c79c9eebc060b6fbed4e3131a7

The `precompute_ecmult.c` and `precompute_ecmult_gen.c` shouldn't be needed; they are for computing the tables, but the tables are checked into the repository. In fact, I'm surprised it even works with those...

@prayank23 Without blockfilters it would take many times longer, and without this PR the functionality to scan blocks doesn't exist at all.

@prayank23 You only need to build the filters once, and you need them too for -peerblockfilters (if you want to support BIP157 protocol).

I think that approach makes sense; it's equivalent to treating every ban as a per-IP thing, and remembering the longest-duration ban for each. One question I have is how to...

> and a single key combo for the seed (the seed is a valid key that we can receive coins at!) 🤦 Nice catch.

* "It's a feature of existing wallets, new wallets should have it too" is not a good argument. It's a new wallet type, it will support whatever features we declare...