Bitcoin full node and query server

Feb 25, 2026

View in context in the libbitcoin Slack #general mirror.

evoskuil07:44
I've completed the direct block/tx fee query optimization. Not only does this significantly optimize the fee accumulator initialization and updates, but also gets us very close to full optimization for all of our REST queries. There were a few that required population of the full block or tx from the store, as well as previous outputs, in order to obtain computed properties (such as fees). This is now reduced to just the sigops property. Sigops can be computed directly via query, just like I've done with fees. I'll probably do that eventually, at least when we get to candidate block generations. sigops and fees were originally intended for caching, but the requirement for prevouts precludes caching during archival, and to preserve the performance of milestone sync, we don't want to compel validation queries just for this. Since these cached values are not used in validation, the query performance is more than sufficient. In the mean time I may drop the sigops field from the native REST queries in order to get the performance benefit of avoiding the full block/tx queries just for this one property.
(edited)