mirror of
https://github.com/Kpa-clawbot/meshcore-analyzer.git
synced 2026-11-11 12:18:47 +00:00
The fallback for nodes never heard zero-hop put them in every region that heard any of their adverts. Adverts flood the whole mesh, so on a statewide deployment most nodes landed in every region. Such nodes now go in the regions that heard them in the fewest hops, and in none when even that is over 3 hops. The recomputer computes this alongside the home regions; the observer-based scan decides only for nodes newer than the last recompute. Perf: one extra pass over in-memory advert observations per recompute (5 min), hop counts from path_json without decoding. 220k adverts x 8 observers: ~50ms under the read lock. RegionNodePubkeys adds one map lookup per heard advert (BenchmarkRegionNodePubkeys ~20ms, unchanged).