The evidence ledger a set of tuning tables was scored on

scripts/0310 writes one ledger per evidence build into scripts/logs/habitat/, so a build checked but not promoted (#111) leaves a newer ledger than the published tables'. Since #111 each summary names the ledger its run read (evidence_ledger); this returns it, so scripts/0325, scripts/0330 and the methodology bundle read the published run's build, not the newest on disk. Tables from before the record fall back to the newest ledger, which is what every reader took until then.

Usage

knw_run_ledger(tune_dir, log_dir = "scripts/logs/habitat")

Arguments

tune_dir

the directory of <key>_summary.csv tables (not recursed).

log_dir

where the ledgers live.

Value

character(1) the ledger's path.

Examples

d <- tempfile(); dir.create(d); dir.create(l <- file.path(d, "logs"))
for (f in c("20261020", "20261027")) {
  writeLines("kind,step,n,note", file.path(l, paste0(f, "_evidence_build_ledger.csv")))
}
utils::write.csv(data.frame(species_code = "CH", stage = "rear",
                            evidence_ledger = file.path(l, "20261020_evidence_build_ledger.csv")),
                 file.path(d, "CH_rear_summary.csv"), row.names = FALSE)
basename(knw_run_ledger(d, l))   # the run's, not the newest
#> [1] "20261020_evidence_build_ledger.csv"