Combine row evidence from several producers.

Merge evidence tables deterministically. Nothing is overwritten: the result is the union of the claims, with exact duplicates collapsed and the surviving rows in the format's canonical order, so the same inputs always give the same output whatever order they arrive in.

Two producers claiming different values for the same row and field is kept as what it is — two claims — instead of one silently winning by argument position. Find those with evidence_conflicts() and decide explicitly.

All inputs must agree on key_columns: evidence keyed on (area_code, year) and evidence keyed on (area_code) address different things, and merging them would change what a row_key means. That aborts rather than being reconciled.

Usage

combine_row_evidence(...)

Arguments

...

Evidence tables from row_evidence(), or lists of them. Each is validated against row_evidence_schema(). Zero-row tables are allowed and contribute no rows or key constraint.

Value

A tibble of row evidence, in canonical order.

Examples

production <- tibble::tibble(area_code = c(724L, 76L), year = 2020L)
key <- c("area_code", "year")
recorded_at <- as.POSIXct("2026-01-01", tz = "UTC")

faostat <- row_evidence(
  production,
  source_id = "FAOSTAT_prod",
  key = key,
  fields = list(method_land = "reported"),
  recorded_at = recorded_at
)
luh2 <- row_evidence(
  production,
  source_id = "LUH2",
  key = key,
  fields = list(method_land = "back-cast"),
  recorded_at = recorded_at
)

combined <- combine_row_evidence(faostat, luh2)
combined
#> # A tibble: 4 × 7
#>   row_key         key_columns   source_id source_version recorded_at field value
#>   <chr>           <chr>         <chr>     <chr>          <chr>       <chr> <chr>
#> 1 "724\u001f2020" area_code,ye… FAOSTAT_… NA             2026-01-01… meth… repo…
#> 2 "724\u001f2020" area_code,ye… LUH2      NA             2026-01-01… meth… back…
#> 3 "76\u001f2020"  area_code,ye… FAOSTAT_… NA             2026-01-01… meth… repo…
#> 4 "76\u001f2020"  area_code,ye… LUH2      NA             2026-01-01… meth… back…

# Both claims survive, and the disagreement is visible.
evidence_conflicts(combined)
#> # A tibble: 2 × 5
#>   row_key         field       n_values values               source_ids        
#>   <chr>           <chr>          <int> <chr>                <chr>             
#> 1 "724\u001f2020" method_land        2 back-cast | reported FAOSTAT_prod, LUH2
#> 2 "76\u001f2020"  method_land        2 back-cast | reported FAOSTAT_prod, LUH2