Skip to contents

Write one replicate's FPSM raw output to a partitioned parquet

write_fps_raw_out_parquet.Rd

The writer counterpart to open_fps_raw_out_dataset(), mirroring write_forcs_log_summary_parquet() including its atomic publish: the parquet is written to a temporary file and then fs::file_move()d into place, so a concurrent reader or a retried write never sees a partial file. scenario is embedded as a data column so several per-scenario roots can be unioned.

Usage

write_fps_raw_out_parquet(
  src_path,
  scenario_dir = NULL,
  subdir = "_aggregates/fps_raw_out",
  staging_dir = NULL
)

Arguments

src_path

Path to one replicate's FPS_raw_out.csv.

scenario_dir

Scenario directory to publish under. When NULL (default) it is derived from src_path, skipping an fps/ working directory.

subdir

Path within the scenario directory for the dataset root.

staging_dir

Optional directory for the temporary parquet; NULL (default) stages in the destination so the move is an atomic rename.

Value

The written parquet path.

See also

Other FPSM helpers: fps_output_files(), fps_pools(), fps_run_docker(), fps_stocks_by_pool(), open_fps_raw_out_dataset(), read_fps_raw_out()