Skip to contents

Initialize checkpoint directory

Source: R/checkpoint.R
init_checkpoint_dir.Rd

Creates the checkpoint directory structure for a new simulation run. This includes the base result path, checkpoints and outcomes subdirectories, run manifest, and initial latest pointer. The ledger subdirectory is created lazily at the first delta write.

Usage

init_checkpoint_dir(
  result_path,
  config_fingerprint,
  config_spec = NULL,
  checkpoint_format = "rds",
  retention_spec = NULL,
  run_policy_spec = NULL
)

Arguments

result_path

Character string giving the base path for results. If NULL, the function returns NULL immediately (no checkpointing).

config_fingerprint

Character string containing a hash of the configuration. This is stored in the manifest for validation during resume operations.

config_spec

Optional normalized configuration spec to persist in the run manifest for future rehydration.

checkpoint_format

Character scalar naming the checkpoint storage format.

Value

Invisible path to the result directory, or NULL if result_path is NULL.

Details

The directory structure created is:


result_path/
+-- run_manifest.json    # run-level metadata and schema versions
+-- latest.json          # pointer to latest valid checkpoint ID
+-- checkpoints/         # atomic checkpoint commit directories
+-- outcomes/            # immutable, redundantly mirrored outcome shards
+-- ledger/              # base ledger plus status deltas (lazy)

The run manifest contains:

  • run_schema_version - Schema version for the run structure

  • result_schema_version - Schema version for result files

  • config_fingerprint - Hash of the configuration

  • created - Timestamp when the run was created

See also

write_checkpoint(), read_checkpoint()

Examples

if (FALSE) { # \dontrun{
result_path <- init_checkpoint_dir(
  result_path = "/path/to/results",
  config_fingerprint = "abc123hash"
)
} # }