Understanding Loss Reports

When converting to a less expressive engine wf2wf writes a <output>.loss.json side-car.

Anatomy of the file

{
  "target_engine": "cwl",
  "entries": [
    {
      "json_pointer": "/tasks/align/retry",
      "field": "retry",
      "lost_value": 3,
      "reason": "CWL has no retry field",
      "severity": "warn",
      "status": "lost"
    }
  ]
}

• severity – info, warn, error (future use)

• status – lost, lost_again, reapplied

CLI workflow

# Convert but abort on any brand-new loss
wf2wf convert -i Snakefile -o workflow.cwl --fail-on-loss

# Validate later
wf2wf validate workflow.cwl            # checks embedded schema
wf2wf validate workflow.cwl.loss.json   # checks that losses are allowed

Tips

• Use a richer target (e.g. DAGMan) if you see critical losses.

• Keep the .loss.json with the workflow so information can be re-applied when converting back.