How aaronxyliu/recall-dataset works
ReCall Dataset is a frozen evaluation corpus, not a recovery implementation: it packages production-bundle inputs and source-to-bundle function-pair ground truth for seven bundler groups. Its user-facing result is a reproducible dataset indexed by qualified-dataset/index.json. The visible internal workflow includes source qualification and function measurement, followed by correspondence collection and dataset auditing. The README excerpt establishes the dataset’s scope and index, while sampled source establishes only some internal calls; unsampled collection and reproduction stages are therefore shown only where their links are documented or supported.
Main entry points: measure.mjs Function measurement · index.mjs POG anchors · functions.mjs Function inspection · audit.mjs Dataset audit
Components 16
Dataset artifacts
- Qualified corpus
qualified-dataset - Dataset index
qualified-dataset/index.json - Reproduction inputs
qualified-dataset - Dataset audit
tools/audit.mjs
Source qualification
- Source preflight
provenance/source-gate/source-functions-3/preflight.mjs - Lock resolution
provenance/source-gate/source-functions-3/locks.mjs - Function inspection
provenance/source-gate/source-functions-3/functions.mjs
Function analysis
- POG anchors
provenance/source-gate/initial-anchors-3/runtime/components/pog/index.mjs - Function measurement
provenance/source-gate/initial-anchors-3/runtime/src/analysis/measure.mjs - Source parsing
provenance/source-gate/initial-anchors-3/runtime/src/analysis/parse.mjs - Project inventory
provenance/source-gate/initial-anchors-3/runtime/src/project.mjs - Artifact checks
provenance/source-gate/initial-anchors-3/runtime/src/shared/artifacts.mjs
Pair collection
- Function inventory
provenance/collector/function-inventory.mjs - Library ownership
provenance/collector/isolated-v1/library-scope.mjs - Pair relations
provenance/collector/isolated-v1/relations.mjs
Other components
- Dataset user
Connections 14
- Dataset user → Qualified corpus: uses corpus
README.md - Qualified corpus → Dataset index: indexed by
README.md - Dataset user → Reproduction inputs: uses inputs
README.md - Source preflight → Lock resolution: resolves packages
provenance/source-gate/source-functions-3/preflight.mjs - Source preflight → Function inspection: inspects files
provenance/source-gate/source-functions-3/preflight.mjs - Function inspection → Function measurement: measures functions
provenance/source-gate/source-functions-3/functions.mjs - POG anchors → Project inventory: creates inventory
provenance/source-gate/initial-anchors-3/runtime/components/pog/index.mjs - POG anchors → Function measurement: measures features
provenance/source-gate/initial-anchors-3/runtime/components/pog/index.mjs - POG anchors → Artifact checks: checks inventory
provenance/source-gate/initial-anchors-3/runtime/components/pog/index.mjs - Function measurement → Source parsing: parses source
provenance/source-gate/initial-anchors-3/runtime/src/analysis/measure.mjs - Pair relations → Library ownership: assigns ownership
provenance/collector/isolated-v1/relations.mjs - Pair relations → Function inventory: builds inventory
provenance/collector/isolated-v1/relations.mjs - Dataset audit → Function inspection: uses function hashes
tools/audit.mjs - Dataset audit → Function measurement: checks runtime
tools/audit.mjs