Skip to content

Latest commit

 

History

History
66 lines (59 loc) · 2.92 KB

File metadata and controls

66 lines (59 loc) · 2.92 KB

VeriSimiser Roadmap

Phase 0: Scaffold (COMPLETE)

  • ✓ RSR template with full CI/CD (17 workflows)

  • ✓ CLI with subcommands (init, start, drift, provenance, history, status, octad)

  • ✓ Manifest parser (verisimiser.toml with tier1/tier2 config)

  • ✓ Tier 1 data types (DriftReport, ProvenanceRecord, TemporalVersion)

  • ✓ ABI module stubs (Idris2 + Zig FFI)

  • ✓ README with two-tier architecture and honest framing

Phase 1: PostgreSQL Tier 1 MVP

  • ❏ PostgreSQL logical replication interception

  • ❏ Provenance sidecar (SQLite) — write-path observer

  • ❏ SHA-256 hash-chain integrity for provenance records

  • ❏ Temporal versioning sidecar — point-in-time queries

  • ❏ Cross-modal drift detection — read-path observer

  • ❏ Drift index with 8-category classification

  • ❏ Idris2 ABI proofs: sidecar isolation, hash-chain integrity, version ordering

  • ❏ Zig FFI bridge: database connection, overlay operations, VCL-total queries

  • ❏ End-to-end test: PostgreSQL → verisimiser overlay → VCL-total query

Phase 2: Multi-Backend Support

  • ❏ SQLite interception via sqlite3_update_hook / WAL monitoring

  • ❏ MongoDB interception via change streams

  • ❏ Redis interception via keyspace notifications

  • ❏ MySQL interception via binlog CDC

  • ❏ Application-level middleware / ORM hooks

  • ❏ Backend-agnostic interception trait abstraction

  • ❏ Per-backend integration tests

Phase 3: Tier 2 Overlays

  • ❏ Graph overlay (RDF triples / property graph edges)

  • ❏ Vector overlay (HNSW embedding similarity search)

  • ❏ Tensor overlay (ndarray multi-dimensional numeric data)

  • ❏ Semantic overlay (CBOR type annotations + proof blobs)

  • ❏ Document overlay (Tantivy full-text search)

  • ❏ Spatial overlay (R-tree geospatial coordinates)

  • ❏ Independent enable/disable per overlay via manifest

Phase 4: VCL-total Integration

  • ❏ VCL-total type-safe query parsing

  • ❏ Cross-tier queries (Tier 1 + Tier 2 in single query)

  • ❏ TypedQLiser integration for compile-time query validation

  • ❏ Query planner for multi-sidecar operations

  • ❏ Performance benchmarks: overhead of augmentation layer

Phase 5: Production Hardening

  • ❏ Retention policies (auto-prune temporal history)

  • ❏ Sidecar compaction and garbage collection

  • ❏ Concurrent access safety (multi-writer provenance chains)

  • ❏ Backup and restore for sidecars

  • ❏ Monitoring and alerting integration

  • ❏ Error recovery and graceful degradation

  • ❏ Shell completions (bash, zsh, fish)

Phase 6: Ecosystem

  • ❏ PanLL panel for drift monitoring dashboard

  • ❏ BoJ-server cartridge (MCP integration)

  • ❏ SqueakWell integration (database recovery via cross-modal constraint propagation)

  • ❏ Migration tooling: Tier 1 → Tier 2 → full VeriSimDB

  • ❏ Publish to crates.io

  • ❏ Chainguard container image