Skip to main content

Roadmap

Nodal is developed in explicit priority rounds. A round describes an intended outcome, not a compatibility or release-date promise. Shipped behavior remains defined by the provider compatibility matrix and package release notes.

Delivered foundations​

P0 delivered the query, tracking, mutation, and migration foundation. P1 delivered typed routes, centrality and community contracts, native Neo4j shortest paths, weighted GDS path compilers, TigerGraph installed-query execution, runtime capability discovery, Neo4j projection lifecycle management, typed options, and compiled analytics factories.

Completed - analytics package boundary correction​

  1. Renamed the analytics package boundary to Nodal.Analytics before the beta line. The current similarity kernel and future path, community, centrality, pattern, temporal, vector, and execution surfaces under the broader analytics boundary. Preserve Nodal.PatternRecognition as a temporary transition package, mark it deprecated on NuGet with Nodal.Analytics as its suggested replacement, and update source, tests, benchmarks, samples, documentation, capability metadata, and publishing workflows together. Treat Nodal.Semantics as a separate future package for ontology, taxonomy, entity resolution, inference, provenance, and knowledge-graph concerns rather than overloading the analytics package.

Next - Sonar quality gate​

  1. Add SonarCloud/SonarQube analysis to the protected CI path. Import the Coverlet coverage report, configure a repository quality profile and quality gate, publish the result on pull requests, and make the gate a required branch check alongside formatting, tests, coverage, package validation, and live provider smoke tests.

P2 - production hardening​

  • Certify Neo4j GDS and TigerGraph installed-query behavior against explicit server versions in live CI.
  • Harden public APIs through real application samples and cross-provider compatibility tests.
  • Expand migration diffing, rollback planning, history, and lifecycle tooling.
  • Define analytics write and mutate semantics without bypassing tracking and unit-of-work guarantees.
  • Add event-driven mutation, outbox, idempotency, and delivery integration points.
  • Publish versioned capability metadata for documentation, tools, and coding agents.
  • Establish versioned documentation and a stable package promotion policy.

P3 - Nodal.Analytics alpha​

P3 introduces an optional package for discovering repeated structures and temporal behavior without changing the responsibilities of Nodal.Core.

Foundation​

  1. Record the package boundaries and execution architecture in an ADR.
  2. Complete the Nodal.Analytics package and dependency rules after the beta boundary correction.
  3. Define canonical path, pattern, observation, and result contracts.
  4. Define event time, observation time, validity time, sessions, and windows.
  5. Add a provider-neutral pattern DSL and canonical analysis plan.
  6. Enforce bounded extraction, path budgets, cancellation, and sampling.

Discovery​

  1. Implement canonical path signatures and stable hashing.
  2. Implement typed positional multi-hot bitsets, XOR/popcount distance, and n-gram similarity kernels with scalar and hardware-accelerated paths.
  3. Implement weighted heterogeneous PathSim/HeteSim-style similarity and dense vector scoring as an optional lane.
  4. Build a sparse top-k similarity graph without an all-pairs materialization, and benchmark exact and approximate candidate strategies.
  5. Detect path communities and select representative paths.
  6. Mine bounded frequent path and event-sequence patterns.

Temporal and provider execution​

  1. Build directed temporal transition graphs with windows and time decay.
  2. Add Neo4j planning and native analytics pushdown where semantics match.
  3. Add TigerGraph planning and GSQL pushdown where semantics match.
  4. Persist versioned patterns, evidence, scores, and explanations explicitly.

Productization​

  1. Add property, differential, live-provider, performance, and load tests.
  2. Publish benchmarks, an e-commerce showcase, documentation, and the first alpha package.

Provider expansion​

  • Nodal.Memgraph: first candidate because local Docker, Bolt, and Cypher make an early compatibility slice practical. It receives its own capability profile; Neo4j compatibility is never assumed implicitly.
  • Nodal.ArangoDB: second candidate because HTTP and AQL exercise a genuinely different transport, graph model, traversal compiler, and transaction model. The first slice targets bounded query, traversal, materialization, and live Docker QA before mutations and migrations.

Additional providers are selected only after a transport, query-language, transaction, migration, local-QA, and maintenance assessment.

P4+ - research horizon​

  • Full frequent-subgraph mining and closed/maximal pattern reduction.
  • Overlapping and evolutionary community detection.
  • Concept-drift detection and automatic window adaptation.
  • Learned path embeddings and approximate vector indexes.
  • Probabilistic next-event and temporal point-process models.
  • Temporal graph neural networks and online model training.
  • Model registry, evaluation, rollback, bias, and drift governance.

The dated execution plan is published in the Nodal Journal. Track implementation status in the GitHub repository.