Skip to content

supported-features.md

Latest commit

 

History

History
124 lines (89 loc) · 4.71 KB

supported-features.md

File metadata and controls

124 lines (89 loc) · 4.71 KB

Supported Features & Limitations

This page is the contract for external users. It intentionally distinguishes between what is ready to evaluate now and what is still experimental or incomplete.

Maturity matrix

Capability Status Notes
Concurrent graph storage and traversal APIs Stable Core graph engine is the strongest part of the repo.
Embedded Database facade for Rust Stable Best current entry point.
Graph algorithms in the workspace Stable Multiple algorithms are implemented and tested.
SQL read subset Stable SELECT, filtering, ordering, limit, COUNT, and GROUP BY subset are implemented.
Cypher read subset Stable MATCH-style reads, filters, ordering, limits, counts, and path traversal subset are implemented.
Gremlin traversal subset Stable g.V(), has, out/in/both, values, count, limit, and ordering subset are implemented.
Document store Stable Collection-based CRUD plus nested-field filtering in-process.
Resource monitor and internal query metrics primitives Stable Useful internally, but not yet a full operator-facing observability surface.
HNSW vector index Experimental Implemented and integrated, but still part of a broader evolving AI/vector story.
Live queries / async graph events Experimental Present in the embedded API, not yet part of a hardened ops surface.
HTTP router Experimental Router exists and is tested, but the repo does not ship a standalone server runtime.
Python and Node bindings Experimental Available in the workspace, but not yet the primary documented adoption path.
Incremental embeddings / inference Experimental Real code exists, but parts of the AI path still contain placeholder/evolving pieces.
Cluster / sharding / HA Experimental Present in the workspace, but not yet documented as production-ready operations.
Docker-first deployment story Planned Not the current entry path.
Production-grade operator controls (timeouts, admission control, slow-query tooling) Planned Internal primitives exist, external surface is still limited.

Query language coverage

BikoDB should currently be understood as supporting documented subsets, not full language compatibility.

SQL

Supported subset today includes:

  • SELECT ... FROM ...
  • WHERE
  • ORDER BY
  • LIMIT
  • COUNT(*)
  • GROUP BY
  • predicates such as =, !=, >, <, >=, <=, LIKE, IN, BETWEEN

Important limitation:

  • the parser contains broader work such as INSERT and DELETE,
  • but the current Database::query_sql() execution path is a read path,
  • so you should not treat SQL writes as a public, end-to-end supported query surface yet.

Cypher

Supported subset today includes:

  • MATCH
  • OPTIONAL MATCH
  • WHERE
  • RETURN
  • ORDER BY
  • LIMIT
  • count(...)
  • variable-length path syntax for supported traversal cases

Important limitation:

  • parser work exists for CREATE, DELETE, and SET,
  • but the public Database::query_cypher() path is currently a read-oriented execution path.

Gremlin

Supported subset today includes:

  • g.V()
  • hasLabel(...)
  • has(...)
  • out(...), in(...), both(...)
  • values(...)
  • count()
  • limit(...)
  • order().by(...)

Important limitation:

  • this is a traversal subset, not full Apache TinkerPop compatibility.

Multi-model scope

What exists today:

  • graph nodes and edges,
  • document collections,
  • HNSW vector search,
  • helper methods that bridge graph, documents, and vectors.

What you should not assume today:

  • a single public transaction API spanning graph + document + vector state,
  • documented conflict semantics across all models,
  • versioned cross-model isolation guarantees.

See multi-model consistency for the current boundaries. See supported guarantees for the shortest summary of what is safe to claim publicly.

Operational limitations

Useful internal pieces already exist:

  • ResourceMonitor for counters and snapshots,
  • AccessTracker for per-operator/query metrics,
  • EXPLAIN formatter in the optimizer.

But the repo does not yet present a complete operator-facing layer for:

  • slow query reporting,
  • query profiling,
  • timeout/cancel controls,
  • admission control and concurrency governance.

See operations & observability for the current operator-facing picture.

Packaging limitations

External users should know this up front:

  • there is no official standalone BikoDB server binary at the repo root,
  • there is no first-class Docker deployment flow in the repo,
  • the best current starting point is still the embedded Rust API.