Orbit-RS maturity
Status is a label earned by a checklist (ADR 0015, Proposed). Compatibility is a measured pass rate (ADR 0002); performance is stated only with a reproducible benchmark (ADR 0016, Proposed). Figures come from scripts/project_facts.py (ADR 0017, Proposed).
What the labels mean
| Label | Requires |
|---|---|
| Experimental | The default. No conformance corpus, or one not run in CI. May change without notice. |
| Preview | A corpus for the target version that the real server passes 100%, run in CI with a ratchet; durable storage with a recovery test; for a replicated component, a test that kills the leader in use; authentication on by default. |
| GA | Preview, plus: corpus at 100% (or the failures listed by name); a tested upgrade path; a tested backup and restore with the standard tools; a soak or chaos run of at least 24 hours; published benchmarks; a recorded security review. |
Components as of 2026-10-27
| Component | Label | Evidence | Not yet met |
|---|---|---|---|
| PostgreSQL wire and durable SQL engine | Preview | Conformance floor 789 of 816 records (96.7%) against the PostgreSQL 18 corpus, which PostgreSQL 18.6 passes 100%; make conformance ratchet; RocksDB persistence; Raft replication with a leader-kill test (raft/sql_tests.rs); schemas, roles and privileges (ADR 0003, 0012, 0013); SCRAM login required unless --dev-mode |
Corpus below 100% and not covering the extended protocol, COPY, ORM traffic, pg_dump, psql \d* or concurrency; pg_dump fails; no 24-hour soak; no published benchmarks |
| MySQL (target LTS 9.7) | Experimental | corpus of 300 records from MySQL 9.7.4: 60 pass (20.0%), make conformance-mysql |
corpus small; most functions, DDL and information_schema unimplemented |
| CQL (target Cassandra 5.0) | Experimental | corpus of 208 statements from Cassandra 5.0.9: 63 pass (30.3%), make conformance-cql |
corpus small |
| AQL (target ArangoDB 3.12) | Experimental | corpus of 258 queries from ArangoDB 3.12.12: 94 pass (36.4%), make conformance-aql; off by default |
collections, graph traversal |
| MongoDB (target 8.3) | Experimental | corpus of 134 commands from MongoDB 8.3.13: 59 pass (44.0%), make conformance-mongodb |
corpus small |
| OrbitQL (specification 1.0) | Experimental | none | no written 1.0 specification or suite |
| Redis RESP (target Redis 8, with modules, ADR 0018) | Experimental | corpus of 666 records from Redis 8.10.2: 247 pass (37.1%), make conformance-redis |
no keyspace engine (ADR 0018); modules not started |
| Cypher/Bolt (target Neo4j 2026.10) | Experimental | corpus of 115 statements from Neo4j 2026.10.1: 9 pass (7.8%), make conformance-cypher |
negotiates Bolt 4.4 only; Bolt 5.x clients cannot connect |
| HTTP REST, gRPC | Experimental | unit tests | no specification-derived suite |
| Storage tiers other than RocksDB | Experimental | unit tests | recovery tests |
The PostgreSQL line is Preview and not GA because the corpus is below 100% and pg_dump fails. Until ADR 0015 is accepted by the owner, these labels are proposals.
Performance
Not yet benchmarked. The earlier figures (message throughput, memory, binary size, cold start, P99 latency and comparisons with JVM systems) were removed because no benchmark backed them. Design targets may be listed as targets. See ADR 0016 for what a result must state.
Figures
Run scripts/project_facts.py. At commit 222ac1c2 on 2026-10-25: 3,986 test functions, 457,233 non-blank lines of Rust, PostgreSQL conformance floor 768 of 816 (94.1%).