Skip to content

Migrate from v0.5.x to v0.6.x

v0.6.x adds the explanation engine (ontologos-explain), ontologos-bridge for horned-owl/oxrdf conversions, and RL/RDFS delegation to reasonable. OWL EL uses in-house completion in ontologos-el (restored in v0.6.1 after a brief whelk experiment that blocked crates.io publish). Public crate names (ontologos-el, ontologos-rdfs, ontologos-rl) are unchanged.

Cargo.toml

ontologos-core = "0.6.1"
ontologos-parser = "0.6.1"
ontologos-profile = "0.6.1"
ontologos-bridge = "0.6.1"   # new — only needed if you call adapters directly
ontologos-rdfs = "0.6.1"
ontologos-rl = "0.6.1"
ontologos-el = "0.6.1"
ontologos-query = "0.6.1"
ontologos-explain = "0.6.1"  # new

Python (alpha): pip install ontologos==0.6.1

Upgrading from v0.6.0 on crates.io

The v0.6.0 tag published only ontologos-core, ontologos-parser, ontologos-profile, and ontologos-query. v0.6.1 publishes the full library set including ontologos-bridge, ontologos-el, ontologos-rdfs, ontologos-rl, and ontologos-explain. Bump all ontologos-* crates to 0.6.1 together.

Breaking changes

MaterializationReport.traces → .trace

RDFS and RL reports use InferenceTrace on the .trace field (renamed from .traces).

Per-rule inference counts

MaterializationReport::inferred_by_rule and rdfs_inferred are empty under the reasonable adapter. Use inferred_total() / axiom counts for saturation metrics until reasonable exposes rule-level diagnostics.

EL vs RL/RDFS output model

  • EL (ElClassifier::classify, --profile el): returns Taxonomy without mutating the ontology.
  • RL/RDFS (RlEngine::saturate, RdfsEngine::materialize): merge inferred axioms back into Ontology.

Removed internals (facade APIs stable)

Custom RDFS/RL rule modules and RL TripleIndex are gone. EL completion internals live in ontologos-el again (not a public API). Do not depend on internal modules — use the public facades or call reasonable directly for RL/RDFS engine-specific features.

New features

  • ontologos explain CLI command and ontologos-explain crate (EL inference traces)
  • ontologos-bridge for core ↔ horned-owl/oxrdf conversions
  • RDFS reports include clashes from reasonable diagnostics

Conformance notes

Some HermiT Tier-A RL tests are #[ignore] where reasonable has known gaps (documented in dependency-first ADR). OntoLogos tracks upstream issues rather than reimplementing rules.

Pizza EL taxonomy is checked against a committed golden baseline (compare-pizza-el-golden.sh in CI). The in-house EL engine may omit some direct C ⊑ owl:Thing edges that other EL tools emit; see conformance.

Further reading