-
Notifications
You must be signed in to change notification settings - Fork 0
OMG SysMLv2
Status: Anchoring review. This page elaborates the scope-reducing assumption stated in Design Spec §3 — that the modeled system is a SysMLv2 model conformant with OMG specifications, and that
omg-sysml:IRIs (openCAESAR OWL rendering) are the canonical model vocabulary inflexo-rtm's RDF graph.
The Systems Modeling Language version 2 (SysMLv2) is an Object Management Group (OMG) standard for model-based systems engineering. The OMG approved final adoption of SysML v2.0 in July 2025 (alongside KerML v1.0 and the SysML v2 API & Services specification); editorial updates in March 2026 prepared the spec for ISO submission. This wiki anchors on the OMG SysML v2.0 final spec.
SysMLv2 is not a cosmetic refresh of SysML 1.x. It is a ground-up redesign with three properties that matter for flexo-rtm:
-
Textual concrete syntax. SysMLv2 models are authored as
.kermland.sysmltext files. The Kernel Modeling Language (KerML) provides the foundational vocabulary; SysML proper is a KerML extension. Textual authoring makes models diff-able, mergeable, and amenable to ordinary git-based workflows — a precondition for the git+RDF foundation described in Design Spec §1. - MOF-based abstract syntax (metamodel). Behind the textual syntax sits a Meta-Object Facility metamodel that defines the abstract structure of every well-formed SysMLv2 element: parts, ports, actions, constraints, requirements, analysis cases, and so on. The metamodel is what makes interoperability tractable.
-
Standardized serializations and API. Models can be serialized to
.sysml.json(the JSON projection defined by the SysML v2 API & Services specification) and accessed through a REST API that the SysMLv2 Pilot Implementation, Magic Systems of Systems Architect, and other tools share. The API is the integration substrate.
For flexo-rtm, "SysMLv2" refers to this triple (textual .kerml/.sysml source, MOF metamodel, .sysml.json / API projection) collectively, as a family of interoperability data standards — not as flexo-rtm's internal model.
The canonical model in flexo-rtm is RDF. flexo-rtm is RDF-native: storage is named graphs of omg-sysml: triples (per the openCAESAR OWL rendering described below), analysis reasons over those triples, and the certification artifact is RDF. The three SysMLv2 file formats above are interface contracts: flexo-rtm accepts each as input (every accepted format maps deterministically into canonical RDF), and produces each on demand from canonical RDF when a downstream tool requires that format. They define the I/O boundary; they do not define the internal model.
This RDF-native posture matters for two reasons. First, interop is by deterministic mapping, not by storage choice — flexo-rtm is not constrained to whichever single file format any one tool prefers, and adding a new accepted format means writing a new mapping, not changing the storage. Second, lossless roundtrip is a property we can check — every accepted file becomes a deterministic RDF graph, and every emission is a deterministic mapping back; the lossless criterion (RDFC-1.0 triple-set equivalence on canonical content, per Lossless Roundtrip Definition) is enforceable on every roundtrip. The same RDF-native pattern is what makes the OSLC adapter (OSLC RM Adapter Contract) work: file inputs in OSLC formats become RDF inside flexo-rtm, and the lossless guarantees follow.
Design Spec §3 frames anchoring on SysMLv2 as a scope-reducing assumption, not a bet. The reasoning is straightforward:
-
Broad but well-specified category. SysMLv2 covers structure, behavior, requirements, constraints, analyses, verifications, and views in one specification. Anchoring there gives
flexo-rtma generous slice of MBSE practice without committing to any tool vendor's idiosyncratic dialect. - OMG governance (open standard). The specification is normative, publicly available at omg.org/spec/SysML/2.0, and on a path to ISO standardization. Adopters are not locked into a vendor's metamodel; the metamodel itself is a public artifact under change-controlled governance.
-
openCAESAR OWL rendering — first-class RDF interop. openCAESAR's "Implementing SysML v2 Ontology" effort produces an OML (Ontological Modeling Language) vocabulary equivalent to the MOF-based SysMLv2 metamodel via a MOF2OML adapter, which in turn renders to OWL2-DL via openCAESAR's
owl-adapter. The result is anomg-sysml:namespace under which every SysMLv2 metaclass appears as an OWL class and every SysMLv2 relationship appears as an OWL property. This is the substrate that letsflexo-rtmkeep models as RDF without inventing a parallel SysMLv2 vocabulary. -
The ADCS prototype already uses this pattern.
structural/satellite.ttlandstructural/parameters.ttlin the ADCS lifecycle demo (see ADCS Prototype Lessons) declare SysMLv2 elements assysml:PartDefinition,sysml:PartUsage,sysml:RequirementDefinition,sysml:AttributeUsage,sysml:ConstraintDefinition, and so on, with thesysml:prefix bound tohttps://www.omg.org/spec/SysML/2.0/. The prototype's local alias maps to openCAESAR'somg-sysml:namespace. This pattern is the empirical proof point: the anchor works in practice.
The alternative — defining flexo-rtm's own RDF vocabulary for systems modeling — would either reinvent SysMLv2 badly or constrain adopters to a non-standard dialect. Anchoring closes that surface area.
flexo-rtm v0.1 reads, writes, and reasons over the following SysMLv2 constructs as omg-sysml: RDF:
-
omg-sysml:Part— system parts. SysMLv2 distinguishesPartDefinition(the type) fromPartUsage(the instance/role). The ADCS corpus uses both:adcs:ReactionWheelDef(definition) andadcs:ReactionWheel_X,_Y,_Z(usages typed by the definition). -
omg-sysml:Action— behavior and process. Actions are SysMLv2's primary behavior primitive; they compose into action sequences, decisions, and concurrent flows.flexo-rtmreferences actions when an artifact attests to behavior (a simulation run, a test execution, a control law evaluation). -
omg-sysml:Constraint— formal constraints.ConstraintDefinitionandConstraintUsageexpress predicates over part attributes (e.g., the ADCS prototype'sPDControllerDefis asysml:ConstraintDefinitionthat definestau = -Kp/2 * theta_error - Kd * omega). Constraints are how engineering math enters the model. -
omg-sysml:RequirementUsage— requirement instances. WhereRequirementDefinitiondeclares the requirement's text and parameters,RequirementUsageinstantiates the requirement in a specific context (a part decomposition, a verification case, a satisfy relationship). The ADCS corpus usessysml:RequirementDefinitiontogether withsysml:SatisfyRequirementUsageto bind a requirement to its satisfying elements. -
Connection points and ports.
omg-sysml:Port(typed byPortDefinition) and connection usages (omg-sysml:ConnectionUsage,omg-sysml:InterfaceUsage) describe how parts couple. The certification surface uses ports to scope analyses to interface contracts.
These five constructs cover the bulk of structural-trace and requirements-trace queries the oracle issues. Behavior round-trip (full Action semantics) is intentionally narrower — see open questions.
flexo-rtm supports two equivalent representations for requirements, and the certification predicate is agnostic between them:
-
Native SysMLv2. A requirement is an
omg-sysml:RequirementDefinition/omg-sysml:RequirementUsagecarried verbatim from the SysMLv2 source. Thesysml:SatisfyRequirementUsagepattern (used in the ADCS corpus) attaches satisfying elements directly. Nortm:Requirementwrapper is introduced. -
flexo-rtmbridge. A requirement is anrtm:Requirementinstance that references the underlying SysMLv2 element viartm:appliesTo. Thertm:Requirementcarries the certification-relevant predicates (rtm:satisfies, attestation links, scope membership) while the SysMLv2 element retains its native semantics.
Both representations are first-class. The basic certification predicate (Design Spec §4.1) computes forward and backward coverage uniformly over either form. The OSLC adapter (Alignment Strategy) prefers the bridge representation because OSLC-RM does not know about SysMLv2 natively — rtm:Requirement aligns cleanly with oslc_rm:Requirement while still carrying a back-pointer to the SysMLv2 element. Adopters who never round-trip to OSLC can stay with the native form.
v0.1 ships the certification core against models already in RDF. SysMLv2 ingestion and emission land in v0.2+:
-
Ingest. Parse
.kermland.sysml.jsonand store the result asomg-sysml:RDF using the openCAESAR OWL rendering. The pipeline is: SysMLv2 textual source → SysMLv2 Pilot parser (or equivalent) → MOF abstract model → MOF2OML/OWL adapter →omg-sysml:triples in the named graph for structural data. -
Emit. RDF →
.kermland.sysml.json. The inverse path uses the same metamodel mapping in reverse, plus a textual unparser. Output is intended for ingestion into SysMLv2 modeling tools (the Pilot Implementation, Magic Systems, future commercial implementations). - Lossless criterion. Identical in spirit to the OSLC adapter (Alignment Strategy): a round-trip is lossless iff the input and output triple-sets are equivalent under RDFC-1.0 canonicalization (W3C RDF Dataset Canonicalization). Whitespace, blank-node renaming, statement ordering, and prefix choice are all permitted to vary; semantic content must not. The conformance suite enforces this on a regression corpus.
This is the minimum surface for flexo-rtm to participate in a SysMLv2 toolchain without forcing adopters to abandon their authoring tools.
The SysMLv2 implementation landscape flexo-rtm targets for interoperability:
-
SysMLv2 Pilot Implementation (Eclipse-based, open source on GitHub at
Systems-Modeling/SysML-v2-Release). The reference implementation; ships the textual parser, the API server, and Jupyter integration. This is the primary regression target. - Magic Systems of Systems Architect (Dassault / No Magic). Commercial; supports SysMLv2 export. Interop target for industrial adopters.
-
openCAESAR tooling (
owl-adapter, OML workbench). The bridge between SysMLv2's MOF metamodel and the RDF worldflexo-rtminhabits. Operationally co-evolves withflexo-rtm's ingest path. -
Flexo MMS (the broader Flexo SysML v2 Team's OWL2-DL ontology work).
flexo-rtmshares vocabulary with Flexo MMS where the SysMLv2 ontology surfaces overlap; divergence is a co-evolution concern.
-
Stability of openCAESAR OWL rendering vs. SysMLv2 spec evolution. The OMG spec is now final but will see point revisions; the openCAESAR rendering is driven by the MOF2OML adapter and may lag or lead.
flexo-rtmpins to a specific openCAESAR rendering version per release and treats version skew between (SysMLv2 spec, openCAESAR rendering,flexo-rtmontology) as a conformance-suite concern. -
Behavior model semantics in RDF. Full round-trip of
Action, state machines, and time-varying behavior over RDF is non-trivial — the metamodel encodes operational semantics that OWL-DL does not natively express. v0.1 carries behavior structurally (the metaclass instances exist as RDF) but does not attempt to reason over behavioral semantics. Whether to lift behavior into SHACL rules, SPARQL property paths, or a separate semantic layer is open. -
Granularity of the
rtm:appliesTobridge. When the bridge representation references a SysMLv2 element, should it reference theRequirementDefinition, a specificRequirementUsagein a containment context, or both? The OSLC adapter's behavior is sensitive to this choice; the regression corpus needs to nail it down.
flexo-rtm's regression corpus needs SysMLv2 example models covering the metaclass range above. Candidate sources:
- OMG-published SysMLv2 sample models distributed with the spec and the Pilot Implementation (the Flashlight Starter Model is one well-known small example).
- openCAESAR example projects that exercise the OWL rendering end-to-end.
-
ADCS lifecycle demo (the
structural/directory). Already encoded asomg-sysml:RDF; useful as a hand-curated fixture against which round-trip from a future.kermlemission can be verified.
Each fixture in the corpus carries the SysMLv2 source, the expected omg-sysml: RDF (canonicalized), and a manifest noting which constructs it exercises.
- Flexo Git Coexistence
- ADCS Prototype Lessons
- MVC Pattern from RIME TRL ANT
- Human-AI Accountability
- Multi-Agent Discourse Graph Precedent
- OSLC RM and QM Review
- INCOSE V2 Review
- OMG SysMLv2
- PROV EARL GSN P-PLAN
- Dragon Architecture and Mission Enterprise
- Traditional Forward and Backward Analysis
- Attestation Infrastructure in v0.1
- Identity Boundaries and Policy Projections
- External URI References
- Signed Envelopes and Established Standards
- Aspect Coverage with Adequacy and Sufficiency
- Federated Audit and Composition
- Certification Predicate
- Gap Taxonomy
- Quantitative Outcomes
- Engineering Lifecycle Stages (v0.2)
- Topological Framework Future Work (research phase)
- Vertices Edges Faces (research phase)
- Three-Layer Architecture
- Operational Layer UX Discipline
- Storage Layer Flexo Conventions
- Analysis Layer Scope Algebra
- OSLC Roundtrip Acceptance
- Identity Adapter Contract
- Flexo REST Binding
- SysMLv2 Ingestion Contract
- External URI Rules
- Signed Envelope Shapes
- Parsimony Manifest
- Lossless Roundtrip Definition
- Vendor Extension Carry-Through
- OSLC RM Adapter Contract
- OSLC QM Adapter Contract
- ADR Template
- ADR-001 Foundations First Approach
- ADR-002 SysMLv2 Anchoring
- ADR-003 Topological Framework Documented as Future Work
- ADR-003a v0.1 Ships Traditional Analysis Only
- ADR-004 Quantitative Certification Outcome
- ADR-005 Adequacy and Sufficiency as Guidance Subtypes
- ADR-006 Three-Layer Architecture
- ADR-007 Scope as First-Class RDF Resource
- ADR-008 Repo Name and Org Transfer Plan
- ADR-009 Two-Repo Strategy
- ADR-010 OSLC-RM and OSLC-QM in v0.1
- ADR-011 Lossless Criterion A plus C
- ADR-012 Direct RDF Properties over Reified Edges
- ADR-013 Simplicial Complex as Derived View When Built
- ADR-014 Parsimony Layer Build-Time Extraction
- ADR-015 GSN Adoption for Adequacy and Sufficiency
- ADR-016 Composable SHACL Profiles
- ADR-017 knowledgecomplex as Optional Extras
- ADR-018 V minus F Invariant Deferred with Topological Framework
- ADR-019 Derived Binary View from Quantitative Metrics
- ADR-020 Vocabulary Alignment with Zargham 2026
- ADR-021 Three Attestation Subclasses Ship in v0.1
- ADR-022 External URI References as Open-Source Foundation
- ADR-023 Cryptography by Composition of Battle-Tested Standards
- ADR-024 Identity by Thin Projection of External Sources
- ADR-025 Reproducibility is Structural and Local
- ADR-026 Cryptographic Agility via Algorithm Profiles
- ADR-027 Bit-Exactness vs Numerical Tolerances Are Both First-Class
- ADR-028 Scope-Level Adequacy and Sufficiency for Federated Audit
- ADR-029 Engineering Lifecycle Stages as Scope Metadata
- ADR-030 Polycentric ASOT Authority Model
- ADR-031 Attestation Status Pass Fail Deferred Deprecated
- ADR-032 Methodology Agnosticism as Foundational Axiom
- ADR-033 Generalized ASOT Principle for All Identified Things