diff --git a/docs/designs/track-f-sysml-kerml-engagement.md b/docs/designs/track-f-sysml-kerml-engagement.md new file mode 100644 index 0000000..6e03e7e --- /dev/null +++ b/docs/designs/track-f-sysml-kerml-engagement.md @@ -0,0 +1,520 @@ +# Track F — SysML v2 / KerML community engagement strategy + +Status: **proposed** — synthesizing two parallel research streams (spar-sysml2 +audit + community state survey) into an actionable engagement plan. +Last update: 2026-04-25. + +Companion to issue #149 (Track D, TSN/WCTT) and #150 (Track E, migration +oracle). This track is **community-and-standards-engagement-led**, not +implementation-led; the technical roadmap follows from the engagement +positioning, not the other way around. + +## Executive summary + +Three findings reshape what we thought we knew: + +1. **`spar-sysml2` is production-grade, not a stub.** 7,167 LOC, zero + `todo!()` / `unimplemented!()` macros, 59+ tests, conformance tests + against official Systems-Modeling spec examples, lossless parsing + verified. Eight concepts are fully bidirectional (parse + lower + + extract + generate), including the **entire requirements roundtrip** + (`satisfy` / `verify` / `refine` / `allocate` / `derive`). That is + the single most important credibility marker for community engagement. + +2. **The OMG `Systems-Modeling/SysML-v2-AADL-Release` repo exists and is + skeletal.** Three commits, no tagged release, four named maintainers + from Galois (Hugues), CMU/SEI (Seibel + Wrage), and Ellidiss (Dissaux). + Goal: SAE+OMG joint AADL+SysML v2 standard for safety-critical + (ARP4754 / DO-178C/330/331 / DO-254). Flows + modes are explicitly + "not yet translated" per the README. **This is exactly where spar's + first contributions belong.** + +3. **spar is not duplicating an existing Rust effort.** Jade Wilson's + `syster-base` / `syster-lsp` (Microsoft, MIT, alpha, ~10 stars) is + the most mature Rust SysML v2 parser. It targets the SysML v2 side; + spar targets AADL with SysML v2 emit. Adjacent, not duplicate. Direct + collaboration on grammar conformance tests is plausible. + +The engagement plan therefore prioritizes the **AADL-Release repo** + the +**OMG RTESC working group** (whose output that repo is) as the primary +venue, with the Google Group + OMG Issue Tracker as supporting channels. +Eclipse SysON is a secondary track, lower priority. + +## §1 — spar-sysml2 audit (verified, 2026-04-25) + +### 1.1 Coverage matrix + +Grouped by status. Full audit data preserved in agent report archive. + +**Fully bidirectional** (parse + lower + extract + generate): + +| Concept | Notes | +|---|---| +| `part def`, `part usage` | Maps to AADL `system implementation` | +| `port def`, `port usage` | Maps to AADL `data port` / `event port` | +| `requirement def`, `requirement usage` | Lossless requirement roundtrip | +| `action def`, `action usage` | Maps to AADL `subprogram` | +| `state def`, `state usage` | Maps to AADL `mode` | +| `satisfy req by` | Requirement-to-architecture link | +| `verify req by` | Requirement-to-test link | +| `refine req by` | Requirement decomposition | +| `allocate task to processor` | AADL processor binding | +| `derive req from` | Requirement traceability | + +**Parse + Lower** (no extract / generate yet): + +| Concept | Why it matters | +|---|---| +| `connect`, `bind` | Lowered to AADL connection; `bind` aliased to `connect` | +| `interface def`, `interface` | Lowered to AADL feature group type | +| `attribute def`, `attribute` | Lowered to AADL data type | +| `enum def`, `enum` | Lowered with `Data_Model::Enumerators` property | +| `constraint def`, `constraint` | Lowered as timing-constraint property | +| `calc def` | Lowered to AADL subprogram | +| `allocation def` | Lowered to processor binding | +| specialization (`:>`, `:>>`, `specializes`, `subsets`, `redefines`) | Lowered to AADL `extends` | +| multiplicity (`[N]`, `[0..*]`) | Lowered to AADL array subcomponents | + +**Parse-only** (keywords recognized, no semantic action): + +- Behavioral: `transition`, `entry`/`do`/`exit` actions, `flow`, `succession`, `perform` +- Variants: `variant`, `variation def`, `select` +- Meta-modeling: `view def`, `viewpoint def`, `metadata def`, `annotation` +- Verification: `assert`, `verify` (as a keyword separate from the relationship verb) +- KerML: `feature def`, `class def` *(not recognized)*, `metaclass def` *(not recognized)* +- Type system: conjugation (`~Type`), end features, complex expressions + +**Missing entirely:** + +- `class def`, `metaclass def` — affects very little real modeling, but they're + KerML kernel concepts; should be addressed before any KerML contribution +- Arithmetic / function-call expressions — parsed as identifiers only + +### 1.2 Test coverage + +| Test file | LOC | Tests | Scope | +|---|---:|---:|---| +| `conformance_tests.rs` | 85 | 8 | Lossless roundtrip on official spec examples (Annex A SimpleVehicleModel + 6 others) | +| `validation_tests.rs` | 251 | 21 | End-to-end parse → lower → AADL on 20 numbered scenarios + 1 lossless guard | +| `fuzz_sysml2.rs` | 520+ | 30+ | Adversarial parse / lower / extract; no panics on malformed input | + +**Gaps the tests admit (worth filing as tracking issues):** + +- No tests for bidirectional roundtrip of behavioral content (state machines, flow specs) +- No tests for expression evaluation +- No tests for variant management constructs + +### 1.3 Public API + +```rust +parse(&str) -> Parse // CST +lower_to_aadl(&Parse) -> ItemTree // → AADL HIR +lower_to_aadl_with_diagnostics(&Parse) -> (ItemTree, Vec) +extract_requirements_list(&Parse) -> Vec +extract_all(&Parse, include_architecture: bool) -> ExtractionResult +extract_all_yaml(&Parse, include_architecture: bool) -> String // rivet YAML +parse_rivet_yaml(&str) -> Vec +generate_sysml2(&[RivetArtifact]) -> String // rivet YAML → SysML v2 +``` + +Consumed by `spar-cli` `extract` / `generate` / `parse` / `lower` commands. + +### 1.4 Spec-conformance signal + +- README and `lower.rs:1–29` document the SEI mapping table SysML v2 ↔ AADL. +- `conformance_tests.rs:1–3` references the official Systems-Modeling/SysML-v2-Release + example corpus. +- No version pinning to a specific OMG spec revision (KerML 1.0 / SysML 2.0 final + was adopted June 2025; spar's parser predates that, so a spec-version annotation + pass is a reasonable v0.8.x housekeeping commit). +- One known limitation noted at `lower.rs:276–279`: specialization-cycle detection + not implemented (deferred to AADL backend). + +### 1.5 What this means for engagement positioning + +- **Don't market spar as "a SysML v2 parser stub".** It's a working bidirectional + translator on the requirements + structural axis. +- **Do market spar as "AADL-side translator with full requirements roundtrip".** + That's the sentence that signals competence to the OMG RTESC WG. +- **Be honest about the behavioral gap.** State machines and flow specs parse + to keywords but don't lower to AADL behavior annex constructs. Calling this + out preempts the obvious reviewer question. + +## §2 — Standards & community landscape (verified, 2026-04-25) + +### 2.1 Spec status + +- KerML 1.0 + SysML v2 1.0 + Systems Modeling API 1.0 final adoption: **30 June 2025** + (announced 21 July 2025). +- Latest release packaged in `Systems-Modeling/SysML-v2-Release`: **2026-03** + (released 2026-04-10), 55 releases lifetime, 819 GitHub stars. +- KerML 1.1 / SysML 2.1 / Systems Modeling API 1.1 RTFs are **active** (no resolutions yet). +- A "round 2" FTF cleanup tracker (`SYSML2_-NNN`) is still being closed out. + +### 2.2 The Systems-Modeling GitHub org — verified inventory + +| Repo | Lang | Stars | Last update | Notes | +|---|---|---:|---|---| +| `SysML-v2-Release` | HTML+spec | 819 | 2026-04-17 | "Start here" — spec PDFs + installers + libraries | +| `SysML-v2-Pilot-Implementation` | Java + Xtext | 219 | 2026-04-22 | Reference impl; LGPL-3.0/GPL-3.0; Maven build; **no `CONTRIBUTING.md`**; issue #571 (LSP/tree-sitter, Jun 2024) has zero maintainer responses 22 months later | +| **`SysML-v2-AADL-Release`** | (n/a) | 8 | 2026-03-01 | **THE strategically critical repo — see §2.3** | +| `SysML-v2-API-Services` | Java | 83 | 2025-06-16 | Reference impl of API; ~10mo old | +| `SysML-v2-API-Java-Client` | Java | 17 | 2025-04-30 | | +| `SysML-v2-API-Cookbook` | Jupyter | 54 | 2025-03-10 | | +| `SysML-v2-API-Python-Client` | Python | 57 | 2021-10-14 | **Effectively abandoned (4.5 yr stale)** | + +### 2.3 `SysML-v2-AADL-Release` — strategic anchor + +- **What:** Domain extension library merging AADLv2 into SysML v2 — translates + "most AADLv2 core language features" to SysML v2 parts/ports/attributes. +- **Status:** 3 commits on master, no tagged releases, README explicitly notes + flows + modes are **not yet translated**. +- **Maintainers (named in README):** + - Jérôme Hugues — Galois + - Joe Seibel — CMU/SEI + - Lutz Wrage — CMU/SEI + - Pierre Dissaux — Ellidiss +- **Governance:** Output of OMG RTESC (Real-Time Embedded Safety-Critical) + working group — an SMC working group. Joint SAE+OMG effort. Stated + alignment goals: ARP4754, DO-178C, DO-330, DO-331, DO-254. +- **Related precedent:** Galois has previously released a SysML→AADL + bidirectional bridge (referenced on Galois LinkedIn). + +This is where spar contributes first. + +### 2.4 Community channels + +| Channel | Cost | What it gets | +|---|---|---| +| `groups.google.com/g/sysml-v2-release` | $0 (apply for membership) | Discussion with RIWG (Friedenthal, Seidewitz, et al.). Approval reportedly fast. **Public archive not available without joining.** | +| `github.com/Systems-Modeling/*` | $0 | Issue + PR submission. No NDA. | +| `issues.omg.org/issues/create-new-issue` | $0 (sign NCLA) | Spec issues against KerML 1.1, SysML 2.1, API 1.1 RTFs. JIRA login + Non-Member Contribution and License Agreement. | +| `github.com/eclipse-syson/syson` | $0 (sign ECA, 3-yr) | PR contributions to Eclipse SysON. Author email must match ECA-registered email. | + +### 2.5 Membership cost matrix (caveat: OMG fee page locked from external fetch) + +| Tier | Annual cost | Voting | SMC seat | Notes | +|---|---:|---|---|---| +| (Just file issues / PRs) | $0 (NCLA) | No | No | One-shot contributions, no WG seat | +| Google Group only | $0 | No | No | Discussion observer status | +| University Member | $550 | TF/SIG (no DTC/PTC) | Free | **Requires academic affiliation** | +| Trial Member | $2,150 once for 1 yr | No | Free | One named individual, time-limited | +| **Influencing Member** (≤$10M revenue tier) | $3,000 | TF/SIG (no DTC/PTC) | Free | Unlimited individuals, renewable, **right-sized for spar** | + +**One indirect search snippet** matched the tier brackets stated above +($3,000 / $5,500 / $11,000 / $21,500 / $37,500 by revenue band), but the +OMG fee page is locked behind authentication and could not be independently +fetched. **Recommend re-confirming with `accounting@omg.org` or via a +logged-in browser session before quoting the exact figures publicly.** + +SMC scope (verified via OMG's 2023-11-30 press release): the Systems Modeling +Community is **free for all eligible OMG member companies**, includes the +RIWG (Reference Implementation Working Group, runs the Google Group) and +the RTESC WG (the AADL+SysML v2 effort). + +### 2.6 Eclipse SysON — separate ecosystem + +- Project page: +- Repo: — 273 stars, 1,422 commits, + 125/8 open issues/PRs. +- Latest release: **v2026.3.0** (2026-03-25), flagged "Major release with API breakage". +- Maintainers: **Obeo + CEA List**. Obeo handles UX/product; CEA List leads + spec compliance for OMG. Powers the SysML v2 editing in Papyrus. +- 8-week release cadence, dual EPL-2.0 / LGPL-3.0. +- ECA: 3-year validity, free for individuals, copyright stays with author. +- Web-based modeler (browser, no install). Native Capella interop. + +**vs. OMG Pilot-Implementation:** Eclipse SysON moves much faster, has a real +contribution path (ECA), and ships a graphical+textual UI. The Pilot-Implementation +is the reference for spec conformance. spar engages both, weighted toward OMG. + +## §3 — Rust SysML v2 ecosystem (positioning) + +| Project | Author | Scope | Status | License | Stars | Last commit | +|---|---|---|---|---|---:|---| +| `syster` (umbrella) | Jade Wilson (Microsoft) | Multi-component ecosystem (parser+codegen+CLI+Python+LSP+VS Code+diagrams) | Alpha | **MIT** | 10 | 2026-02-20 | +| `syster-base` | Jade Wilson | Parser + AST + Salsa-driven HIR | Alpha | **NONE** ⚠️ | 5 | 2026-02-19 | +| `syster-lsp` | Jade Wilson | LSP server | Alpha | **NONE** ⚠️ | 1 | 2026-02-20 | +| `syster-codegen` | Jade Wilson | KEBNF → parser/lexer code generator | Alpha | MIT | 0 | 2026-02-14 | +| `syster-cli` | Jade Wilson | CLI wrapper | Alpha | MIT | 2 | 2026-02-24 | +| Sensmetry `Sysand` | Sensmetry | Rust SysML v2 package manager + registry | Active 2026 | MIT/Apache-2.0 | (varies) | active | +| `tree-sitter-sysml` | Community | tree-sitter grammar | Active 2026-03 | (varies) | (varies) | active | +| `kerml` (crate) | Community | Stand-alone | Lower activity | (varies) | (varies) | quiet | +| `sysml-parser` (crate) | Community | "Heavy construction" Mar 2025 | Stalled? | (varies) | (varies) | quiet | +| `sysml.rs` | artob | "🚧" marked | Unclear | (varies) | (varies) | quiet | +| **`spar-sysml2`** | PulseEngine | AADL-side parser + lower + extract + generate | Production | (workspace, MIT pending) | (in-tree) | active | + +### 3.1 syster — corrected picture (deeper inspection 2026-04-26) + +Initial Track F research saw "alpha, MIT, ~10 stars" and characterized +syster as a single Rust SysML v2 parser. Direct GitHub inspection showed +something more nuanced: + +- **Mixed licensing.** Umbrella `syster`, `syster-codegen`, `syster-cli` + are MIT. `syster-base` and `syster-lsp` — the actual parser and + language server — **have no LICENSE file.** Default copyright is + "all rights reserved", so they cannot be consumed by spar (or anyone) + legally without explicit permission. **This is a hidden blocker for + any code-level consumption.** +- **Activity slowed.** Last commits Feb 14–24, 2026, mostly + repo-restructure + submodule shuffling. Two-month quiet since. + Not abandonment, but momentum has cooled. +- **Architectural breadth.** README confirms full ecosystem: parser → + codegen → CLI → Python wrapper → LSP → VS Code extensions (language + client + modeller + viewer) → diagram core + UI. Far more ambitious + than spar-sysml2's scope. +- **KEBNF-driven codegen.** `syster-base`'s parser is *generated* from + KEBNF grammar files (the OMG-shipped grammar definition format) via + `syster-codegen`. spar-sysml2's parser is hand-rolled. **This is a + real architectural asymmetry** — auto-conformance to future SysML v2 + spec revisions is a property syster has and spar doesn't. + +### 3.2 spar's positioning (refined) + +*spar is the AADL-side Rust toolchain that produces SysML v2 artifacts +and round-trips requirements bidirectionally; complementary to `syster` +(which targets SysML v2 directly), to the OMG Pilot-Implementation +(Java/Xtext), and to Sensmetry's `Sysand` (package manager).* + +**Avoid:** any messaging that reads as "another Rust SysML v2 parser". +spar is on the AADL side and emits SysML v2; that's the differentiator. + +**Cooperation policy:** spar pursues active cooperation with `syster` +**only when it's a strategic asset**, not a courtesy. Default mode is +"stay code-aware, don't duplicate work spar doesn't need" (e.g. spar +will not build a VS Code modeller when syster has one). Active +cooperation is conditional on syster solving its license issue first +(see §4.2 below). + +### 3.3 Architectural lesson worth tracking — KEBNF codegen (v0.9.0+ consideration) + +`syster-codegen` (MIT) reads KEBNF grammar files and generates a Rust +parser. The OMG ships KEBNF for SysML v2 / KerML; using `syster-codegen` +or porting its approach into spar-sysml2 would give automatic +conformance to spec revisions. + +**Trade-off:** spar-sysml2's hand-rolled parser is mature and tested +(7,167 LOC, 59+ tests). Replacing it is a substantial rewrite; the +benefit is alignment to future spec revisions without manual port work. +Not a v0.8.0 candidate; revisit as a v0.9.0+ option once SysML v2 1.1 +ships and we see how often the KEBNF revisions land. + +## §4 — Strategic anchors + +### 4.1 Direct contribution opportunities (high-leverage, low-cost) + +1. **`SysML-v2-AADL-Release`** flows + modes mapping. The README explicitly + names them as gaps. spar already lowers many of these AADL constructs + in its own analyses; converting the mapping rules into the AADL-Release + repo's library form is a focused contribution. +2. **OMG issue tracker** — KerML 1.1 / SysML 2.1 RTFs are active. Recent + issues like the ones referenced in the research (e.g., `stakeholder-node` + not defined in BNF, `FlowConnectionDefinitions` violating KerML structure + restrictions) demonstrate the genre. spar's experience parsing the + official spec examples puts it in a position to surface similar issues. +3. **Pilot-Implementation issue #571** (LSP/tree-sitter request, 22 months + without maintainer response). Both spar's parser and `syster`'s LSP + exist; offering them as community options is a visible contribution. + +### 4.2 syster — minimum-viable engagement only + +Per §3.1's findings (mixed licensing, two-month commit slowdown), spar +pursues *minimum-viable* engagement with `syster`: + +- **One issue against `jade-codes/syster-base`** asking for an explicit + LICENSE file (MIT to match umbrella). Content-free, low-stakes, + unblocks any future cooperation. **Don't go further until this lands.** +- If/when license clarifies, *re-evaluate* — don't auto-engage. The + cooperation criterion is "complete strategic value to spar", not + community courtesy. +- Until then: stay code-aware, avoid duplicating work spar doesn't need + (don't build a VS Code modeller, don't reinvent KEBNF codegen for + v0.8.0). +4. **`syster-cli` issue #4** (`sysml.library` doesn't parse cleanly in + syster). spar has parsed the same library successfully in conformance + tests; cross-validation would help both projects. + +### 4.2 Tailwinds — DARPA / DoD / industry signals + +- **DARPA PROVERS / INSPECTA** (Collins Aerospace + CMU + Dornerworks + + UNSW + Kansas State / Hatcliff group) — extends Sireum HAMR to use + SysML v2 instead of AADL, generates code to **Rust + seL4 microkit** + among other targets. Direct US-defense validation that SysML v2 + AADL + + Rust + safety-critical is the right combination. +- DoDI 5000.97 (Digital Engineering) + Aug 2025 SysML v2 info sheet from + DoD CTO — public guidance on transitioning to v2. +- INCOSE + NDIA-TVC running 2026 MBSE symposia with v2 tracks. +- Tool maturity caveat: 2027–2028 is the production-tooling window per + industry consensus. The current period favors implementation-grade + contributors. + +## §5 — 30 / 60 / 90 day plan + +### Day 0–7 (this week) + +1. **Apply to `sysml-v2-release` Google Group** with the application text + in §7 below. ~5 minutes; days-to-hours approval. Provides a discussion + surface to the RIWG. +2. **Sign the Eclipse Contributor Agreement (ECA)** at . + Free, 3-year validity. No project commitment yet — just enables PRs + to Eclipse SysON if/when one becomes useful. +3. **Run the OMG Pilot-Implementation locally** following its README + (Eclipse 2025-12 + Maven). Translate one of spar's existing AADL + fixtures to SysML v2 via spar's `generate` command, then load into + the Pilot-Implementation's Jupyter kernel. Document what survives + the roundtrip and what doesn't. +4. **Watch `Systems-Modeling/SysML-v2-AADL-Release`** repo. Read the + commit history. Note the build process (which is barely documented + today — itself a contribution opportunity). + +### Day 8–30 (this month) + +5. **First OMG issue submission** via NCLA at issues.omg.org. Use spar's + parsing experience as the source. Specifically look for: BNF tokens + that aren't defined, examples in the spec that don't parse cleanly, + property semantics that are spec-ambiguous. spar's existing + `spec_gaps.md` memory is a draft list. +6. **First PR or issue against `SysML-v2-AADL-Release`** with a flow + or mode mapping rule. Even a draft is signal. +7. **Direct outreach to one of the four named maintainers** (Hugues at + Galois is the natural first contact given his SysML→AADL bridge + precedent). Subject: introducing spar as the AADL-side Rust toolchain + that complements their library work. Don't pitch — offer to discuss. + +### Day 31–60 + +8. **Cross-validate against `syster-base`** by parsing the same `sysml.library` + and comparing diagnostics. File a joint issue if both find the same + spec ambiguity. +9. **Attend at least one RIWG / RTESC meeting if invited** (the Google + Group post velocity should make this clear by then). +10. **v0.8.0 commit**: align spar-sysml2 with current spec rev. Add a + spec-version annotation to the parser ("aligned to SysML v2 1.0 + final, 2026-03 release") and document conformance status per + §1.1's coverage matrix. + +### Day 61–90 + +11. **Decision point: Influencing Member ($3,000/yr) via PulseEngine.** + Triggers if: at least one merged PR / accepted issue has landed, + AND at least 3 spar-filed RTF issues are accepted, AND v0.8.0 + Track D/E work demonstrates spar's strategic seriousness. +12. **Roadmap commit for v0.9.0+**: Track F technical milestones — + KerML kernel coverage (`class def`, expression grammar), behavioral + lower (state transitions, flow specs), variant lower + (`variation def`, `variant`, `select`). + +## §6 — Investment ladder & decision triggers + +``` +$0 path $3,000/yr path $X,XXX commitment +───────── ───────────── ──────────────── +NCLA + Google Group Phase 1: triggered by ≥1 Phase 2: triggered by ++ GitHub PRs AADL-Release PR or issue sustained engagement ++ direct outreach accepted by named maintainer. + voting weight desired ++ ECA (Eclipse parallel) Path: Trial Member ($2,150 in DTC/PTC. + one-time, no voting) for 1 yr Path: Domain or Platform + evaluation, then upgrade to Member tier (~$5,500+ + Influencing Member ($3,000/yr at <$10M revenue per + renewable, full TF/SIG voting, indirect snippet). + free SMC seat). +``` + +**Phase 1 trigger criteria (specific):** +- A PR or issue submitted by spar to `SysML-v2-AADL-Release` is accepted. +- A spar-filed issue at issues.omg.org is acknowledged by an RTF chair. +- spar's roundtrip output on at least one industry-relevant model is referenced + publicly (LinkedIn, conference, paper). + +**Phase 2 trigger criteria (longer horizon):** +- spar-as-PulseEngine has produced ≥2 specification influence outputs (PR + merged, issue accepted, paper presented at INCOSE / MODELS / FMICS). +- AADL-Release reaches its first tagged release with spar contributions visible. +- The SAE/OMG joint AADL+SysML v2 standard publishes its first draft. + +## §7 — Application text + +For the `sysml-v2-release` Google Group. Both lengths shipped — pick the +one that fits the form; the long version is preferred for setting up +direct credibility. + +### Long version (~180 words — recommended) + +> Name: Ralf Anton Beier +> Organizational affiliation: PulseEngine (pulseengine.eu) — independent +> open-source toolchain for safety-critical systems +> Interest in SysML: +> I maintain spar (github.com/pulseengine/spar), an open-source AADL v2.3 +> toolchain in Rust with 27+ analysis passes, deployment allocation with +> ASIL decomposition, and an LSP server. spar has a working SysML v2 +> parser-and-translator (spar-sysml2: 7,000+ LOC, lossless parsing, +> bidirectional roundtrip for the requirements domain via the +> satisfy/verify/refine/allocate/derive relationships, conformance +> tests against the official Systems-Modeling spec examples). +> +> I am also developing rivet (github.com/pulseengine/rivet), a +> schema-driven traceability validator for safety-critical SDLCs, +> where SysML v2's textual notation and the Systems Modeling API are +> directly relevant for AI-agent-driven workflows. +> +> My interest in joining is to follow the Reference Implementation work, +> understand spec ambiguities and edge cases as they're discussed, and +> contribute test models, OMG issues, and AADL-side mappings into the +> SysML-v2-AADL-Release repo. Background: 22 years of automotive software +> architecture, focused on formal verification, MBSE, and AI-assisted +> safety-critical development. + +### Short version (~70 words) + +> I maintain spar (an open-source AADL v2.3 toolchain in Rust with formal +> verification integration) and rivet (a schema-driven traceability +> validator), both at github.com/pulseengine. spar-sysml2 is a working +> bidirectional SysML v2 translator on the requirements domain. I would +> like to follow the Reference Implementation work, contribute mappings +> into SysML-v2-AADL-Release, and file OMG issues from spar's parsing +> experience. Background: 22 years automotive software architecture. + +## §8 — Risks & unknowns + +| Risk | Mitigation | +|---|---| +| OMG fee figures could not be independently fetch-verified (page locked) | Re-confirm with `accounting@omg.org` before quoting publicly | +| Google Group post velocity opaque without joining | Phase 0 application is itself the cost-of-information | +| `syster-base` (Microsoft, alpha) might accelerate | Quarterly re-check; if it does, spar's Rust angle weakens — but spar's AADL angle is not affected | +| Pilot-Implementation external contribution path is undocumented | Engage via Google Group + named maintainers, not via blind PRs | +| Influencing Member ($3,000/yr) is a real recurring cost | Phase 1 trigger criteria are explicit; don't upgrade without them | +| AADL-Release has only 3 commits — could stall | Outreach to maintainers de-risks this; if it stalls, pivot to direct OMG issue track | +| Behavioral / variant gaps in spar-sysml2 are real | Address in v0.8.x roadmap; messaging stays honest about the gap | + +## §9 — Open questions for the user + +1. **Re-confirm OMG fees** in a logged-in browser before any public quote. +2. **Decide $3,000/yr threshold** — is PulseEngine ready to budget that + if Phase 1 triggers fire? (Not a blocker; just sets the deciding + criterion when triggers arrive.) +3. **University-affiliation path** — is there a Fraunhofer / TU / industry + research collaboration that could make University Member ($550/yr) + viable? Best $/voting-power on the menu if so. +4. **Direct outreach order** — Hugues (Galois, AADL+formal-methods natural + match) → Seibel/Wrage (CMU/SEI, AADL canonical authors) → Dissaux + (Ellidiss, real-time/safety) is the recommended sequence. Confirm. +5. **Eclipse SysON priority** — secondary track. If a small first + contribution there (doc fix, test case) happens early, it broadens + the surface; if not, fine. + +## §10 — Appendix: agent-collected raw data + +Both research agents produced detailed reports during this session: + +- **spar-sysml2 audit** — full coverage matrix (SysML v2 + KerML core + + cross-cutting), test inventory, public-API surface, spec-conformance + notes. Available in agent-completion log. +- **Community state research** — verified Systems-Modeling repo inventory, + Pilot-Implementation deep dive, AADL-Release deep dive (the §2.3 above), + OMG Issue Tracker semantics, Eclipse SysON parameters, OMG fee tier + table (with auth-locked-page caveat), Rust ecosystem inventory, vendor + landscape, DoD/DARPA/AUTOSAR signals, ~50 sources cited inline. + +The synthesized doc above is the actionable derivative; raw data is +preserved in the run logs and in the project memory files +(`project_sysml_v2_engagement.md`, `project_post_v070.md`).