Home / Docs / Case studies — overview

Warp Commerce Model — Domain Case Studies (reconciled at canonical v1.0.0)

These are the adversarial test corpus referenced in the Commerce Model spec’s Formal Sufficiency Test — now executable against the canonical schema.

The spec claims the five primitives were “tested adversarially across 22 commerce domains.” This directory turns that claim into evidence: one worked case study per domain — 22 domains, 22 executable fixtures — each backed by an executable canonical scene fixture that validates + audits clean against the canonical Warp Commerce Model schema v1.0.0 (schema/structure/*.schema.json) via the canonical runner.

node conformance/runner/run.mjs            # 51/51 (29 core + 22 case studies)
node conformance/case-studies/validate-aux.mjs   # 5/5 auxiliary records vs canonical

What “reconciled” means

The case studies were authored by Agent D (PR #1) against a bespoke minimal schema and a standalone audit.mjs. Both are superseded. This corpus was re-pointed at the canonical schema:

  • Each domain is re-authored as a canonical scene fixture (PascalCase .type states, numeric Money, subject.{offered,requested}, {kind,payload} envelope) by conformance/case-studies/_generate.mjs.
  • They are judged by the canonical runner — the same one that judges the core conformance suite. No bespoke schema or runner is used.
  • Auxiliary records (AuctionProcess incl. ScoredSelection, AwardProtest, ResolutionProcess, EntitlementConsumption) are validated against the canonical schema by conformance/case-studies/validate-aux.mjs.

The per-domain markdown files below are D’s original narratives (lifecycle walkthroughs). Where a markdown says a construct is “UNREPRESENTABLE in schema v1.0.0” or “pending-v1.1”, that was true of D’s bespoke schema, not the canonical one — see the banner at the top of each file and the Findings below.

Result

Domains documented + lifecycle-walked 22
Executably validated against canonical v1.0.0 22 / 22
Blocked / pending-v1.1 (cannot express against canonical) 0
Unmodelable by the five primitives 0
Auxiliary records validated vs canonical 5 / 5

The honest split is 22 conformant / 0 pending / 0 unmodelable — because the canonical schema already expresses every v0.3 construct D had flagged as a gap. Two genuinely new follow-ups surfaced (BACKLOG B-2, B-3); neither blocks any domain.

Status table — domain → validates v1.0.0? → signature constructs exercised

All fixtures are kind: scene, expect: accept, under conformance/case-studies/<domain>/<domain>.json.

Domain Validates v1.0.0 Signature canonical constructs exercised
physical-ecommerce ✅ base five primitives; return as new role-reversed Commitment (I-2)
gifting ✅ parent→3-children tree (I-6 sum); ResolutionProcess (aux)
pos ✅ InPersonHandover; StaffDiscount; split tender (loyalty+card+cash)
services ✅ ServicePerformance; NoShowPolicy; GracePeriod; subscription Active
bnpl ✅ PaymentTiming::Installments; financing child (I-6)
escrow ✅ PaymentTiming::AfterGoodsReceived; Guarantor intermediary
fx ✅ two single-currency Commitments; Simultaneous; currency_conversion
saas ✅ DigitalGood NonExclusive License; DigitalDelivery; AccessGrant
streaming ✅ AccessModel::Stream; GracePeriod; subscription Active (see B-2)
api-metering ✅ AccessModel::APIAccess; Metered; overage Commitment; EntitlementConsumption (aux)
nft ✅ DigitalGood Exclusive NFT; RoyaltyDistribution condition
auction-family ✅ Tendered bids; supersession; AuctionProcess/English (aux)
real-estate ✅ FinancingContingency; InspectionContingency; TitleTransfer; RegistryRecording
healthcare ✅ PostFulfillment(InsuranceAdjudication); PrescriptionRequired; NoReturnPolicy; MedicalRecord
government-procurement ✅ ComplianceDocumentation; ScoredSelection + AwardProtest (aux)
wholesale ✅ RecurringDelivery; Net; VolumePricing; blanket-PO tree (I-6)
marketplace ✅ CommissionSplit (DoubleSided); payout+commission tree (I-6)
trade-finance ✅ DocumentaryCollection; DocumentsAgainstPayment; CustomsRelease; BillOfLading+CustomsClearance
events ✅ EventAccess; CascadeCancellation; EventCancellationPolicy; ticket tree (I-6)
loyalty ✅ LoyaltyEarnTerm; custom-currency points (PTS); split cash/points
group-buying ✅ ThresholdActivation; simultaneous activation
carbon-credits ✅ AccessModel::CarbonCredit; ValueState::Retired; RegistryVerification; RegistryRetirement; RetirementCertificate

“(aux)” = the signature record is an auxiliary record, validated by validate-aux.mjs against the canonical schema (the scene runner has no kind for standalone auxiliary records — BACKLOG B-3).

Findings — what the reconciliation revealed

  1. The model spine is sound and now executable against the canonical schema. All 22 domains express as canonical scene fixtures and audit clean (I-1…I-6). No domain needed a sixth primitive; 0 unmodelable.

  2. D’s “pending-v1.1” gaps were artifacts of D’s bespoke schema, not the canonical one. ScoredSelection, the full CommitmentCondition layer, CascadeCancellation, AwardProtest, and the v0.3 Evidence / DeliveryMethod / PaymentTiming / AccessModel variants are all present in canonical v1.0.0 — proven by the re-authored fixtures (which carry them in terms) and by validate-aux.mjs. 0 of D’s listed constructs is a real v1.1 gap.

  3. One genuine schema gap surfaced — B-2: canonical ValueState has no digital-access lifecycle (AccessGranted / AccessSuspended / AccessRevoked / AccessExpired). streaming/saas model access suspension/revocation at the Commitment level instead. This does not block those domains (they validate); it is a candidate v1.1 refinement.

  4. One conformance-coverage item — B-3: the runner has no fixture kind for standalone auxiliary records; they are covered by validate-aux.mjs as a supplement. Not a schema gap.

The claim, stated precisely: every domain is documented and lifecycle-walked; all 22 are executably validated against the canonical schema v1.0.0; 0 await a v1.1 schema construct to be expressible; the only enumerated follow-ups are one optional ValueState refinement (B-2) and one runner-coverage enhancement (B-3).

Regenerating 

node conformance/case-studies/_generate.mjs   # rewrites fixtures + manifest block
node conformance/runner/run.mjs               # must stay 51/51
node conformance/case-studies/validate-aux.mjs
Source: docs/case-studies/README.md in the warp-lang repository.