Sprint 16 Release Notes

Table of Contents

April 2026

Sprint 16 delivered a generalised event-driven workflow engine, full ORE XML round-trip coverage across 17 instrument families, a new market-data domain and storage service, per-asset-class typed instrument models, and a Qt plugin architecture for the client.

✅ Highlights

  • Generalised workflow engine — ores.workflow.service, saga-based executor with FSM state UUIDs, startup recovery, correlation IDs, and a dedicated Qt Workflow Monitor plugin.
  • ORE integration (Phases 1–17) — end-to-end ORE XML import/export: golden round-trip tests for every instrument family (rates, FX, bond, credit, equity, commodity, composite, scripted, exotic/hybrid), plus server-side ores.ore.service orchestrating imports over NATS + ores.storage.
  • ORE market-data domain model — new ores.marketdata.{api,core,service} with TimescaleDB schema for series/observations/fixings, ORE key decomposition, server-side import of market.txt / fixings.txt, and a Market Data MDI UI.
  • Typed instrument domain model — replaced the generic ores_trading_instruments_tbl with per-asset-class tables for rates, FX, and equity (bond/credit to follow); instrument_family renamed to product_type.
  • Qt plugin architecture — ores.qt.api extracted as a shared framework; domain-specific plugins (ores.qt.admin, ores.qt.compute, ores.qt.mktdata, ores.qt.refdata, ores.qt.trading, ores.qt.scheduler) behind an IPlugin interface.

🛠️ Key Improvements

Workflow Orchestration

  • Saga engine orchestrates cross-service workflows (party provisioning, ORE import, report execution) with durable state, compensation steps, and startup recovery.
  • Status migrated from ad-hoc strings/enums to database-driven FSM state UUIDs (ores_workflow_fsm_*); fsm_state_map utility resolves state names dynamically.
  • Nats-Correlation-Id propagated across every handler via log_handler_entry; correlation IDs surfaced on the Tenant Provisioning Wizard summary page.
  • Workflow Monitor Qt plugin + NATS query handlers; idempotent self-service JetStream stream provisioning replaces central bootstrap.

ORE Integration

  • Forward + reverse instrument mappers for credit, rates, FX exotics (barriers, digitals, accumulators, variance swaps), equity (options/forwards/swaps), bond (incl. BondRepo, BondTRS), commodity (swaps, APOs, option strips), composite, scripted, exotic/hybrid.
  • Unified map_instrument dispatch via std::variant; round-trip fidelity tests for every family.
  • Server-side ores.ore.service handles XML import over NATS; ores.storage provides S3-like object storage with PUT/GET/DELETE HTTP routes (replaces compute-specific buckets).
  • Portfolio export to ORE XML, "Open Instrument" / "Export XML" actions in BookMdiWindow; auto-import of market.txt + fixings.txt alongside trade XML.
  • ORE XML calendar adjustments and financial conventions (Zero, Deposit, Swap, OIS, FRA, IBOR, Overnight, FX, CDS) imported into refdata.

Market Data

  • ores_marketdata_series_tbl, ores_marketdata_observations_tbl, ores_marketdata_fixings_tbl with bitemporal history and tenant isolation.
  • ORE key decomposition via series_key_registry for lossless parse/serialize.
  • Qt Market Data MDI: list views for series + index fixings, time-series detail views, MarketDataController.
  • DataLibrarianWindow and publish dialogs moved from mktdata to refdata plugin.

Instrument & Trade Model

  • Typed per-asset-class instrument tables and domain models for rates (Phase 0/1), FX (incl. barriers, digitals, accumulators, variance swaps), and equity (options/forwards/swaps/positions).
  • TradeDetailDialog unification: FX, swap/rates, bond+credit, equity+commodity, composite, scripted instruments merged into a single dialog (6-phase refactor).
  • IInstrumentForm registry + per-family form widgets; dynamic form activation via trade type cache; create-mode instrument capture.
  • Data-driven trade-type metadata: typed product_type_t enum, trading.v1.trade-types.list NATS protocol, DB ores_trading_trade_types_tbl.

Asset Class Unification

  • instrument_family → product_type rename across SQL, C++, NATS, Qt (clarifies routing discriminator vs. risk taxonomy).
  • Data-driven asset classes: new refdata service, repository, NATS protocol (refdata.v1.asset-classes.list); Qt combos loaded asynchronously; asset_class column added to trades.

Analytics

  • New ores.analytics.{api,core,service} component managing pricing engine types, model configurations, products, and parameters.
  • Analytics CRUD UI with codegen support for combo-box widgets in detail dialogs; repository integration tests for all four entities.
  • DQ project layout restructured; report definitions sourced from DQ artefact table (28 standard ORE analytics reports seeded) instead of hardcoded C++.

Service & Infrastructure

  • All services normalised onto NATS-based hosting via ores.service runners (domain, signing, Wt); unified heartbeats now the single source of truth for the service dashboard.
  • New service controller: DB entities, NATS handlers, lifecycle events, replica-specific logging; service dashboard shows exit codes + last 20 log lines on failure.
  • Multi-phase process shutdown (SIGTERM → SIGQUIT → SIGKILL) with Windows-safe shutdown_signals abstraction; TLS cert validation on NATS connections.
  • UTC-everywhere policy: canonical UTC API in ores.platform, DB sessions pinned to UTC, ISO 8601 on the wire, display-only local time in UI.
  • Session-level Nats-Session-Id header for per-session log correlation.

Report Execution

  • report_execution_handler for trade gathering + finalization; report execution workflow pipeline (Phase 3) with concurrency checks; risk_report_config domain model and repository.
  • Party picker redesigned: business-center codes, party categories, system vs. operational filtering, flag icons via image cache.

Scheduler

  • Qt UI completed: Job Instances execution history + live Scheduler Monitor, backed by new NATS protocols and handlers.

Build, Portability & DX

  • Windows portability: replaced strptime=/=_wfopen=/=SIGQUIT with portable equivalents; MSVC /Z7 embedded debug info to fix PDB contention under parallel Ninja; trading.core built as static lib to dodge 64 K DLL export limit; registrar.cpp split to avoid C1202 constexpr depth.
  • macOS portability: portable file-time conversion replaces clock_cast.
  • CI: MSYS2 path-mangling fixes in NATS cert generation; WiX 2 GB limit handled; stack size bumped to 8 MB for XML-parsing tests.
  • Split-component codegen profiles (api=/=core=/=service) with automated file placement.
  • Valgrind suppressions broadened for GLib call_init UB.

⚠️ Known Issues & Postponed

  • Asset class unification — Phases 1 (product_type rename) and 2–5 (refdata seeding, validation, Qt data-driven combo) landed; remaining integration work (full Qt wiring of asset_class on trades) continues in Sprint 17.
  • Typed instruments — rates, FX, and equity tables landed; bond and credit typed tables deferred to Sprint 17.
  • Party isolation RLS — additional RLS policies across IAM/refdata/scheduler/trading landed (#681), but full enforcement via parent-table joins on trading instrument subtables remains open.
  • OpenTelemetry distributed tracing, three-level provisioning E2E tests, async workflow progress for large party hierarchies, trade-expiry and barrier-event workflows, positions domain model, party wizard UX improvements, and automated service registration remain on the Sprint 16 backlog and are carried into Sprint 17.
  • Instrument creation in TradeDetailDialog — several PRs (#649, #651) flagged regressions in create-mode flows (hidden tabs, broken save wiring); follow-up fixes tracked for Sprint 17.

📊 Time Summary

  • Total effort: not tracked
  • PRs merged: not tracked
  • Sprint duration: 2026-03-30 → 2026-04-18

Next sprint: Sprint 17.

Emacs 29.1 (Org mode 9.6.6)