Sprint 12 Release Notes

Table of Contents

February 2026

This sprint focused on a massive expansion of the Legal Entity and Party framework, the introduction of Trading & Book hierarchies, and a major infrastructure shift toward a unified database model with Row-Level Security (RLS). It represents a foundational shift in the system's architecture, moving from a basic reference data store to a sophisticated, secure, and hierarchy-aware financial platform. We have significantly matured the organizational model, enhanced security protocols, and established the core infrastructure for trade lifecycle management.

✅ Highlights

  • Party & Counterparty Ecosystem: Implemented a comprehensive C++ and SQL stack for managing global legal entities, supporting GLEIF data integration, hierarchical corporate structures, and multi-identifier mapping (LEI, BIC).
  • Trading & Risk Foundation: Launched the ores.trading component, establishing the "Trade Envelope" and a dual-layer hierarchy of Portfolios (for risk aggregation) and Books (for trade capture).
  • Unified Security Architecture: Migrated to a single-database model, replacing physical database cloning with Row-Level Security (RLS) to isolate tenant and party-level data.
  • Automated Evaluation Onboarding: Introduced a "First Login" wizard for Evaluation Tenants, featuring automated GLEIF hierarchy imports and synthetic data generation for immediate platform testing.
  • Finite State Machine (FSM): Developed a native PostgreSQL FSM framework to govern complex entity lifecycles, currently powering trade status transitions and audit-trailed authorizations.

🛠️ Key Improvements

Identity & Access Management (IAM)

  • System & Operational Roles: Introduced "System" vs. "Operational" party distinctions and separated "Tenant Admin" from "Super Admin" roles for granular platform control.
  • Human-Readable Short Codes: Implemented a mnemonic generation engine to replace UUIDs in daily operations (e.g., "GLDMN" for Goldman Sachs) with built-in collision resolution.
  • Audit Integrity: Hardened the audit trail by ensuring modified_by fields are derived from the authenticated database session, preventing identity spoofing in communication protocols.

Architecture & Infrastructure

  • Single-Database Migration: Eliminated ores_admin and ores_template databases in favor of a unified model using a two-phase setup (Superuser vs. DDL user).
  • SQL Security: Extended the sqlgen library to support bound parameters, significantly reducing the surface area for SQL injection by moving away from raw libpq queries.
  • Portable Build System: Fixed macOS CI failures by replacing GNU-specific getopt logic with portable shell parsing for database scripts.

UI & Developer Experience

  • Qt GUI Enhancements: Upgraded the entity list views with a Strategy Pattern to reduce code duplication and added a visual Party Hierarchy Graph for navigating parent-child relationships.
  • Advanced List Widgets: Centralized UI metadata (column widths, alignment, persistence) into a single source of truth, ensuring a consistent experience across all 24 MDI windows.
  • Interactive Shell: Integrated a REPL shell widget into ores.qt, allowing administrators to perform bulk tasks (like tenant provisioning) directly from the GUI.

Testing & Data Quality

  • FK-Aware Generators: Overhauled synthetic data generation to respect Foreign Key constraints, allowing for the creation of complex, interconnected test datasets (e.g., linking a party_id to its identifiers).
  • Coverage Hardening: Added 110+ test cases and 12 new unit test files, successfully re-enabling 19 previously skipped tests in the DQ and IAM modules.
  • GLEIF Integration: Established an automated pipeline to ingest global LEI data, including specialized detection logic for 93+ central banks and supranational entities.

⚠️ Known Issues & Postponed

  • Caching Architecture: Analysis of the ores.caching library (using immer for immutable data structures) has been POSTPONED to a future sprint to prioritize core trading functionality.
  • libpq Transition: Full removal of raw libpq calls is BLOCKED pending the official release and package update of the enhanced sqlgen library.
  • Multi-Schema Model: The story to "Split create schemas from main populate script" was CANCELLED as the project has moved to a unified RLS-based database design.

📊 Time Summary

  • Total effort: not tracked
  • PRs merged: not tracked
  • Sprint duration: 2026-02-06 → 2026-02-20

Next sprint: Sprint 13.

Emacs 29.1 (Org mode 9.6.6)