Skip to content

Configuration Extensibility - Implementation Phases

This document tracks the progress of the Configuration Extensibility project (feature/configuration-extensibility).

🟢 Phase 1: Infrastructure & Domain Configuration (Completed)

Goal: Establish the foundation for loading and serving domain-specific configurations.

  • Domain Model: Created DomainConfig, WorkflowConfig, AiStrategiesConfig models.
  • Configuration Service: Implemented DomainConfigurationService to load masala_domains.yaml.
  • Data Persistence: Added DomainId to Ticket entity.
  • Global Defaults: Defined default "IT" domain behavior.

🟢 Phase 2: Custom Fields & Dynamic UI (Completed)

Goal: Enable domains to define custom data fields that are rendered dynamically.

  • Validation Service: Created ICustomFieldValidationService for type safety (number, date, select).
  • Dynamic Rendering: Implemented _CustomFieldsPartial.cshtml for Create/Edit views.
  • Persistence: Implemented JSON storage (CustomFieldsJson) for custom data.
  • Read-Only View: Added _CustomFieldsDisplayPartial.cshtml for Ticket Details.
  • Search/Indexing: (Addressed via Architecture critique: JSONB/Computed Columns recommendation).

🟢 Phase 3: Workflow Engine (Completed)

Goal: Enforce domain-specific state transitions and business rules.

  • Rule Engine: Implemented IRuleEngineService for transition validation.
  • Logic Enforcement: Integrated checks in TicketService.UpdateTicketAsync.
  • UI Filtering: Updated Edit view to filter "Status" dropdown based on valid next states.
  • Exception Handling: Added graceful error messages for invalid transitions.

� Phase 4: AI Strategy Extension (Completed)

Goal: Allow domains to configure which GERDA AI strategies are used for ranking, estimating, and dispatching.

  • Strategy Factory: create a factory to resolve IJobRankingStrategy, IEstimatingStrategy, etc., by name.
  • Configuration Integration: Update services to ask DomainConfig which strategy key to use (e.g., "TaxLaw" -> "RiskScoreRanking").
  • Safety Checks: Validate that configured strategies exist in the DI container at startup.
  • Testing: Verify different behaviors for different domains (e.g., IT uses WSJF, Gardening uses SeasonalPriority).

� Phase 4.5: Configuration Versioning (Critical Prerequisite)

Goal: Implement "Snapshot Strategy" to prevent rule changes from breaking existing tickets.

  • Data Model: Create DomainConfigVersion entity.
  • Versioning: Store ConfigVersionId on Ticket creation.
  • Rule Engine: Update service to request rules by Version ID.

�🟢 Phase 5: Performance Optimization (Rule Compiler) (Completed)

Goal: Refactor Rule Engine from Runtime Interpreter to Compiled Expression Trees to prevent performance degradation at scale.

Guidance:

  • Cache Key: Dictionary<(string DomainId, string VersionHash), CompiledPolicy>

  • Safety Valve: Wrap Expression.Compile() in try/catch; fallback to safe delegate on error.

  • Rule Compiler Service: Implement RuleCompilerService using System.Linq.Expressions.

  • Startup Compilation: Compile and cache all YAML rules on app startup/reload.
  • Field Extractor: Optimize JSON field access with FieldExtractor helper.

🟢 Phase 6: Advanced AI (Feature Extraction) (Completed)

Goal: Move beyond basic Strategy selection to dynamic Feature Extraction for ML models (ONNX/ML.NET).

  • Feature Mapping: Add feature_mapping format to masala_domains.yaml.
  • Feature Extractor: Implement IFeatureExtractor to convert Ticket/JSON to float[].
  • Integration: Update Strategies to use extracted features for inference.

� Phase 7: UI Localization & Branding (Completed)

Goal: Domain-aware UI labels and theming.

  • UI Localization Service: Replace hardcoded "Ticket" labels with config lookups.
  • Domain Switcher: Add domain switcher to layout for multi-domain deployments.
  • Domain Styles: Add domain-specific icons and color themes support.

� Phase 8: Scalable Ingestion (Gatekeeper) (Completed)

Goal: Decouple ingestion to handle high throughput (IoT/Webhooks) and long-running syncs (ERP). Adopt Event Driven Architecture for intake.

Decision: Use Scriban for template rendering (Mapper).

  • Gatekeeper API: Create separate Minimal API project for accepting webhooks.
  • Message Bus: Implement simple producer/consumer (System.Threading.Channels).
  • Digestion Worker: Create BackgroundService to process queue items.
  • Ingestion Mapping: Detailed ingestion configuration in masala_domains.yaml using Scriban templates.

Status Legend:

  • 🟢 Completed
  • 🟡 In Progress / Next
  • 🔴 Not Started