Dev Update: March 28 — The 1.0 Push

What shipped

Wave 0 cleared out all dead database references — dropped tables and columns that were causing runtime crashes. ScrapLogPage deleted, NCR disposition renamed, vendor status column fixed. Both builds passing, all tests green.

Wave 1 laid the foundation: number sequence atomicity (no more duplicate invoice numbers under load), idempotency middleware (double-click protection on every POST), fiscal period close endpoint, and 9 shared frontend components that every entity page will use going forward.

New API endpoints

Resource rate management, part cost tracking, and fiscal period close — three new business object route files with full CRUD. Schema columns wired into invoice, quote, sales order, and shipment routes. The API now exposes every column the database has.

MRP + Scheduling engine designed

The biggest design decision this week: MRP and scheduling are one engine, not two separate systems. Demand explosion, netting, lot sizing, backward scheduling, and finite capacity loading all happen in a single planning run.

Key decisions: minute-level granularity, three job states (unfirmed / firm planned / released), time-based operation overlap instead of quantity-based, and cross-resource scheduling that splits work across shifts automatically. Full design document published internally.

What's next

Wave 2 is in progress: PDF generation for quotes, travelers, invoices, and POs. Approval workflows for purchase orders. Backflush pipeline for auto-issuing materials when operations complete. Labor detail endpoints. The planning engine implementation starts in Wave 3.

We're early in the 1.0 execution sequence, shipping the execution sequence slice by slice.