Why the generator is shaped the way it is. The development axioms these pages lean on live in the Principles section.
-
Pipeline overview: how a
.graphqlsschema flows from parse through capture, the fact store’s derived strata, planning and the render fold to consumer compile. -
The fact model: the modeling discipline behind the fact store; natural keys, provenance as separate relations, derived reads as views, the closed command graph, every rule with its enforcer named.
-
Naming the row: a gentle introduction to the store’s naming habit, with examples, a pantry metaphor, and the honest costs; start here if the fact model page is your first contact with the store.
-
Dispatch axes — the decomposed dispatch model behind DataLoader-backed source-side fields: the batch-key residue, the record-parent lift, and the loader identity.
-
Typed rejection — why rejection is a typed variant, the
Rejectiontaxonomy, and the candidate-hint contract.