-
Add custom conditions: attach a developer-supplied predicate to a field, argument, or input field via @condition.
-
Apollo Federation transport: wire the federation runtime around a Graphitron schema.
-
Batch lookups: drive a positional batch fetch with @lookupKey and get one result per input position.
-
Computed fields with @externalField: project a derived column from a developer-supplied jOOQ Field<T>.
-
Stacking and overriding conditions: when explicit @condition predicates stack with implicit ones, and what override: suppresses.
-
Cursor-paginated connections: synthesised vs declared connection types, totalCount cost, and what the rewrite caps (and doesn’t).
-
Custom scalars: declare a custom scalar in the SDL and bind it to a GraphQLScalarType constant via the @scalarType(scalar:) directive.
-
The dev loop (Quarkus + LSP): the inner edit-save-reload loop, why macOS needs the LSP, and how to connect IntelliJ to it.
-
The errors channel: route Java exceptions through a service payload’s errors: field as typed data.
-
Wire external Java code: the shared mechanics for @service, @externalField, @tableMethod, @condition, @sourceRow, and @enum.
-
Federation key fields: declare entity keys with @key, layer them onto @node types, and stub non-resolvable entities owned by another subgraph.
-
Global object IDs: Relay Global Object Identification end-to-end, including stable IDs, keyColumns migrations, and Query.node dispatch.
-
Handle services: bind GraphQL arguments to Java parameters and pick a return shape the framework recognises.
-
Join with references: pin foreign-key paths between table-bound types so the generator emits a JOIN.
-
Map types to tables: bind types and fields with @table + @field across the four directive sites.
-
Agent context over MCP: connect an MCP-aware agent (Claude Code, Cursor) to graphitron:dev so it can answer questions about your schema, catalog, and directives.
-
Migrating from the legacy generator: walk a schema authored against graphitron-codegen-parent to the rewrite.
-
Multi-hop @nodeId filters: reach a related node type through one or more intermediate tables when the FKs carry the target’s keys positionally by SQL name.
-
Polymorphic types: choose between single-table and multi-table layouts for GraphQL interfaces and unions.
-
Result-type variants: understand the backing class reflected from a class-backed parent’s producer (jOOQ TableRecord, jOOQ Record, Java record, plain POJO).
-
Sort results: pick a column source, build stable keyset cursors with explicit tie-breakers, and avoid the collation pitfalls.
-
When to split queries: choose between inline correlated subqueries and @splitQuery-batched fetches.
-
Tenant scoping: per-request routing through Graphitron.newExecutionInput(dsl, …). The full recipe (per-tenant DataLoader partitioning, @tenantId directive, byTenant factory overload) ships under R45; until then, single-tenant apps pass a single DSLContext.
-
Test your schema: wire a consumer-side execution-tier test against a real Postgres.