Storm is code-first today: you write the records and the schema has to match. Most teams that try Storm start from an existing database, and right now that means writing every record by hand.
Add generation from a live schema: connect, introspect tables, columns, primary and foreign keys, and emit Java records or Kotlin data classes annotated with @DbTable/@DbColumn/@PK/@FK, ready for the metamodel processor. JDBC metadata gets most of the way; dialect-specific type mapping fills in the rest.
The npm CLI only configures AI assistants and runs on Node, so the generator itself belongs on the JVM: a storm-codegen module with Maven plugin and Gradle task entry points, plus a runnable jar. The CLI can front it later.
Tasks:
Storm is code-first today: you write the records and the schema has to match. Most teams that try Storm start from an existing database, and right now that means writing every record by hand.
Add generation from a live schema: connect, introspect tables, columns, primary and foreign keys, and emit Java records or Kotlin data classes annotated with
@DbTable/@DbColumn/@PK/@FK, ready for the metamodel processor. JDBC metadata gets most of the way; dialect-specific type mapping fills in the rest.The npm CLI only configures AI assistants and runs on Node, so the generator itself belongs on the JVM: a storm-codegen module with Maven plugin and Gradle task entry points, plus a runnable jar. The CLI can front it later.
Tasks:
Ref<T>