Author ltree columns

Prisma 8 has two contract-authoring lanes, and prisma-ltree supports both:

• PSL lane: contract.prisma, the surface the official extensions document
• TypeScript lane: contract.ts via defineContract

Both emit the same compiled contract. Choose the surface that fits your project.

Composing the extension

Whichever lane you author in, add ltree to extensions in prisma.config.ts so the ltree namespace resolves during emit:

import { definePrismaConfig } from "prisma/config";
import { defineConfig as ormConfig } from "@prisma/orm-postgres/config";
import ltree from "prisma-ltree/control";

export default definePrismaConfig({
  orm: ormConfig({
    contract: "./contract.prisma", // or "./contract.ts"
    extensions: [ltree],
  }),
});

Omitting ltree from extensions while a contract.prisma references ltree.Ltree() fails emit with PSL_EXTENSION_NAMESPACE_NOT_COMPOSED.

Declaring ltree columns

Two named-type constructors cover the Postgres types: ltree.Ltree() for a single ltree path (codec pg/ltree@1) and ltree.LtreeArray() for an ltree[] array (codec pg/ltree-array@1). Author them in either lane:

// use prisma

types {
  // Single ltree path → codec `pg/ltree@1`, native type `ltree`.
  Path = ltree.Ltree()
  // ltree[] array → codec `pg/ltree-array@1`, native type `ltree[]`.
  Paths = ltree.LtreeArray()
}

/// A page in a hierarchy. `path` is its position in the tree;
/// `breadcrumbs` is the array of ancestor paths.
model Page {
  id          String @id @default(uuid())
  path        Path
  breadcrumbs Paths

  @@map("page")
}

After editing either contract, re-emit:

pnpm prisma contract emit

Runtime, migrations, and indexes

See Get started for runtime wiring, the baseline migration, and your first queries.

Add a GiST index on path columns so ancestor, descendant, and pattern queries can use an index. See Add a GiST index.