A server in ten lines

A UDC server is a directory of collections and a few lines that serve it. This one keeps three books in memory; swap kind: memory for Postgres, a directory of files or an HTTP API later without touching the server.

Install the server, the protocol package and Fastify:
Terminal window
```
mkdir books-server && cd books-server && npm init -y && npm pkg set type=module
npm install @jsudc/server @jsudc/core fastify zod
```
npm also installs drizzle-orm and pg, the server’s other peers; a memory collection never opens a connection.

Lay out one collection, books:

server.ts
Directorycollections/
- Directorybooks/
  - schema.json the fields
  - view.yaml how the data view shows it
  - source.yaml where the rows live
  - seed.json the rows a memory collection starts with

{
  "version": "v1",
  "core_fields": [],
  "fields": {
    "title": { "type": "string", "required": true },
    "author": { "type": "string" },
    "year": { "type": "integer", "min": 0 },
    "pages": { "type": "integer", "min": 1 },
    "read": { "type": "bool" }
  },
  "field_order": ["title", "author", "year", "pages", "read"]
}

default_renderer: table
available_renderers: [table, cards]
search:
  fields: [title, author]
columns:
  - { field: title, width: 260 }
  - { field: author, width: 200 }
  - { field: year, width: 90, align: right }
  - { field: pages, width: 90, align: right }
  - { field: read, width: 80, align: center }
card_options:
  title_field: title
  subtitle_field: author

kind: memory

[
  {
    "title": "The Master and Margarita",
    "author": "Mikhail Bulgakov",
    "year": 1967,
    "pages": 384,
    "read": true
  },
  {
    "title": "The Bridge on the Drina",
    "author": "Ivo Andrić",
    "year": 1945,
    "pages": 314,
    "read": true
  },
  {
    "title": "Dictionary of the Khazars",
    "author": "Milorad Pavić",
    "year": 1984,
    "pages": 338,
    "read": false
  }
]

The server:

import { createDispatcher, loadRegistry, makeSource, udcPlugin } from "@jsudc/server";
import Fastify from "fastify";

const registry = loadRegistry("./collections", (asset, registry) =>
  makeSource(asset, { registry, env: process.env }),
);
const app = Fastify({ logger: true });
const dispatcher = createDispatcher(registry, { log: app.log });
await app.register(udcPlugin, { dispatcher, registry, swagger: false });
await app.listen({ port: 5100 });

loadRegistry reads every directory under collections/, validates its three files and fails at start on a mistake; makeSource turns each source.yaml into its source. udcPlugin mounts POST /udc/<action> on your own Fastify instance.

Run it and ask for the books:

node server.ts
curl -s localhost:5100/udc/list -H 'content-type: application/json' \
  -d '{"collection":"books","sort":[{"field":"year","dir":"asc"}]}'

{
  "kind": "udc.list.ok",
  "id": "",
  "payload": {
    "rows": [
      { "__meta": { "id": "…", "version": 1 }, "title": "The Bridge on the Drina", "year": 1945, … },
      …
    ],
    "next_cursor": null,
    "total": 3
  }
}

Every answer is HTTP 200; the envelope’s kind says ok or err. Try describe with {"collection":"books"} to see what a client learns before it draws anything, or mutate to add a fourth book: a memory collection keeps it until the process ends.

The OpenAPI document of every action at /docs: drop swagger: false and install @fastify/swagger @fastify/swagger-ui fastify-type-provider-zod.
Draw the books: a React data view pointed at this server.
Keep them in Postgres: the full stack.

A server in ten lines

Next