Page 3: Running your first query

With the server from page 1 still running on http://localhost:8080, open a second terminal and send a query. (If you prefer the in-browser GraphiQL playground, open http://localhost:8080/graphiql/ and paste the query body into the left pane instead of running curl.)

Send the query

curl -s -X POST http://localhost:8080/graphql \
  -H 'Content-Type: application/json' \
  -d '{"query":"{ customers { firstName lastName email } }"}'

The response is JSON containing the five rows the example seed loaded into the customer table:

{
  "data": {
    "customers": [
      {"firstName": "Mary", "lastName": "Smith", "email": "mary.smith@example.com"},
      {"firstName": "Patricia", "lastName": "Johnson", "email": "patricia.johnson@example.com"},
      {"firstName": "Linda", "lastName": "Williams", "email": "linda.williams@example.com"},
      {"firstName": "Barbara", "lastName": "Jones", "email": "barbara.jones@example.com"},
      {"firstName": "Elizabeth", "lastName": "Brown", "email": "elizabeth.brown@example.com"}
    ]
  }
}

What just happened

Graphitron’s generated resolver for Query.customers produced a single SQL statement against PostgreSQL, projected only the columns your query asked for, and returned the rows as a GraphQL response.

Roughly, it ran:

SELECT customer.first_name AS "firstName",
       customer.last_name  AS "lastName",
       customer.email      AS "email"
  FROM customer

If you are running with quarkus.log.level=DEBUG (or want to inspect the executed SQL), Quarkus’s dev mode prints jOOQ’s SQL log to the terminal where mvn quarkus:dev is running. The actual statement uses parameter placeholders and quoted identifiers, but the shape is the one above.

The selection set drives the projection: a query that asks only for firstName produces a SELECT with only customer.first_name. Add lastName to the query and the SELECT grows to two columns. Graphitron does this projection-narrowing for every query, every time, because each generated resolver inspects the request’s selection set and threads the chosen fields through to jOOQ.

Filter with the active argument

The active argument from page 2 filters by the activebool column. The seed data has three active customers and two inactive:

curl -s -X POST http://localhost:8080/graphql \
  -H 'Content-Type: application/json' \
  -d '{"query":"{ customers(active: true) { firstName } }"}'

{"data":{"customers":[
  {"firstName":"Mary"},
  {"firstName":"Patricia"},
  {"firstName":"Linda"}
]}}

The @field(name: "ACTIVEBOOL") on the active argument is what wired this filter to the activebool column. Without it, the generator would have used the GraphQL name (active) to look up a column called active, found one (the customer table happens to have an unused integer column named active too), and wired the filter to the wrong place. Argument-level @field is how you steer the column choice when names collide.

You have just learned

How a generated resolver translates a GraphQL selection set into a column-narrowed SELECT, and how an argument-level @field directive maps a query argument to a specific column.

Page 4: Joining tables adds the address field and shows how @reference produces a JOIN.

← Previous: The starter schema

↑ Tutorial overview

Next: Joining tables →