More Batching

Batching can also be done through a many-to-many relationship. After the first SQL query, Join Monster will select from the junction table, and then join it on the related table. The junction object will need a couple more properties, including a uniqueKey for itself, and sqlBatch. This can be string for a single unique column or an array for a composite.

const User = new GraphQLObjectType({
  // ...
  fields: () => ({
    following: {
      description: 'Users that this user is following',
      type: new GraphQLList(User),
      // batching many-to-many is supported too
      junction: {
        sqlTable: 'relationships',
        // this table has no primary key, but the combination of these two columns is unique
        uniqueKey: [ 'follower_id', 'followee_id' ],
        sqlBatch: {
          // the matching column in the junction table
          thisKey: 'follower_id',
          // the column to match in the user table
          parentKey: 'id',
          // how to join the related table to the junction table
          sqlJoin: (junctionTable, followeeTable) => `${junctionTable}.followee_id = ${followeeTable}.id`

This introduces another batch to the plan. In addition to the changes made on the previous page, the plan now has 3 database queries.


Requests for the followees and for the comments are independent, and are sent concurrently.