Table of Contents


Namespaces allow you to have multiple schemas under the same application, making sure they are isolated while also sharing common components.

Let’s imagine a scenario where your application has an admin and a client side, and you want to run both in GraphQL. But, the admin will have its set of queries, mutations, and subscriptions, while the client side will have a different set. On top of that, some common components should be available on both sides, so you don’t have to repeat the code.

Such a scenario is possible with the usage of namespaces.

How it works

You just have to remember three things:

  1. Schemas only refer to a single namespace;
  2. Any other component can have multiple namespaces;
  3. The :base namespace is the fallback.

With that in mind, the solution is to create two schemas, each one with its own namespace. Then add isolated components to their respective namespace and the shared ones to either the :base or to both namespaces.

Setting the Namespace

There are two ways to define a component’s namespace:

# app/graphql/objects/user.rb
set_namespace :admin     # This forces the value
namespace :admin         # This adds to the list

You can call .namespaces from any component to see its value.

Note Namespaces are inherited by default, but you can use the above methods to replace or increment the value.

Module Association

To reduce the number of places you might have to set the namespace manually, the Type Map allows you to associate a module to a namespace. When doing this, any component inside of that method or its descendants will be automatically set to that namespace unless told otherwise.

Here is how you can configure that:

GraphQL.type_map.associate(:admin, GraphQL::Admin)
# OR
# app/graphql/admin_schema.rb
namespace :admin, GraphQL::Admin

The schema has its own version of the namespace method, which allows you to set the namespace and, at the same time, associate a module to it.