Support for GraphQL persisted queries

[Mosnar]

Description

The concept of persisted queries in GraphQL is essentially that you keep track of your queries in an allow-list with queries mapped to IDs. When executing the query, instead of passing the entire thing, you just pass a queryId. The GraphQL server is then responsible for looking up that query and ensuring it's allowed. Generally, the process of generating the allow list is automated using a frontend build tool to extract queries and generate the allow-list for you, replacing queries with their IDs in your compiled code.

The feature already exists on the webonyx/graphql-php server via the persistentQueryLoader config; however, it would need to be implemented in Craft's implementation.

Benefits

• Makes it easier to utilize GraphQL as an API in production since you don't need to worry about permissions leaking data
• Smaller client requests since the payload only needs to carry the query ID
• Smaller production assets since your queries would be replaced with IDs
• Improves cachability of http requests since not all full-length queries can fit into GET request parameters to be cached on CDNs

Related Reading

Relay Persisted Queries Reference: https://relay.dev/docs/en/persisted-queries.html
Apollo Server Reference: https://www.apollographql.com/docs/apollo-server/performance/apq/
Example Webpack Loader: https://github.com/leoasis/graphql-persisted-document-loader

[denisyilmaz]

Is there any plan to implement this feature?

[denisyilmaz]

Hi @andris-sevcenko any updates on this feature?

[mark-wff]

@brandonkelly is there an update when the feature will be implemented?

[denisyilmaz]

@brandonkelly any updates regarding this feature?

[brandonkelly]

We’re planning to do a big GraphQL push soon, once the dust has cleared from the Craft 5 launch!

[denisyilmaz]

That sounds great. Using Craft CMS in Headless mode is our default setup for clients, so i'm happy to see progress in this area
.

[denisyilmaz]

Hi @andris-sevcenko besides the idea of persisted queries, would something like "RESTified GraphQL Endpoints" be easier to implement? Hasura GraphQL Engine is using this idea: https://hasura.io/docs/latest/api-reference/restified/
You basically save GraphQL queries in the database and create a RestAPI endpoint for it.

[brandonkelly]

Can you post that as a new discussion?

[denisyilmaz]

created the discussion here: #12503

[denisyilmaz]

@brandonkelly are there any plans to integrate this? maybe for craft 6?

For bigger projects we see GraphQL being the biggest issue in regards to performance when the IA is using a lot of relationship fields and the content gets updated regularly. any update for a stronger/better caching/performance for graphql would highly appreciated

[denisyilmaz]

@brandonkelly @andris-sevcenko Any update on this?

[brandonkelly]

Sorry, no update yet, but it’s on our list.