Hi All! New to Prisma and GraphQL so I have what is likely a stupid question: Why do I need to define my data schema essentially twice - once in my

prisma.datamodel

file and again in my

schema.graphql

file? As per this example repo: https://github.com/prisma/prisma-examples/tree/master/node/graphql. Is it because the former is not really graphQL syntax? Also, why does the example repo define a separate graphQL server using graphql-yoga when the prisma playground “speaks” GraphQL and I can make graphQL queries against my prisma server/database.

Hey, the datamodel is essentially going to configure your database. The second is your public facing API schema that you would expose to the client.

Gotcha. So the playground that the prisma server stands up is essentially a direct conversion of the prisma datamodel into a graphQL schema? Which is fine for development but may not be desirable for public use with additional complexity (e.g. permissions/authn/z) ?

Correct. You’d almost never want to expose the underlying Prisma API (while it’s authenticated - If you enable a secret key) it can be confusing to consume so the watered down API infront is often a great barrier

Makes sense. Thanks for clarifying!

You might also want to check out this article https://www.prisma.io/blog/prisma-and-graphql-mfl5y2r7t49c/ @Craig 🙌

Thanks @notrab. I had this exact confusion about the seeming duplication of gql serves and your distinction clears it up!