New rule request: union subtypes with conflicting field value type

[caccialdo]

Is your feature request related to a problem? Please describe.

When 2 subtypes of the same union type share a field but with a different value type, it's possible to run into query-time errors like below if the client doesn't alias the field with conflicting value types:

Minimal reproducible setup where a Book is either a Novel or a Poem but their id field is Drn and Int respectively.

Sample schema

type Drn {
  country: String
  id: String
}

type Poem {
  id: Int
  title: String
  author: String
}

type Novel {
  id: Drn
  title: String
  author: String
}

union Book = Poem | Novel

type Query {
  books: [Book]
}

Describe the solution you'd like

Creating such scenario in a schema isn't forbidden per se but it would be nice to have the option of enforcing a linting rule preventing it in the first place. The rule would suggest renaming the field(s) in the conflicting subtype(s).

Describe alternatives you've considered

apollographql/eslint-plugin-graphql but the tool is also missing a rule for this scenario.

Additional context

–

[dimaMachina]

Accepted, this seems could be fixed by integrating federation validation rules into graphql-eslint #1824

If somebody want to give a try to integrate it, feel free, i’ll review it

[dimaMachina]

@caccialdo After some searching, seems your lint error in Apollo Studio goes not from federation rules but from graphql-js rules, use @graphql-eslint/overlapping-fields-can-be-merged rule and you can try it in playground https://the-guild.dev/graphql/eslint/play?or=overlapping-fields-can-be-merged

[caccialdo]

Thanks @dimaMachina ! It does look like this is the linting rule used in the playground indeed (cf. original screenshot in the ticket description). I'll look into leveraging it in our clients.

I think the feature request is still valid though as enforcing it on the server-side at the schema level (vs. operation level) would be valuable for clients lacking linting capabilities.

[dimaMachina]

@caccialdo this rule is accepted, if you want you can work on it and send a PR

[caccialdo]

Thanks! I'll give it a shot in the new year if nobody attempts it before me