Is this still a valid way of handling errors?

General
1

I’ve picked up a project that returns different entities depending if its an error or not.

I really like their errors within the data object (plus i know a few calls are being made with this set up) and lets developers loop through nested results and handle errors as and when they occur (specifically 404 errors)

My question is: Is this still a valid way of operating and if so is there documentation and use with data sources?

In the example below:
If there is a 404 or 401 error when querying the root the developer can quickly see/handle the error

If there is a user id stored in an array of organisation->userIds that does not have a corresponding user->id its is easy to handle / see where the error is

{
  Organisation(id: "123456-12345678-123456") {
    ...on Error404 {
      message
    }

  ...on Error401 {
      message
    }
    
    ...on Organisation {
      name
      id
      users {
        ...on Error404 {
# E.g. User with id 'qwerty', not found
          message
          id
        }

       ...on Error404 {
         message
       }
        
        ...on User {
          email
          roles
        }
      }
    }
  }
}

Here is the article they used to base their existing error handling:
https://blog.logrocket.com/handling-graphql-errors-like-a-champ-with-unions-and-interfaces/