When creating types in GraphQL, the list of fields on a type can become quite long after a while. One way of keeping the number of fields down would be to put some relevant subset of the fields in to one "grouping" type, so instead of user { last_name, first_name, height, weight, friends, favorite_clubs } one would have user { identification, friends, favorite_clubs } where the schema for identification is of course { last_name, first_name, height, weight }.
However, I'm not convinced that doing this is a good idea. The advantage is that the user type is more readable, but the things that worry me are:
You are creating a new GraphQLObjectType just for the sake of organisation, it's not a new node in my mental model of the graph...
The root parameter of the identification resolver would still be the user object, which feels like I'm doing something wrong...
Is this a good way of organising your API, or are my worries correct?
As it all comes down to what you prefer, there is no real answer to this question.
I would personally simply list all the individual fields. :)
Doesn't mean the other way is wrong.
Anyways, don't overthink it.