Why is the concept of N+1 important for Graph QL and Hasura independently?

[jonas-kgomo]

For context here is a study of the N+1 problem and data-loader query problem

[wtrocki]

Dataloader can add complexity. What I see is that current community moving towards understanding GraphQL better and trying to make it work same way as REST does. To do that developers need to:

• Ignore graphql execution plan and use only root resolver
• Parse and utilize resolver info object
• Use generated joins - because no one wants to write joins themselves.

I truly think that Hasura approach is way more efficient and sophisticated than dataloader. Since hasura does things only for postgress there is need for more generic solution to make it popular and show obvious benefits

Questions:

Do you think that root level resolvers are te good idea?

What are your opinions about libraries like GraphQL-Query-Mapper that allow developers to use root level resolvers?

https://github.com/aerogear/graphql-query-mapper

Is there value to modify execute method of graphql layer to ignore execution of nested resolvers?