Select only necessary columns for optimized queries

What problem does this feature proposal attempt to solve?
I am trying out GraphQL, testing various features in order to include it in some of the web apps I am developing with laravel. Being curious, I created a Schema and fired up all sorts of queries, followed by poking around in the SQL server to see what was going on.

What I discovered was this:

Suppose I have a Type defined which corresponds to a Model, and the database table has an ID column and 26 more columns (A, B C, ..., Z).

No matter what the Type definition is, which in my case only includes some of the columns ( say, ID and A, B, C), I can see that the SQL server is asked to fetch all 27 columns from the DB, using a SELECT * FROM [Table] statement, even though given the definition of the type at most 4 of them would be returned via the resolver.

This is an unnecessary data transfer between the DB and PHP, in my opinion.

Which possible solutions should be considered?

I propose that we construct a more refined SQL query, selecting only the columns that are requested in the GraphQL query.

I am not sure I am competent enough to try and implement this on my own, but I will give it a shot. In any case, I thought I'd share my thoughts.

Start with the GraphQL selection, Laravel model, and resolver behavior described in the issue, then compare the requested fields with the generated SELECT * query. Done means queries fetch only the columns requested by the GraphQL query without breaking resolver results or related behavior.