Performance on selecting all columns instead of specified attributes

Is there a reason the _model_class queries all columns?

https://github.com/cerebris/jsonapi-resources/blob/bf4b4cd7d79da453372880068e167066f0d05f47/lib/jsonapi/active_relation_resource.rb#L286-L288

This is a huge performance impact on models with many columns that may not ever be used by the resource.

It seems for performance reasons it makes sense to only select the attributes defined in the JSONAPI::Resource class.

A workaround for now is to override each resource using .select to retrieve only the fields that are specifically defined.

attributes :name, :address

def self.records(_options = {})
  super.select(self.fields)
end

This should also take into account sparse fieldsets parameters, which would further limit which columns are selected.

Empieza en lib/jsonapi/active_relation_resource.rb, en la consulta _model_class vinculada, y sigue cómo se representan los atributos de Resource y los sparse fieldsets. El trabajo está terminado cuando esas consultas seleccionen únicamente los campos necesarios para el resource y el fieldset solicitado, preservando el comportamiento existente de records y cubriendo las rutas de consulta relevantes.