Querying for allSpecies and homeworld executes 38 SQL queries

从 issue 中的 allSpecies 查询和 SQL trace 开始，然后检查加载每个 homeworld 的 Graphene-Django resolver 和 Django ORM 路径。比较此关系使用 select_related() 和 prefetch_related() 的效果，并以报告中的查询作为基线；完成的标准是示例将请求减少到一条或两条 SQL 查询。

This query:

{
  allSpecies {
    edges {
      node {
        name
        homeworld {
          name
          gravity
        }
      }
    }
  }
}

Executes 38 sql queries, looking something like this:

(0.001) SELECT COUNT(*) AS "__count" FROM "starwars_species"; args=()
(0.000) SELECT "starwars_species"."id", "starwars_species"."created", 
"starwars_species"."edited", "starwars_species"."name", 
"starwars_species"."classification", "starwars_species"."designation", 
"starwars_species"."average_height", "starwars_species"."skin_colors", 
"starwars_species"."hair_colors", "starwars_species"."eye_colors", 
"starwars_species"."average_lifespan", "starwars_species"."homeworld_id", 
"starwars_species"."language" FROM "starwars_species" LIMIT 37; args=()

(0.000) SELECT "starwars_planet"."id", "starwars_planet"."created", 
"starwars_planet"."edited", "starwars_planet"."name", 
"starwars_planet"."rotation_period", "starwars_planet"."orbital_period", 
"starwars_planet"."diameter", "starwars_planet"."climate", 
"starwars_planet"."gravity", "starwars_planet"."terrain", 
"starwars_planet"."surface_water", "starwars_planet"."population" FROM 
"starwars_planet" WHERE "starwars_planet"."id" = 9; args=(9,)

(0.000) SELECT "starwars_planet"."id", "starwars_planet"."created", 
"starwars_planet"."edited", "starwars_planet"."name", 
"starwars_planet"."rotation_period", "starwars_planet"."orbital_period", 
"starwars_planet"."diameter", "starwars_planet"."climate", 
"starwars_planet"."gravity", "starwars_planet"."terrain", 
"starwars_planet"."surface_water", "starwars_planet"."population" FROM 
"starwars_planet" WHERE "starwars_planet"."id" = 14; args=(14,)

... repeat another 36 times

Full list here: https://gist.github.com/simonw/762a98de85a2f113d1d7ec6026f73ee4

Is it possible to detect this case and then use select_related() or prefetch_related() here to join against the relevant table and get everything done in just one or two queries?

If so, an example showing how to do that would be a fantastic illustration of a slightly more complex use-case for graphene-django (I tried myself and couldn't figure out how to do it).