I’ve been trying to connect 2 tables (for over a week) using a join table as described in Ecto Association Guide – Ecto v2.2.11. Best docs I could find.

I have three tables: organizations, employees, organization_employees. There is no schema for organization_employees

The latest issue is a PK error. I insert and preload the “objects”, but when I try to update the DB with put_assoc I get an error. The “objects” do exist in the DB and PK matches.

A breif snipped is below while the full code is here: Loading put_assoc · GitHub

org_update = Ecto.Changeset.put_assoc(org_changeset, :employees, [e_insert])
#Ecto.Changeset<
  action: nil,
  changes: %{
    email: "toys@rus.com",
    employees: [
      #Ecto.Changeset<action: :update, changes: %{}, errors: [],
       data: #Employee<>, valid?: true>
    ],
    name: "Toys R Us",
    phone: "777777777",
    slug: "toys-r-us"
  },
  errors: [],
  data: .Organization<>,
  valid?: true
>
Repo.update(org_update) 

** (Ecto.NoPrimaryKeyValueError) struct `%TurnStile.Company.Organization{__meta__: #Ecto.Schema.Metadata<:built, "organizations">, id: nil, email: nil, name: nil, slug: nil, phone: nil, employees: #Ecto.Association.NotLoaded<association :employees is not loaded>, inserted_at: nil, updated_at: nil}` is missing primary key value
    (ecto 3.9.4) lib/ecto/repo/schema.ex:977: anonymous fn/3 in Ecto.Repo.Schema.add_pk_filter!/2
    (elixir 1.14.1) lib/enum.ex:2468: Enum."-reduce/3-lists^foldl/2-0-"/3
    (ecto 3.9.4) lib/ecto/repo/schema.ex:416: Ecto.Repo.Schema.do_update/4
    iex:13: (file)

You can see in the full code that I do preload, so I don’t know why it says employees is not loaded. Even after preloading, employees is an empy list [], even though there is an employee in the DB. Vice versa for organizations. Not sure if this is how it’s supposed to be.

This has been SO error prone and tedious. Any help is appreciated.