Generating a model referencing self column name error - phoenix-framework

Question

Given (from scratch )
mix phoenix.new sandpit
mix ecto.create
mix phoenix.gen.json Part parts name part_of:references:parts
mix ecto.migrate
and editing the route so
scope "/api", Sandpit do
pipe_through :api
resources "/parts", PartController, except: [:new, :edit]
end
then running the server and going to
http://localhost:4000/api/parts
I get the errors
info] Running Sandpit.Endpoint with Cowboy using http on port 4000
13 Jul 23:11:37 - info: compiled 5 files into 2 files, copied 3 in 744ms
[info] GET /api/parts
[debug] Processing by Sandpit.PartController.index/2
Parameters: %{}
Pipelines: [:api]
[debug] SELECT p0."id", p0."name", p0."part_of_id", p0."inserted_at", p0."updated_at" FROM "parts" AS p0 [] ERROR query=78.0ms
[info] Sent 500 in 140ms
[error] #PID<0.318.0> running Sandpit.Endpoint terminated
Server: localhost:4000 (http)
Request: GET /api/parts
** (exit) an exception was raised:
** (Postgrex.Error) ERROR (undefined_column): column p0.part_of_id does not exist
(ecto) lib/ecto/adapters/sql.ex:185: Ecto.Adapters.SQL.query!/5
is this a bug with the generator? it seems to want to name things with _id either it seems like a hidden "rule" or it failed to add in _id. If it doesn't auto add _id, I'd think either it would error doing the generation, or use the names as given without crashing?
Trying with
mix phoenix.gen.json Thing things name thing_of_id:references:things
and migrating causes no errors
or is something else going on?

Answer 1

Known issue, currently under debate on what the best solution is :-
https://github.com/phoenixframework/phoenix/issues/1605
Current solution is to remember to suffix references with _id