schema_name config option ignored about guardian_db HOT 11 CLOSED

token_source would be a function to load the table name, for example:

defp token_source() do
  Application.fetch_env!(:guardian_db, GuardianDb)
  |> Keyword.get(:schema_name, "guardian_tokens")
end

This isn't documented anywhere in one place, unfortunately. Some bits are in the Ecto.Schema docs, some in Ecto.Query and some in Ecto. This would use the same technique as the one described in the polymorphic belongs_to docs.

from guardian_db.

I opened a PR to ecto with improvements to the docs around the __meta__ field: elixir-ecto/ecto#2228

from guardian_db.

It seems that the error was because I had changed the schema_name value without recompiling the dependencies. I'll submit a PR to add a note about this to the documentation.

from guardian_db.

@davejlong the correct solution for this is to change

Talking to Michal about it, use module attributes for reading this configurations is a bad practice because of the situation you ran into, which is pretty common to run into

from guardian_db.

hhhhmmm I think I can't switch it to some function call 😢 so it will be better to actually add some documentation I think

from guardian_db.

Ecto supports using {"table_name", Schema} as the source of a query. This can be leveraged by changing, for example:

GuardianDb.repo.get_by({token_source(), Token}, jti: jti, aud: aud)

It's also supported when creating structs like here:

%Token{}
|> Ecto.put_meta(source: token_source())
#...

Ecto makes sure that structs extracted from tagged queries when used with update or delete issue correct queries.

This should cover all the use cases.

from guardian_db.

@michalmuskala what is that token_source the schema_name in this case? What about the schema "name over here" situation?

Could you point me to the doc spot so I can read about it please

from guardian_db.

@michalmuskala I notice this https://github.com/elixir-ecto/ecto/blob/v2.2.3/lib/ecto/schema.ex#L378 if I use embedded_schema I dont need any random name on the schema name because I will be using what you recommend.

What are your thoughts about it?

from guardian_db.

embedded_schema generates a slightly different struct from schema - in particular, it doesn't track the lifetime of the struct (since it doesn't matter for embedded associations). The difference is the __meta__ key.

from guardian_db.

@michalmuskala wouldn't be worth to add the differences in Ecto documentation?

Situation like this one people dont have to ask you what the differences are and it's important to know for apply the correct use of it.

For example right now, I don't know what you meant by lifetime of the struct and other situations.

What do you think?

from guardian_db.

@michalmuskala you are the boss!

from guardian_db.