Configuration and State

We need to fill in implementations for each of the required functions, but we won't need all of these to work just yet.

First, you'll see that we define two types: Configuration, and State.

Let's define those now above the connector and start function:

type Configuration = {
  tables: TableConfiguration[];
  filename: string;
};

type TableConfiguration = {
  tableName: string;
  columns: { [k: string]: Column };
};

type Column = {};

type State = {
  db: Database;
};

Configuration is the type of the connector's configuration, which will be read from a directory on disk. By convention, this configuration should be enough to reproducibly determine the NDC schema, so for our SQLite connector, we configure the connector with a list of tables that we want to expose. Each table is defined by its name and a list of columns. Columns don't have any specific configuration yet, but we leave an empty object type here because we might want to capture things like column types later on.

State

The State type is for things like connection pools, handles, or any non-serializable state that gets allocated on startup, and which lives for the lifetime of the connector. For our connector, we need to keep a handle to our sqlite database.

Cool, so now that we've got our types defined, we can fill in the function definitions which the connector requires in order to interact with our SQLite database and Hasura DDN. Let's do that in the next step.