Currently there is no way to bind values to placeholders in statements. This can be implemented in SqliteStatement.

See https://sqlite.org/c3ref/bind_blob.html

Currently the project builds in its own little world. I'm unfamiliar with pony-stable but it'd be nice if this project could work well with it.

SQLite supports registering custom functions so they can be used from SQL. We should support this feature.

See https://sqlite.org/c3ref/create_function.html

We should support the Iterator protocol either on SqliteStatement itself or as another class that wraps SqliteStatement. The latter feels right but an experiment should yield the right solution.

We should see about setting up CI with various platforms so we can know whether this library works outside of the official Pony docker image.

Currently if a column is null and is retrieved as a particular type it will be a zero-value for that type. Is this the right way to accomplish this or should we be more explicit about the return type? The API user can check if a column is explicitly null via SqliteStatement.data_type(column_index).

Considering that all values in SQLite are coercible does it make sense to introduce None onto the methods that return values from SQLite? Or should we keep it as is?