SQLAlchemy implementation

[somada141]

Hi there and thanks a lot for posting this :). Would you happen to have figured out how to glue together Graphene, Graphene-SQLAlchemy, and Falcon?

[alecrasmussen]

I haven't worked with Graphene-SQLAlchemy myself, but it looks to me like you could create your schema the same way you usually would and then use that. To copy from Graphene-SQLAlchemy's first example, you could try something like this:

Base = declarative_base()

class UserModel(Base):
    __tablename__ = 'department'
    id = Column(Integer, primary_key=True)
    name = Column(String)
    last_name = Column(String)

class User(SQLAlchemyObjectType):
    class Meta:
        model = UserModel

# Define a GraphQL query schema.
class Query(graphene.ObjectType):
    users = graphene.List(User)

    def resolve_users(self, info):
        query = User.get_query(info)  # SQLAlchemy query
        return query.all()

# Create the schema that will be used to resolve GraphQL requests.
schema = graphene.Schema(query=Query)

The main other thing that you may need to add is to pass in your db handler via context when calling schema.execute.

[somada141]

I think that's where my primary hangup was, namely passing the SQLAlchemy session to the context while ensuring that we're using per-request sessions instead of a hanging one that eventually times out.

Further tinkering and http://docs.graphene-python.org/projects/sqlalchemy/en/latest/tips/#querying makes me think that there are three options here.:

1. Create a scoped_session and pass it to the resource-class which in-turn passes it to the context.
2. Create a scoped_session and boil it in the SQLAlchemy Base much like https://medium.com/@fasterpancakes/graphql-server-up-and-running-with-50-lines-of-python-85e8ada9f637
3. Allow the resource-class to create an engine to the DB and explicitly create sessions per request which it then passes to the context.

Thanks a lot for the code again! Any plans to make it into a pypi package for easier integration?