Hi folks I am using the latest version of dagster, I got a problem where an op returns a data frame but outside doesn't recognize it and throws an error

@op(required_resource_keys={"analytic_db"},)
def query(context) -> pd.DataFrame:
    engine = create_engine(context.resources.analytic_db)
    query = """
        select  * from foo;
    """
    df = pd.read_sql_query(query, con=engine.connect())
    context.log.debug(df.shape)
    return df

@graph
def main():   
    result = query()
    print(result.shape)

context.log.debug

can show my data frame shape, but outside said

AttributeError: 'InvokedNodeOutputHandle' object has no attribute 'shape'

Not sure what I missed? I used all default configs

hi @Quy! the code inside an

@graph

decorated function is purely for defining dependencies between ops (or other graphs). so when the code inside

main

is executed, it's not actually executing the ops themselves, it's just telling dagster which outputs should be fed into which inputs.