Hi
@Tim we can help you work through this, but in general we’re not fans of dynamic pipelines like this as it really damages the principles of reusability we’re trying to encourage.
In general we try to push users to follow the following rules:
• Nodes themselves should be pure python functions (with no side effects, or concept of IO), that should be delegated to the catalog. Your SQL call from a node here violates this assumption already.
• If you are going to generate a dynamic catalog, the
after_context_created
or
before_pipeline_run
hooks are the right way to do that