Stephen Bailey
02/07/2024, 7:12 PMScout
02/07/2024, 7:12 PMgraph_asset
, you can use the ins
parameter of the @graph_asset
decorator to map input names to AssetIn
objects that specify the dependencies. The AssetIn
object can be configured with dagster_type=Nothing
, indicating that the asset does not expect a runtime value but does have a dependency on another asset. Here's an example from the documentation that demonstrates how to specify dependencies for a `graph_asset`: python from dagster import graph_asset, asset, AssetIn, op, Nothing, In @asset def upstream(): ... @op(ins={"upstream": In(Nothing)}) def my_op(): return ... @graph_asset( ins={"upstream": AssetIn(dagster_type=Nothing)}, ) def my_graph_asset(upstream): return my_op(upstream)
In this example, my_graph_asset
has a non-argument dependency on the upstream
asset. The upstream
input to the my_graph_asset
function is mapped to an AssetIn
with dagster_type=Nothing
, which establishes the dependency without passing a runtime value.