Is there a good way to wire up input assets to ops...
# ask-community
s
Is there a good way to wire up input assets to ops. I have a graph of ops and a job for it. I have some assets produced in a software defined assets fashion. How can I best wire this up. Producing AssetMaterializtions/Observations from ops is easy but I am missing the other way. Do I need an asset_sensor and ins={“dfs”: In(AssetKey())}. It still does not show in the asset graph. Do I have to use @assets or is there a way to integrate normal ops into this. I would like to avoid spamming assets with op outputs of which some are quite temporary. Ops approach is quite flexible but assets approach is more logical and clean. I also once saw an example of having an asset which contained a more detailed graph a folder how can I achieve that?
b
I don't know but asked the same question awhile back so am keenly following this
c
This is not currently possible, but we know that this is a main area of improvement for software defined assets. We're currently working on building out functionality to combine assets and ops within a job which will be released in the next couple months.
b
Perfect! Thanks, Claire!
dagster yay 1