I’m looking to build a transforms in which I have the acl for each row in a “markings” column. Is it possible to have the transforms output directly to a restricted view without materializing a backing dataset?
I’m thinking if I output to a dataset then I will need to set all markings in the data onto the dataset. Making dataset acls depend on dynamic data like that seems a bit unwieldy, so asking to see if there’s a better way.
Hey @npham Pipeline Builder does not currently support outputting a restricted view directly (I believe code repositories also doesn’t support this but can have someone from their team confirm)
However, wouldn’t outputting a restricted view directly still make the RV policies dependent on dynamic data?
I think the policy shouldn’t change (i.e. you still have rules like “Users’ classification access needs to satisfy markings column”). It’s just the acl requirement of each row will depend on what’s in the markings column, but that’s how restricted view always works.
Would it work for you to have an ‘admin’ marking on the direct output dataset and then just remove that marking in the RV?