When I'm updating our data model, sometimes I want...
# gooddata-cloud
m
When I'm updating our data model, sometimes I want to "lift" a primary key out to a new, parent data set. For example, I have a "Web Conversion" table, which had a primary key of "Conversion ID". I'm lifting that up to it's own SQL data set, since other tables use the same conversion IDs, and I'd like a single model to relate them all (see image below). However, whenever I try to adjust a field to be a relationship instead of an attribute, it screws up any downstream dependencies (see pic). Is there not a way to do this w/o screwing up all my insights, metrics, and dashboards?
If I undo "Delete attribute "Conversion ID"", I don't lose anything but now I have two conversion IDs in the mode. The attribute and the relationship.
For clarity, this is the model before making the changes I described:
m
Hi Michael, in this case I think the best approach is to: • create new empty dataset • move the attribute to this new dataset • make it primary key of this dataset • create a reference from this new dataset to the original one (and map it to the column there) this way you should not lose any of your reports as it will not delete the object but just moves it to another dataset (while keeping its identifier).