One Relationship per row
Luz Calero Forero
Would be nice to have the option to generate one relationship per row instead of one relationship per from_id <> to_id (merging multiple relationships into one). Which effectively makes it so that in the case that there are multiple rows in your csv with the same from and to nodes, only the last one is being imported into the graph.
Mark Faga
Our use-case is single value only right now.
I'd say if that is significantly easier than supporting compound keys, there are reasonable work arounds to make that OK. If it's the same amount of work to support compound keys though, than why not.
Gregory King
Mark Faga Thanks, useful input.
Gregory King
Mark Faga as a follow on question, do you have one single value that would uniquely identify relationships - or would you be reliant on a compound of values?
Mark Faga
Awesome, thanks for the quick response!
Gregory King
Thanks for the feedback Mark. We'll look into adding an optional Id and corresponding uniqueness constraint for relationships as part of the Import UI to address this.
Mark Faga
+1 to this. The import utility is effectively not-usable until this exists for us. To answer Michael Hunger's question - yes, there is other metadata we are loading onto the relationships that differentiates them.
Michael Hunger
Do you have then ID-like properties that identify your relationships uniquely? Otherwise it becomes hard to create idempotent import operations.