Hi there…
I’m pleased to see that bulk loading can read URIs and map to UIDs directly, while saving an xid edge…
I’d like to keep my reference data in that format, but then I can’t easily live load some incremental data through the mutations API. What I need to do is crawl through my file, query dgraph to find the xid → uid maps that exist, assign new blanks to those that don’t, and do all of that externally to the database. That’s a lot slower than anything… Wouldn’t there be a way I could live load some RDF that have some <whatever:node_id> as subject, and have the program do the mapping, and assignation of new UIDs?
Why the two different behaviors?
btw, thanks for the great database 