Yes, I think you are on the right track.

I think the issue with splitting the XML is that XML::RDB tries to extract shared values into a one to many relationship, to avoid redundancy. This is great, but it defeats the possibility of splitting.

If I'm right, then during the second data import, it will try to create the shared values as foreign key entries, but without checking to see if they already exist.

I was hoping someone would suggest a way to get it to check for pre-existing fk entries, or some such.

Looks like I might have to go down a more manual route, which is a shame as XML::RDB looks pretty elegant otherwise.