Feature Request: Allow automatic creation of relationship on duplicate during create/update

[ravensorb]

One of the great things about a graph database is the ability to leverage relationships between existing entities to both describe a larger concept and at the same time reduce the amount of duplication of data. It would be a great feature if structr had an option to automatically associate entities during create/update operations based on a unique constraint violation.

Consider the following (Note: Address has a unique constraint on Line1, City, State, and Postal Code)

• Create: User U1 is located at Address A1 and Works at A2
• Create: User U2 is located at Address A3 and Works at A2
• Update: User U1 is now located at Address A3 and now Works at A5

When I create U1, I would expect A1 and A2 to automatically be created -- this is possible today using the "Automatic Creation of Related Nodes" option.

When I create U2, I would expect A3 to be automatically created and a relationship between U2 and A2 to be created -- this does NOT seem to be possible today (unless I am missing something)

When I update U1, I would expect a relationship be created to Address A3 and a new address created for A5 -- note, I would expect I need to manually remove the relationship to Address A1.

While I know ALL of this is manually possible by creating the individual records and manually creating the relationships it requires a very large number of queries to be executed.

Consider what currently needs to be done

For Create

• Query for U1, if not found Create User U1 (2 API calls)
• Query for A1, if not found Create Address A1 (2 API calls)
• Query for relationship between U1 and A1, if not found create relationship (2 API calls)

For Update

• Query for U1, if found Update User U1 (2 API calls)
• Query for A1, if found Update Address A1, if not found create Address A1 (3 API calls)
• Query for relationship between U1 and A1, if not found create relationship (2 API calls)

For just one instance you have at best 3 API calls and at worst 7 API calls. Where as the idea I am suggestion would amount to at most 2 API calls (Query for User, Create/Update with related records) and structr would internally take care of Creating/Updating related records.

Thoughts?