Virtual properties/implicit join support?

[FelixZY]

I've got a graph-like dataset where I also need to be able to perform geographical queries (e.g. "which node is closest to lat/lng?"). My idea was therefore to have both a vertex with a Location label as well as a locations table. Each row in the locations table would map to a Location vertex, preferably with an ON DELETE CASCADE foreign key.

This would allow me to create a geographical index on the locations table using postgis. It would also allow me to add other properties, provide fuzzy search via pg_trgm and add other indexes specific to my needs. Finally, this setup would provide me with type safety and consistency guarantees.

Using this setup, I would like to be able to write cypher queries that automatically join in my locations values when needed:

MATCH (:Person)-[:BORN_AT]->(l:Location)
WHERE l.lat > 40

Unfortunately, I cannot find a way to do this in Apache AGE at the moment. I understand that I could use regular SQL joins on the dataset returned by cypher(...) but this seems inefficient and perhaps even insufficient for more complicated queries.

So to summarize: does AGE support

• storing vertex/edge properties in a separate table,
• including automatic joining-in of properties from this table when needed,
• as well as ON DELETE CASCADE referencing to the vertex/edge detailed by each row in this table?

If not, what are your thoughts on this as a feature in general? Yay/nay?

[jberzy]

Kind of related to a POC I was experimenting with

https://github.com/jberzy/age/tree/computed-graph.

The POC is a little more generic as it lets you register an existing postgres view as a vertex/edge table.