Multi-Hop Traversal

Where we left off

Lesson 9's checkpoint chained two graph.get_neighbors() calls by hand to walk two hops. graph.traverse() is the function that does that in one call, for any number of hops, and it's the workhorse this whole intermediate tier builds on.

Growing the example: reporting structure and subsidiaries

This tier extends the running schema. people gains manager_id (a self-referencing foreign key) and seniority_years; companies gains parent_company_id and ownership_pct; a new projects table tracks who leads what. Four edges now exist: works_at (unchanged), reports_to (people.manager_id -> people.id, one-way, Dan reports to Bob who reports to Alice), subsidiary_of (companies.parent_company_id -> companies.id, one-way, weighted by ownership_pct), and led_by (projects.lead_person_id -> people.id). Every lesson script in this tier sets up this full schema itself, so each one still runs on its own.

`graph.traverse()`

SELECT depth, node_table_name, node_id FROM graph.traverse(
  'public.people'::regclass, 'p4', 2,
  edge_types := ARRAY['reports_to'],
  direction := 'out',
  hydrate := false
);

The three required positional arguments are the seed table, seed ID, and max_depth. Starting from Dan (p4) and following reports_to outward two hops reaches Bob at depth 1, then Alice at depth 2, Dan's manager's manager. hydrate := false skips reading the full row for each result, useful here since only the chain of IDs matters, not the row contents. This is what Lesson 9's two chained get_neighbors() calls were really doing, one function call instead of writing the loop yourself, and it works for any depth, not just two.