fixed code for theia data

compose.yml

@@ -19,7 +19,7 @@ services:
       - BATCH_SIZE
       - SINGLE_STATEMENT_NODES
     volumes:
-      - ./data/cadets.zip:/data.zip:ro
+      - ./data/theia.zip:/data.zip:ro
       - ./scripts/pg_insert.py:/app/script.py:ro
     depends_on:
       postgres:

insert_tests.py

@@ -177,6 +177,14 @@ if __name__ == "__main__":
             ),
             "env": env_default,
         },
+        {
+            "name": "experiment_no_index_single",
+            "compose_files": (
+                "compose.yml",
+                "compose.edge-id.yml",
+            ),
+            "env": env_single,
+        },
     )
     for experiment in experiments:
         run_experiment(**experiment)

postgres/initdb/03-initdb_trigger_edge_id.sql

@@ -21,9 +21,14 @@ begin
 		where type='Subject'
 		UNION ALL
 
+		select id, content#>>'{localPrincipal,UUID}', 'has_owning_principal'
+		from new_vertices
+		where type='FileObject'
+		UNION ALL
+
 		select id, content->>'localPrincipal', 'has_owning_principal'
 		from new_vertices
-		where type='Subject' or type='FileObject'
+		where type='Subject'
 		UNION ALL
 
 		select id, content#>>'{sourceUUID,UUID}', 'ingests'

postgresql.conf

@@ -192,7 +192,7 @@ dynamic_shared_memory_type = posix	# the default is usually the first option
 #backend_flush_after = 0		# measured in pages, 0 disables
 #effective_io_concurrency = 1		# 1-1000; 0 disables prefetching
 #maintenance_io_concurrency = 10	# 1-1000; 0 disables prefetching
-max_worker_processes = 6		# (change requires restart)
+#max_worker_processes = 8		# (change requires restart)
 #max_parallel_workers_per_gather = 2	# limited by max_parallel_workers
 #max_parallel_maintenance_workers = 2	# limited by max_parallel_workers
 #max_parallel_workers = 8		# number of max_worker_processes that

scripts/pg_insert.py

@@ -22,7 +22,7 @@ NODE_TYPES: Final[dict] = {
         ("has_parent", ("parentSubject", "UUID")),
         ("has_owning_principal", ("localPrincipal",)),
     ),
-    "FileObject": (("has_owning_principal", ("localPrincipal",)),),
+    "FileObject": (("has_owning_principal", ("localPrincipal", "UUID")),),
     "UnnamedPipeObject": (
         ("ingests", ("sourceUUID", "UUID")),
         ("outputs", ("sinkUUID", "UUID")),