Performance within instances()

Continuing from Ie43beecf37945b2bb7fff0aaa597a597293daa18,
also observed is the overhead of PathRegsitry memoized token
functions, as these paths are not cached in the case of
long joinedloader chains.  The memoizations here were made
with short paths in mind, and have been replaced with
an inlined straight create of these paths up front, producing
callcounts very similar to 0.8. Combined with the previous
optimizations, 1.1 now runs the "joined eager load of one row"
worst case test in about 40% fewer calls than 0.8 and 60%
fewer than 1.1.5.

Change-Id: Ib5e1c1345a1dd8edfbdb3fed06eb717d4e164d31
Fixes: #3915

1 files changed, 20 insertions, 0 deletions

diff --git a/test/aaa_profiling/test_orm.py b/test/aaa_profiling/test_orm.py
index 1380ef5c3..61fd532e5 100644
--- a/test/aaa_profiling/test_orm.py
+++ b/test/aaa_profiling/test_orm.py
@@ -655,3 +655,23 @@ class JoinedEagerLoadTest(fixtures.MappedTest):
                 q._compile_context()
         go()
 
+    def test_fetch_results(self):
+        A, B, C, D, E, F, G = self.classes('A', 'B', 'C', 'D', 'E', 'F', 'G')
+
+        sess = Session()
+
+        q = sess.query(A).options(
+            joinedload(A.bs).joinedload(B.cs).joinedload(C.ds),
+            joinedload(A.es).joinedload(E.fs),
+            defaultload(A.es).joinedload(E.gs),
+        )
+
+        context = q._compile_context()
+
+        @profiling.function_call_count()
+        def go():
+            for i in range(100):
+                obj = q._execute_and_instances(context)
+                list(obj)
+                sess.close()
+        go()