[Sqlalchemy-tickets] Issue #3526: unused systems in bulk (zzzeek/sqlalchemy)

[Mike B. <iss...@bi...>]

New issue 3526: unused systems in bulk
https://bitbucket.org/zzzeek/sqlalchemy/issues/3526/unused-systems-in-bulk

Mike Bayer:

based on things found via #3525, _postfetch is called in bulk update for unknown reasons, as no tests fail removing it.  need to isolate why _postfetch is needed and add tests, or not:

```
#!diff


diff --git a/lib/sqlalchemy/orm/persistence.py b/lib/sqlalchemy/orm/persistence.py
index c785a4d..6487228 100644
--- a/lib/sqlalchemy/orm/persistence.py
+++ b/lib/sqlalchemy/orm/persistence.py
@@ -672,6 +672,8 @@ def _emit_update_statements(base_mapper, uowtransaction,
                         connection, value_params in records:
                     c = cached_connections[connection].\
                         execute(statement, params)
+
+                    # TODO: why with bookkeeping=False?
                     _postfetch(
                         mapper,
                         uowtransaction,
@@ -694,6 +696,8 @@ def _emit_update_statements(base_mapper, uowtransaction,
                     execute(statement, multiparams)
 
                 rows += c.rowcount
+
+                # TODO: why with bookkeeping=False?
                 for state, state_dict, params, mapper, \
                         connection, value_params in records:
                     _postfetch(
@@ -964,6 +968,8 @@ def _postfetch(mapper, uowtransaction, table,
     after an INSERT or UPDATE statement has proceeded for that
     state."""
 
+    # TODO: bulk is never non-False, need to clean this up
+
     prefetch_cols = result.context.compiled.prefetch
     postfetch_cols = result.context.compiled.postfetch
     returning_cols = result.context.compiled.returning

```