why does query.statement add no_replacement_traverse

[sqlalchemy-bot]

Migrated issue, originally created by Michael Bayer (@zzzeek)

e.g.

diff --git a/lib/sqlalchemy/orm/query.py b/lib/sqlalchemy/orm/query.py
index 272fed3e23..c623f10a95 100644
--- a/lib/sqlalchemy/orm/query.py
+++ b/lib/sqlalchemy/orm/query.py
@@ -514,7 +514,7 @@ class Query(object):
 
         # TODO: there's no tests covering effects of
         # the annotation not being there
-        return stmt._annotate({'no_replacement_traverse': True})
+        return stmt
 
     def subquery(self, name=None, with_labels=False, reduce_columns=False):
         """return the full SELECT statement represented by

here is a confusing effect of it being there:

from sqlalchemy import *
from sqlalchemy.orm import *
from sqlalchemy.ext.declarative import declarative_base

Base = declarative_base()


class MyTable(Base):
    __tablename__ = 'my_table'
    score = Column(Integer, primary_key=True)

my_alias = aliased(MyTable, name="my_alias")

s = Session()

query_2 = s.query(MyTable).union(s.query(MyTable))

query_2 = query_2.filter(MyTable.score > 5)

stmt1 = exists().where(my_alias.score > MyTable.score)
stmt2 = s.query(my_alias).filter(my_alias.score > MyTable.score).exists()

print(query_2.add_column(stmt1))

print(query_2.add_column(stmt2))

# correct:
SELECT anon_1.my_table_score AS anon_1_my_table_score, EXISTS (SELECT * 
FROM my_table AS my_alias 
WHERE my_alias.score > anon_1.my_table_score) AS anon_2 
FROM (SELECT my_table.score AS my_table_score 
FROM my_table UNION SELECT my_table.score AS my_table_score 
FROM my_table) AS anon_1 
WHERE anon_1.my_table_score > :score_1

# incorrect:
SELECT anon_1.my_table_score AS anon_1_my_table_score, EXISTS (SELECT 1 
FROM my_table AS my_alias, my_table 
WHERE my_alias.score > my_table.score) AS anon_2 
FROM (SELECT my_table.score AS my_table_score 
FROM my_table UNION SELECT my_table.score AS my_table_score 
FROM my_table) AS anon_1 
WHERE anon_1.my_table_score > :score_1

[sqlalchemy-bot]

Michael Bayer (@zzzeek) wrote:

running it through at https://gerrit.sqlalchemy.org/#/c/zzzeek/sqlalchemy/+/823/ to see if this impacts anything

[sqlalchemy-bot]

Michael Bayer (@zzzeek) wrote:

Don't apply no-traverse to query.statement

Fixed long-standing issue in :class:.Query where a scalar subquery such
as produced by :meth:.Query.exists, :meth:.Query.as_scalar and other
derivations from :attr:.Query.statement would not correctly be adapted
when used in a new :class:.Query that required entity adaptation, such as
when the query were turned into a union, or a from_self(), etc. The change
removes the "no adaptation" annotation from the :func:.select object
produced by the :attr:.Query.statement accessor.

Change-Id: I554e0e909ac6ee785ec3b3b14aaec9d235aa28cf
Fixes: #4304

→ fb37722

[sqlalchemy-bot]

Changes by Michael Bayer (@zzzeek):

• changed status to closed