sqlalchemy-tickets — trac tickets and updates forwarded to this list

[Sqlalchemy-tickets] Issue #3969: negation of labeled element? (zzzeek/sqlalchemy)

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

New issue 3969: negation of labeled element?
https://bitbucket.org/zzzeek/sqlalchemy/issues/3969/negation-of-labeled-element

Michael Bayer:

this is technically something that semantically doesn't make sense, but in the "what I expect" category it does, which is to negate the inner element of a label.


```
#!python

from sqlalchemy import *

expr = or_(column('x') == 1, column('y') == 2)

bug = not_(expr.label('foo'))

from sqlalchemy.dialects import sqlite

print bug.compile()
print bug.compile(dialect=sqlite.dialect())

```

```
#!


NOT x = :x_1 OR y = :y_1
x = ? OR y = ? = 0
```

so the parenthesis are lost because we don't unwrap the label(), and we in fact hit a rule in ColumnElement._negate() where it assumes we're negating a true/false, so for a non-native-boolean it then sticks on "= 0" to the whole thing to make things worse.

[Sqlalchemy-tickets] Issue #3968: Idle in transaction connections can not be returned to pool (zzzeek/sqlalchemy)

[caijinrong <iss...@bi...>]

New issue 3968: Idle in transaction connections can not be returned to pool
https://bitbucket.org/zzzeek/sqlalchemy/issues/3968/idle-in-transaction-connections-can-not-be

caijinrong:

Hi Michael,

SQLA is used in my python project.
And in my code, I have the green thread used. And the green thread was killed in some of my code. After the the thread was killed, and the transaction was just began. And the database connection was still in postgresql server side with status "idle in transaction".
This caused my connection was exhausted in the pool.

Traceback (most recent call last):
  File "/usr/lib/python2.7/dist-packages/eventlet/hubs/hub.py", line 457, in fire_timers
    timer()
  File "/usr/lib/python2.7/dist-packages/eventlet/hubs/timer.py", line 58, in __call__
    cb(*args, **kw)
  File "/usr/lib/python2.7/dist-packages/eventlet/greenthread.py", line 214, in main
    result = function(*args, **kwargs)
  File "/tmp/test.py", line 48, in operate_db
    query = session.query(Test).all()
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/orm/query.py", line 2588, in all
    return list(self)
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/orm/query.py", line 2736, in __iter__
    return self._execute_and_instances(context)
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/orm/query.py", line 2749, in _execute_and_instances
    close_with_result=True)
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/orm/query.py", line 2740, in _connection_from_session
    **kw)
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/orm/session.py", line 905, in connection
    execution_options=execution_options)
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/orm/session.py", line 910, in _connection_for_bind
    engine, execution_options)
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/orm/session.py", line 334, in _connection_for_bind
    conn = bind.contextual_connect()
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/engine/base.py", line 2039, in contextual_connect
    self._wrap_pool_connect(self.pool.connect, None),
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/engine/base.py", line 2074, in _wrap_pool_connect
    return fn()
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/pool.py", line 376, in connect
    return _ConnectionFairy._checkout(self)
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/pool.py", line 714, in _checkout
    fairy = _ConnectionRecord.checkout(pool)
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/pool.py", line 480, in checkout
    rec = pool._do_get()
  File "/usr/lib/python2.7/dist-packages/sqlalchemy/pool.py", line 1054, in _do_get
    (self.size(), self.overflow(), self._timeout))
TimeoutError: QueuePool limit of size 2 overflow 0 reached, connection timed out, timeout 30

Questions:
1. When will the connection returned to the pool? 
In my understanding, when the rollback/commit was listened, the connection was returned.

2. Will the connection returned to the pool if my green thread was killed?
>From the code, we can see the connection object seems still in the pool with active status.

3. Will the connection returned to the pool if the db transaction(id in transaction) was rolled back by my database(idle_in_transaction_session_timeout).
If not, how can we handle the connection whose status is in "idle in transaction"?

SQLA version: 1.0.11
PostgreSQL version: 9.6
See attachment for test case.
Please let me know if you need more information.
Thanks.

Responsible: zzzeek

[Sqlalchemy-tickets] Issue #3967: subquery eager loading doesn't use join_depth correctly (zzzeek/sqlalchemy)

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

New issue 3967: subquery eager loading doesn't use join_depth correctly
https://bitbucket.org/zzzeek/sqlalchemy/issues/3967/subquery-eager-loading-doesnt-use

Michael Bayer:

the join_depth parameter w/ subquery eager loading causes it to just load infinitely as long as there are objects

```
#!diff


diff --git a/test/orm/test_subquery_relations.py b/test/orm/test_subquery_relations.py
index 1396281..2b8068d 100644
--- a/test/orm/test_subquery_relations.py
+++ b/test/orm/test_subquery_relations.py
@@ -517,6 +517,29 @@ class EagerTest(_fixtures.FixtureTest, testing.AssertsCompiledSQL):
         eq_(self.static.user_address_result,
             sess.query(User).order_by(User.id).all())
 
+    def test_cyclical_explicit_join_depth(self):
+        """A circular eager relationship breaks the cycle with a lazy loader"""
+
+        Address, addresses, users, User = (self.classes.Address,
+                                           self.tables.addresses,
+                                           self.tables.users,
+                                           self.classes.User)
+
+        mapper(Address, addresses)
+        mapper(User, users, properties=dict(
+            addresses=relationship(Address, lazy='subquery', join_depth=1,
+                                   backref=sa.orm.backref(
+                                       'user', lazy='subquery', join_depth=1),
+                                   order_by=Address.id)
+        ))
+        is_(sa.orm.class_mapper(User).get_property('addresses').lazy,
+            'subquery')
+        is_(sa.orm.class_mapper(Address).get_property('user').lazy, 'subquery')
+
+        sess = create_session()
+        eq_(self.static.user_address_result,
+            sess.query(User).order_by(User.id).all())
+
     def test_double(self):
         """Eager loading with two relationships simultaneously,
             from the same table, using aliases."""

```

will cause a crash.

similarly, the test "test_lazy_fallback_doesnt_affect_eager" doesn't make sense; the subquery loads are emitted for all levels deep when the join_depth=1.   it should only eagerly load one level deep.

[Sqlalchemy-tickets] Issue #3966: ancient mysql utc_timestamp rule seems unnecessary (zzzeek/sqlalchemy)

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

New issue 3966: ancient mysql utc_timestamp rule seems unnecessary
https://bitbucket.org/zzzeek/sqlalchemy/issues/3966/ancient-mysql-utc_timestamp-rule-seems

Michael Bayer:

from https://github.com/zzzeek/sqlalchemy/pull/355

```
#!


commit a21dc3d
Author: Mike Bayer mi...@zz...
Date: Wed Sep 10 21:09:04 2008 +0000

- the function func.utc_timestamp() compiles to UTC_TIMESTAMP, without
the parenthesis, which seem to get in the way when using in
conjunction with executemany().

```

we have a hardcoded UTC_TIMESTAMP render that seems unnecessary and gets in the way of parameters like fsp.    the ancient assertion that this "gets in the way with executemany()" must be tested.   It would only be some ancient mysqldb driver issue at best.

[Sqlalchemy-tickets] Issue #3965: Bug with hybrid property ? (zzzeek/sqlalchemy)

[Olivier Le M. <iss...@bi...>]

New issue 3965: Bug with hybrid property ?
https://bitbucket.org/zzzeek/sqlalchemy/issues/3965/bug-with-hybrid-property

Olivier Le Moign:

I use SQLALchemy ORM with a declarative base and SQLITE.
I use hybrid properties to encrypt TOTP secrets as follows:

```
#!python

class FactorsTOTP(Model):
    app_secret = 'foobar'

    user_id = Column(Integer, ForeignKey('user.id'), nullable=False)
    _key = Column('key', String, nullable=False)

    @hybrid_property
    def key(self):
        cipher_key = urlsafe_b64encode(self.app_secret.ljust(32, b'0'))
        f = Fernet(cipher_key)
        return f.decrypt(self._key.encode('utf-8'))

    @key.setter
    def key(self, value):
        cipher_key = urlsafe_b64encode(self.app_secret.ljust(32, b'0'))
        f = Fernet(cipher_key)
        self._key = f.encrypt(value.encode('utf-8')).decode('ascii')
```

My problem is, when creating a row: factors_totp = models.FactorsTOTP(key=totp_key)

I'm not sure why, but the key getter is being called after the setter and fails on self._key.encode('utf-8'). Simple Python operations work (self._key + 'suffix'), but not calling a function (bytes(self.key, encoding='utf-8) fails as well).

```
#!python

Traceback (most recent call last):
  File "/Users/olemoign/.pyenv/versions/rta/bin/rta_add_admin", line 11, in <module>
    load_entry_point('rta', 'console_scripts', 'rta_add_admin')()
  File "/Users/olemoign/Developer/Parsys/rta/rta/scripts/add_admin.py", line 27, in main
    password, totp_key = _command_line_add_user(env['request'], admin)
  File "/Users/olemoign/Developer/Parsys/rta/rta/services/users.py", line 363, in _command_line_add_user
    factors_totp = models.FactorsTOTP(key=totp_key)
  File "<string>", line 4, in __init__
  File "/Users/olemoign/.pyenv/versions/3.5.2/envs/rta/lib/python3.5/site-packages/sqlalchemy/orm/state.py", line 414, in _initialize_instance
    manager.dispatch.init_failure(self, args, kwargs)
  File "/Users/olemoign/.pyenv/versions/3.5.2/envs/rta/lib/python3.5/site-packages/sqlalchemy/util/langhelpers.py", line 66, in __exit__
    compat.reraise(exc_type, exc_value, exc_tb)
  File "/Users/olemoign/.pyenv/versions/3.5.2/envs/rta/lib/python3.5/site-packages/sqlalchemy/util/compat.py", line 187, in reraise
    raise value
  File "/Users/olemoign/.pyenv/versions/3.5.2/envs/rta/lib/python3.5/site-packages/sqlalchemy/orm/state.py", line 411, in _initialize_instance
    return manager.original_init(*mixed[1:], **kwargs)
  File "/Users/olemoign/.pyenv/versions/3.5.2/envs/rta/lib/python3.5/site-packages/sqlalchemy/ext/declarative/base.py", line 653, in _declarative_constructor
    (k, cls_.__name__))
TypeError: 'key' is an invalid keyword argument for FactorsTOTP
```

This isn't so bad, as I can do 

factors_totp = models.FactorsTOTP()
factors_totp.key = totp_key

but I just wanted to raise the issue. Thank you for your great library.

[Sqlalchemy-tickets] Issue #3964: sqltypes.ARRAY does not link to postgresql.ARRAY (?!?!) (zzzeek/sqlalchemy)

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

New issue 3964: sqltypes.ARRAY does not link to postgresql.ARRAY (?!?!)
https://bitbucket.org/zzzeek/sqlalchemy/issues/3964/sqltypesarray-does-not-link-to

Michael Bayer:

the bind/ result processors of postgresql.ARRAY are not invoked with the sqltypes.ARRAY.   I'm not sure how the ARRAY enhancement got committed with an omission like this.

[Sqlalchemy-tickets] Issue #3963: aliased vs. non-aliased specificity in loader options on lazy load (zzzeek/sqlalchemy)

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

New issue 3963: aliased vs. non-aliased specificity in loader options on lazy load
https://bitbucket.org/zzzeek/sqlalchemy/issues/3963/aliased-vs-non-aliased-specificity-in

Michael Bayer:

all of these cases emit eager load for B.cs when they are specified as lazyload:


```
#!python

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

Base = declarative_base()


class A(Base):
    __tablename__ = 'a'
    id = Column(Integer, primary_key=True)
    bs = relationship("B")

class B(Base):
    __tablename__ = 'b'
    id = Column(Integer, primary_key=True)
    a_id = Column(ForeignKey('a.id'))
    cs = relationship("C")

class C(Base):
    __tablename__ = 'c'
    id = Column(Integer, primary_key=True)
    b_id = Column(ForeignKey('b.id'))

e = create_engine("sqlite://", echo=True)
Base.metadata.create_all(e)

s = Session(e)
s.add(A(bs=[B(cs=[C()])]))
s.add(A())
s.commit()
s.close()

aa = aliased(A)
q = s.query(aa, A).filter(aa.id == 1).filter(A.id == 2).options(joinedload(A.bs).joinedload(B.cs))
for a, real_a in q:
    for b in a.bs:
        # was eager loaded, when should lazy load
        for c in b.cs:
            print c


s.close()

print "--------------------------------"

aa = aliased(A)
q = s.query(aa, A).filter(aa.id == 1).filter(A.id == 2).options(defaultload(A.bs).joinedload(B.cs))
for a, real_a in q:
    for b in a.bs:
        # was eager loaded, when should lazy load
        for c in b.cs:
            print c


s.close()

print "--------------------------------"

aa = aliased(A)
q = s.query(A, aa).filter(A.id == 1).filter(aa.id == 2).options(defaultload(aa.bs).joinedload(B.cs))
for a, real_a in q:
    for b in a.bs:
        # was eager loaded, when should lazy load
        for c in b.cs:
            print c
```


_chop_path is important here, this fixes:

```
#!diff


diff --git a/lib/sqlalchemy/orm/strategy_options.py b/lib/sqlalchemy/orm/strategy_options.py
index bae2b73..60ed2be 100644
--- a/lib/sqlalchemy/orm/strategy_options.py
+++ b/lib/sqlalchemy/orm/strategy_options.py
@@ -376,7 +376,8 @@ class _UnboundLoad(Load):
                         _WILDCARD_TOKEN,) and c_token != p_prop.key:
                     return None
             elif isinstance(c_token, PropComparator):
-                if c_token.property is not p_prop:
+                if c_token.property is not p_prop or \
+                        c_token._parententity is not p_mapper:
                     return None
         else:
             i += 1

```

[Sqlalchemy-tickets] Issue #3962: Unsupported SQLAlchemy of mssql instance (zzzeek/sqlalchemy)

[Ramin <iss...@bi...>]

New issue 3962: Unsupported SQLAlchemy of mssql instance
https://bitbucket.org/zzzeek/sqlalchemy/issues/3962/unsupported-sqlalchemy-of-mssql-instance

Ramin:

Hi,

i going to connect of SQLAlchemy to other mssql instance exm: `<ip>\SQLInstance` but i see your [docs](http://docs.sqlalchemy.org/en/latest/dialects/mssql.html#module-sqlalchemy.dialects.mssql.pymssql) SQAlchemy default connect to `.`  instance, 

default instance code :
```python
from sqlalchemy import create_engine
engine = create_engine('mssql+pymssql://<user>:<pass>@<ip>:1433/database_name')

from sqlalchemy.ext.declarative import declarative_base
Base = declarative_base()
from sqlalchemy import Column, Integer, String

class teeetttt(Base):
     __tablename__ = 'testtetetet'
     id = Column(Integer, primary_key=True)
     name = Column(String)
     fullname = Column(String)
     password = Column(String)

     

print teeetttt.__table__
ed_user = teeetttt(name='ed', fullname='Ed Jones', password='edspassword')
from sqlalchemy.orm import sessionmaker
Session = sessionmaker(bind=engine)
Session = sessionmaker()
Session.configure(bind=engine)
session = Session()
session.add(ed_user)
session.commit()
print ed_user.name
session.query(teeetttt).all()
f  = session.query(teeetttt).all()
print f[0].name
print session.query(teeetttt).all()

```

my goals connect to ```engine=create_engine('mssql+adodbapi://sa:a`123456@172.16.20.181:1433\SQLInstance/DB')``` that `SQLInstance` this is my instance and DB is database name

Responsible: zzzeek

[Sqlalchemy-tickets] Issue #3961: mysql partition options need to be after all the table options (zzzeek/sqlalchemy)

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

New issue 3961: mysql partition options need to be after all the table options
https://bitbucket.org/zzzeek/sqlalchemy/issues/3961/mysql-partition-options-need-to-be-after

Michael Bayer:

e.g. https://dev.mysql.com/doc/refman/5.7/en/create-table.html

[Sqlalchemy-tickets] Issue #3960: Declarative backref not filtering by polymorphic identity (zzzeek/sqlalchemy)

[Mischa <iss...@bi...>]

New issue 3960: Declarative backref not filtering by polymorphic identity
https://bitbucket.org/zzzeek/sqlalchemy/issues/3960/declarative-backref-not-filtering-by

Mischa:

I am trying to declare a backref relationship between a related table and a single-table polymorphic subclass. I expect the relationship to filter by the polymorphic discriminator column when I select the relationship, but instead it selects ALL related rows, regardless of polymorphic discriminator identity.
This is confusing, and does not appear to be documented in http://docs.sqlalchemy.org/en/latest/orm/extensions/declarative/inheritance.html#declarative-inheritance
I think this is a common use case, but I see very little documentation in the declarative inheritance relationship documentation. Please let me know if I'm missing something.
Attached is a test case that fails to discriminate when selecting the related rows via backref. SQL logs are at: https://0bin.net/paste/P2VTuIUsIDyZ9vw4#zItNB6zFxnNm1usjqzqflA4eJ+JqGELuUq8uIm2pf2i

[Sqlalchemy-tickets] Issue #3959: postgresql interval restricted types (zzzeek/sqlalchemy)

[lv <iss...@bi...>]

New issue 3959: postgresql interval restricted types
https://bitbucket.org/zzzeek/sqlalchemy/issues/3959/postgresql-interval-restricted-types

lv:

as [mentioned Ilja Everilä](http://stackoverflow.com/a/43231662)
according to [this code](https://github.com/zzzeek/sqlalchemy/blob/master/lib/sqlalchemy/dialects/postgresql/base.py#L1303),

there is only
'interval': INTERVAL,
'interval year to month': INTERVAL,
'interval day to second': INTERVAL,
and no [others](https://www.postgresql.org/docs/9.6/static/datatype-datetime.html)
YEAR
MONTH
DAY
HOUR
MINUTE
SECOND
YEAR TO MONTH
DAY TO HOUR
DAY TO MINUTE
DAY TO SECOND
HOUR TO MINUTE
HOUR TO SECOND
MINUTE TO SECOND

types of interval ?

found [460 ticket](https://bitbucket.org/zzzeek/sqlalchemy/issues/460/postgresql-interval-type) but there only two additional types 'interval year to month', 'interval day to second'

[Sqlalchemy-tickets] Issue #3958: Sqlalchemy 1.1 doesn't remove duplicate rows in some cases while 1.0 does (zzzeek/sqlalchemy)

[anti-social <iss...@bi...>]

New issue 3958: Sqlalchemy 1.1 doesn't remove duplicate rows in some cases while 1.0 does
https://bitbucket.org/zzzeek/sqlalchemy/issues/3958/sqlalchemy-11-doesnt-remove-duplicate-rows

anti-social:

After the [commit](https://bitbucket.org/zzzeek/sqlalchemy/commits/ceeb033054f09db3eccbde3fad1941ec42919a54) next test case begun to fail:


```
#!python

class FilterUniqueJoinTest(fixtures.MappedTest, AssertsCompiledSQL):
    run_setup_mappers = 'once'

    @classmethod
    def define_tables(cls, metadata):
        Table('users', metadata,
              Column('id', Integer, primary_key=True,
                     test_needs_autoincrement=True),
              Column('name', String(50)),
              Column('date_created', DateTime, default=datetime.datetime.now))

        Table('orders', metadata,
              Column('id', Integer, primary_key=True,
                     test_needs_autoincrement=True),
              Column('user_id', Integer, ForeignKey('users.id')))

    @classmethod
    def setup_classes(cls):
        users = cls.tables.users
        orders = cls.tables.orders

        class User(cls.Comparable):
            pass

        class Order(cls.Comparable):
            pass

        mapper(User, users, properties={
            'orders': relationship(Order, order_by=users.c.id.desc())
        })

        mapper(Order, orders)

    @classmethod
    def insert_data(cls):
        User = cls.classes.User
        Order = cls.classes.Order

        sess = create_session()
        o1 = Order()
        o2 = Order()
        o3 = Order()

        u1 = User(name='Mike')
        u1.orders = [o1, o2]
        u2 = User(name='John')
        u2.orders = [o3]

        sess.add(u1)
        sess.add(u2)
        sess.flush()
        sess.close()

    def test_filter_unique_column(self):
        User = self.classes.User

        sess = create_session()

        users = (
            sess.query(
                User.date_created.label('dt'),
                User,
            )
            .options(joinedload(User.orders))
            .all()
        )
        assert len(users) == 2

    def test_filter_unique_func_column(self):
        User = self.classes.User

        sess = create_session()

        users = (
            sess.query(
                func.date(User.date_created, 'start of month').label('month'),
                User,
            )
            .options(joinedload(User.orders))
            .all()
        )
        assert len(users) == 2

```

Sqlalchemy 1.0:

```
#!text

test/orm/test_joins.py::FilterUniqueJoinTest::test_filter_unique_column PASSED
test/orm/test_joins.py::FilterUniqueJoinTest::test_filter_unique_func_column PASSED
```

Sqlalchemy 1.1:

```
#!text
test/orm/test_joins.py::FilterUniqueJoinTest::test_filter_unique_column PASSED
test/orm/test_joins.py::FilterUniqueJoinTest::test_filter_unique_func_column FAILED

```

Is it expected behaviour?

[Sqlalchemy-tickets] Issue #3957: incorrect rendering of sqltext for column-level check constraint (zzzeek/sqlalchemy)

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

New issue 3957: incorrect rendering of sqltext for column-level check constraint
https://bitbucket.org/zzzeek/sqlalchemy/issues/3957/incorrect-rendering-of-sqltext-for-column

Michael Bayer:

```
#!python

from sqlalchemy import *
from sqlalchemy.sql.expression import ColumnClause

class CodenameConstraint(CheckConstraint):
    """Check that the column uses a limited alphabet."""
    def __init__(self, column_name):
        column = ColumnClause(column_name)
        super(CodenameConstraint, self).__init__(
            column.op("~")("^[A-Za-z0-9_-]+$"))


m = MetaData()
t = Table(
    't', m, Column('x', String, CodenameConstraint('x')),

)

from sqlalchemy.schema import CreateTable

print CreateTable(t)

```


a string compiler is being invoked separately, needs this:


```
#!diff

diff --git a/lib/sqlalchemy/sql/compiler.py b/lib/sqlalchemy/sql/compiler.py
index 4415028..b18f903 100644
--- a/lib/sqlalchemy/sql/compiler.py
+++ b/lib/sqlalchemy/sql/compiler.py
@@ -2602,7 +2602,9 @@ class DDLCompiler(Compiled):
             formatted_name = self.preparer.format_constraint(constraint)
             if formatted_name is not None:
                 text += "CONSTRAINT %s " % formatted_name
-        text += "CHECK (%s)" % constraint.sqltext
+        text += "CHECK (%s)" % self.sql_compiler.process(constraint.sqltext,
+                                                         include_table=False,
+                                                         literal_binds=True)
         text += self.define_constraint_deferrability(constraint)
         return text
 

```

been this way for years so 1.2 to allow for any surprises.

[Sqlalchemy-tickets] Issue #3956: Mutable + column properties broken (zzzeek/sqlalchemy)

[Marcin K. <iss...@bi...>]

New issue 3956: Mutable + column properties broken
https://bitbucket.org/zzzeek/sqlalchemy/issues/3956/mutable-column-properties-broken

Marcin Kurczewski:

Trying to use both mutable and column properties within the same table results in mapper exception. 

The stacktrace is:


```
#!text
Traceback (most recent call last):
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/sql/elements.py", line 682, in __getattr__
    return getattr(self.comparator, key)
AttributeError: 'Comparator' object has no attribute 'info'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/srv/www/tmp.sakuya.pl/f/test", line 25, in <module>
    session.query(User).all()
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/orm/session.py", line 1328, in query
    return self._query_cls(entities, self, **kwargs)
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/orm/query.py", line 139, in __init__
    self._set_entities(entities)
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/orm/query.py", line 150, in _set_entities
    self._set_entity_selectables(self._entities)
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/orm/query.py", line 180, in _set_entity_selectables
    ent.setup_entity(*d[entity])
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/orm/query.py", line 3595, in setup_entity
    self._with_polymorphic = ext_info.with_polymorphic_mappers
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/util/langhelpers.py", line 764, in __get__
    obj.__dict__[self.__name__] = result = self.fget(obj)
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/orm/mapper.py", line 1948, in _with_polymorphic_mappers
    configure_mappers()
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/orm/mapper.py", line 2869, in configure_mappers
    mapper, mapper.class_)
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/event/attr.py", line 218, in __call__
    fn(*args, **kw)
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/orm/events.py", line 616, in wrap
    fn(*arg, **kw)
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/ext/mutable.py", line 603, in listen_for_type
    prop.columns[0].info.get('_ext_mutable_orig_type')
  File "/home/rr-/.local/lib/python3.6/site-packages/sqlalchemy/sql/elements.py", line 688, in __getattr__
    key)
AttributeError: Neither 'Label' object nor 'Comparator' object has an attribute 'info'
```


The code works fine on 1.1.7 and doesn't on 1.1.8. The GitHub hash of the change that introduces this behavior is 07b63894cb8ff9529b406f196b5d7cb9af209e9e: "Track SchemaEventTarget types in as_mutable()".

I attach a code sample that triggers this behavior.

[Sqlalchemy-tickets] Issue #3955: connectionless execution + autocommit + ON CONFLICT DO NOTHING + no rows == boom (zzzeek/sqlalchemy)

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

New issue 3955: connectionless execution + autocommit + ON CONFLICT DO NOTHING + no rows == boom
https://bitbucket.org/zzzeek/sqlalchemy/issues/3955/connectionless-execution-autocommit-on

Michael Bayer:

```
#!diff

@@ -96,6 +97,33 @@ class OnConflictTest(fixtures.TablesTest):
                 [(1, 'name1')]
             )
 
+    def test_on_conflict_do_nothing_connectionless(self):
+        users = self.tables.users_xtra
+
+        with testing.db.connect() as conn:
+            result = conn.execute(
+                insert(users).on_conflict_do_nothing(
+                    constraint='uq_login_email'),
+
+                dict(name='name1', login_email='email1')
+            )
+            eq_(result.inserted_primary_key, [1])
+            eq_(result.returned_defaults, (1,))
+
+        result = testing.db.execute(
+            insert(users).on_conflict_do_nothing(
+                constraint='uq_login_email'
+            ),
+            dict(name='name2', login_email='email1')
+        )
+        eq_(result.inserted_primary_key, None)
+        eq_(result.returned_defaults, None)
+
+        eq_(
+            testing.db.execute(users.select().where(users.c.id == 1)).fetchall(),
+            [(1, 'name1', 'email1', None)]
+        )
+
     @testing.provide_metadata
     def test_on_conflict_do_nothing_target(self):
         users = self.tables.users

```

the lack of row returned sends the connection into autoclose via result.fetchone().  then autocommit causes an exception, then it continues to be confused because the connection has been closed.

[Sqlalchemy-tickets] Issue #3954: mainstream the "baked" lazyloader (zzzeek/sqlalchemy)

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

New issue 3954: mainstream the "baked" lazyloader
https://bitbucket.org/zzzeek/sqlalchemy/issues/3954/mainstream-the-baked-lazyloader

Michael Bayer:

looking to use "baked" for the two "IN" loaders, this should be mainstreamed in LazyLoader as well.   no issues have been reported over the past year so it's time to see if this really holds up.

[Sqlalchemy-tickets] Issue #3953: expanding parameter (zzzeek/sqlalchemy)

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

New issue 3953: expanding parameter
https://bitbucket.org/zzzeek/sqlalchemy/issues/3953/expanding-parameter

Michael Bayer:

this is for the purpose of IN() in conjunction with query caching.    Idea is to allow this to be viable with "select in" loading and "select in" inheritance so that these are compatible w/ "baked".

    stmt = select([table]).where(table.c.col.in_(bindparam('foo', expanding=True))
    conn.execute(stmt, {"foo": [1, 2, 3]})

the string statement is converted at ExecutionContext time with an extra search and replace.  this so that all the caching steps can be used against the above query w/ a dynamic list for IN.

[Sqlalchemy-tickets] Issue #3952: with_variant with dialect type and parameters doesn't work when query. (zzzeek/sqlalchemy)

[Isaku Y. <iss...@bi...>]

New issue 3952: with_variant with dialect type and parameters doesn't work when query.
https://bitbucket.org/zzzeek/sqlalchemy/issues/3952/with_variant-with-dialect-type-and

Isaku Yamahata:

The following code snipet doesn't work as expected from sqlalchemy 1.1.5
The cultprit is the change set of df9b6492e5ca47e26d539d2283fa816a2d5c8ad6
change-id of I7b7b45d604a4ae8d1dc236a5a1248695aab5232e
https://bitbucket.org/zzzeek/sqlalchemy/commits/df9b6492e5ca


import sqlalchemy as sa
from sqlalchemy.dialects import sqlite

class Data(Base):
   created_at = sa.Column(
        sa.DateTime().with_variant(
            sqlite.DATETIME(truncate_microseconds=True), 'sqlite'),
        server_default=sa.func.now())

# insert row and query based on created_at doesn't work as expected.
row = Data()
session.add(row)
session.flush()

rows = sessio.query(data.created_at).all()
row = rows[0]

# query by created_at
found = sessoin.squery(Data).filer(Data.created_at == row.created_at).all()

found should include added row. but after the change set pointed above, found is empty list.


by setting engine.echo=True, the following was gotten.
SELECT Data.created_at AS Data_created_at
WHERE Data.created_at = ?
2017-03-31 17:42:51,951 INFO sqlalchemy.engine.base.Engine.sqlite@vgluzzatoz ('2017-04-01 00:42:45.000000',)
('2017-04-01 00:42:45.000000',)

please notice the it includes microseconds as ".000000" even with truncate_microseconds=True

sqlite.DATETIME(truncate_micorseconds=True) should be used when bind_processor() is called, but plain DateTime is used. This is because of type_api.Variant.coerce_compared_value ignores dialects.

[Sqlalchemy-tickets] Issue #3951: PostgreSQL Sequence not created by metadata.create_all() from declarative mixin (zzzeek/sqlalchemy)

[Andrew P. R. <iss...@bi...>]

New issue 3951: PostgreSQL Sequence not created by metadata.create_all() from declarative mixin
https://bitbucket.org/zzzeek/sqlalchemy/issues/3951/postgresql-sequence-not-created-by

Andrew Paulo Robillo:

I'd been copying this [snippet](https://bitbucket.org/snippets/shimofuri/bALAz) from my old cookbook to add a surrogate key to a declarative table class:


```
#!python

class MixinIntKey(object):
    @declared_attr
    def key(cls):
        return Column(
            pg.INTEGER,
            Sequence(f"sqk_{cls.__tablename__}", 1, 1),
            nullable=False,
            primary_key=True
            )
```
This code was working before (some 3-4 years ago, I think around version 0.8 ): it generates the Sequence when I build the database objects through metadata.create_all(). The only change I made to the snippet was to update it to use format strings (Python 3.6).

After some googling around and testing (see attached file/[link](https://bitbucket.org/snippets/shimofuri/bALAz)), the Sequence can be generated only if the metadata is passed during Sequence init:

```
#!python
Sequence(f"sqk_{cls.__tablename__}", 1, 1, metadata=metadata)
```

My environment: psycopg2==2.7.1, SQLAlchemy==1.1.8, Python 3.6 64bit on Windows 8.1 64bit, PostgreSQL 9.6 server

[Sqlalchemy-tickets] Issue #3950: 1.1 regression due to TypeDecorator copy() from Column.copy(), interferes with mutable (zzzeek/sqlalchemy)

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

New issue 3950: 1.1 regression due to TypeDecorator copy() from Column.copy(), interferes with mutable
https://bitbucket.org/zzzeek/sqlalchemy/issues/3950/11-regression-due-to-typedecorator-copy

Michael Bayer:

this is a bug that exists since sqlalchemy.ext.mutable started, however it only is apparent when the type object is copied.  The copy is uncommon except in the case of SchemaEventTarget, which TypeDecorator becomes in 1.1.

```
#!python

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

Base = declarative_base()


class MyType(TypeDecorator):
    impl = Text

    def process_bind_param(self, value, dialect):
        if value is not None:
            value = json.dumps(value)
        return value

    def process_result_value(self, value, dialect):
        if value is not None:
            value = json.loads(value)
        return value

data = Column(mutable.MutableDict.as_mutable(MyType))


class A(Base):
    __tablename__ = 'a'
    id = Column(Integer, primary_key=True)

    # works fine
    # data = data

    # fails in 1.1 because TypeDecorator is a SchemaType,
    # so Column copies the type, so the event we attached
    # to scan mapper columns can't identify that this is the "data"
    # we copied
    data = data.copy()  # as happens in declarative

e = create_engine("sqlite://", echo=True)
Base.metadata.create_all(e)

s = Session(e)

a1 = A(data={"foo": "bar"})
s.add(a1)
s.commit()

a1.data["foo"] = "bat"
s.commit()

assert a1.data["foo"] == "bat"

```

[Sqlalchemy-tickets] Issue #3949: sqlalchemy compiler.py bug (zzzeek/sqlalchemy)

[Jorge S. <iss...@bi...>]

New issue 3949: sqlalchemy compiler.py  bug
https://bitbucket.org/zzzeek/sqlalchemy/issues/3949/sqlalchemy-compilerpy-bug

Jorge Sierra:

conn_str = string.Template('db2+ibm_db://$user:$password@localhost:50000/Sample').substitute(user = user , password = password)

for t in metadata.sorted_tables:
   logger.info ("CreateTable %s %s" % (t.name, **CreateTable**(t).compile(db2)))

When trying to run **CreateTable**

C:\Python27\Lib\site-packages\sqlalchemy\sql\**compiler.py**

around line 2588
under function 

def **visit_foreign_key_constraint**(self, constraint):

this function is assuming **f.parent** is never  'None' and in my case it was, same as  **f.column**

This is the fix below


```
#!python


list_text1 = [] 
for f in constraint.elements:
 if f.parent is not None:
   list_text1.append(preparer.quote(f.parent.name))
   
list_text2 = [] 
for f in constraint.elements:
 if hasattr(f, 'column'):
   if f.column is not None:
     list_text2.append(preparer.quote(f.column.name))           
     
text += "FOREIGN KEY(%s) REFERENCES %s (%s)" % (
      ', '.join(list_text1),
      self.define_constraint_remote_table(
          constraint, remote_table, preparer),
      ', '.join(list_text2))
text += self.define_constraint_match(constraint)
text += self.define_constraint_cascades(constraint)
text += self.define_constraint_deferrability(constraint)
return text

```

[Sqlalchemy-tickets] Issue #3948: post-load of joined-inh subtables (zzzeek/sqlalchemy)

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

New issue 3948: post-load of joined-inh subtables
https://bitbucket.org/zzzeek/sqlalchemy/issues/3948/post-load-of-joined-inh-subtables

Michael Bayer:

using the post_load architecture of #3944, add `post_with_polymorphic` argument to mapper.    Subclasses will be loaded using a query like query(Subclass).options(defer(everything on superclass), noload(everything on superclass)), as opposed to a custom SELECT statement the way the lazy load works, so that relationship loaders and other loader options on the subclass take effect normally.

[Sqlalchemy-tickets] Issue #3947: race condition in joined eager load perf enhancement (zzzeek/sqlalchemy)

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

New issue 3947: race condition in joined eager load perf enhancement
https://bitbucket.org/zzzeek/sqlalchemy/issues/3947/race-condition-in-joined-eager-load-perf

Michael Bayer:

e.g. #3915

test:


```
#!python

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

Base = declarative_base()


class A(Base):
    __tablename__ = 'a'
    id = Column(Integer, primary_key=True)
    bs = relationship("B")

class B(Base):
    __tablename__ = 'b'
    id = Column(Integer, primary_key=True)
    a_id = Column(ForeignKey('a.id'))

e = create_engine("sqlite:///foo.db", echo=False)
Base.metadata.create_all(e)


import threading
import random
import time
import traceback

def work(e):
    s = Session(e)
    while True:
        try:
            s.query(A).options(joinedload(A.bs)).all()
        except Exception as e:
            traceback.print_exc()
            print "Exception: %s" % e
        finally:
            time.sleep(random.random())

def chaos():
    while True:
        time.sleep(.001)
        A.bs.property._get_strategy((("lazy", "joined"),))._aliased_class_pool[:] = []

s = Session(e)
s.add_all([A(bs=[B(), B(), B()])])
s.commit()


threads = [threading.Thread(target=work, args=(e, )) for i in range(20)]
chaos = threading.Thread(target=chaos)

for t in threads:
    t.start()
chaos.start()

```


traceback


```
#!

Traceback (most recent call last):
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/query.py", line 2679, in all
  File "test.py", line 31, in work
    s.query(A).options(joinedload(A.bs)).all()
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/query.py", line 2679, in all
yes
    return list(self)
    return list(self)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/loading.py", line 90, in instances
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/query.py", line 2831, in __iter__
    return self._execute_and_instances(context)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/query.py", line 2854, in _execute_and_instances
    result = conn.execute(querycontext.statement, self._params)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/engine/base.py", line 945, in execute
    util.raise_from_cause(err)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/util/compat.py", line 203, in raise_from_cause
    reraise(type(exception), exception, tb=exc_tb, cause=cause)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/loading.py", line 57, in instances
    for query_entity in query._entities
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/query.py", line 3686, in row_processor
    polymorphic_discriminator=self._polymorphic_discriminator
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/loading.py", line 330, in _instance_processor
    context, path, mapper, result, adapter, populators)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/interfaces.py", line 532, in create_row_processor
    return meth(self, multiparams, params)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/sql/elements.py", line 269, in _execute_on_connection
    mapper, result, adapter, populators)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/strategies.py", line 1594, in create_row_processor
    eager_adapter)
    return connection._execute_clauseelement(self, multiparams, params)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/loading.py", line 318, in _instance_processor
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/engine/base.py", line 1053, in _execute_clauseelement
    col = adapter.columns[col]
    compiled_sql, distilled_params
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/sql/util.py", line 713, in __getitem__
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/engine/base.py", line 1189, in _execute_context
    context)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/engine/base.py", line 1393, in _handle_dbapi_exception
    exc_info
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/util/compat.py", line 203, in raise_from_cause
    reraise(type(exception), exception, tb=exc_tb, cause=cause)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/engine/base.py", line 1182, in _execute_context
    context)
    self.parent.include_fn and not self.parent.include_fn(key)
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/orm/util.py", line 326, in _include_fn
  File "/home/classic/dev/sqlalchemy/lib/sqlalchemy/engine/default.py", line 504, in do_execute
    entity = elem._annotations.get('parentmapper', None)
AttributeError: 'NoneType' object has no attribute '_annotations'
Exception: 'NoneType' object has no attribute '_annotations'
    cursor.execute(statement, parameters)

```

[Sqlalchemy-tickets] Issue #3946: ROLLBACK outside of transactions (zzzeek/sqlalchemy)

[Daniel E. <iss...@bi...>]

New issue 3946: ROLLBACK outside of transactions
https://bitbucket.org/zzzeek/sqlalchemy/issues/3946/rollback-outside-of-transactions

Daniel Eisner:

Hi,

If I try to run a bad "SELECT" statement without starting a transaction, SQLAlchemy tries to roll back the transaction, which fails, and then raises a ProgrammingError execption, obscuring the original exception.

I think the problem is with this bit of code in sqlalchemy.engine.base:

    def _autorollback(self):
        if not self._root.in_transaction():
            self._root._rollback_impl()


That logic looks backwards to me. Was this the original intent? Below is an example stack trace from a case like this.



---------------------------------
Traceback:



/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/pandas/io/sql.pyc in read_sql(sql, con, index_col, coerce_float, params, parse_dates, columns, chunksize)
    492             sql, index_col=index_col, params=params,
    493             coerce_float=coerce_float, parse_dates=parse_dates,
--> 494             chunksize=chunksize)
    495 
    496 

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/pandas/io/sql.pyc in read_query(self, sql, index_col, coerce_float, parse_dates, params, chunksize)
   1126         args = _convert_params(sql, params)
   1127 
-> 1128         result = self.execute(*args)
   1129         columns = result.keys()
   1130 

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/pandas/io/sql.pyc in execute(self, *args, **kwargs)
   1020     def execute(self, *args, **kwargs):
   1021         """Simple passthrough to SQLAlchemy engine"""
-> 1022         return self.engine.execute(*args, **kwargs)
   1023 
   1024     def read_table(self, table_name, index_col=None, coerce_float=True,

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in execute(self, statement, *multiparams, **params)
   2053 
   2054         connection = self.contextual_connect(close_with_result=True)
-> 2055         return connection.execute(statement, *multiparams, **params)
   2056 
   2057     def scalar(self, statement, *multiparams, **params):

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in execute(self, object, *multiparams, **params)
    937         """
    938         if isinstance(object, util.string_types[0]):
--> 939             return self._execute_text(object, multiparams, params)
    940         try:
    941             meth = object._execute_on_connection

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in _execute_text(self, statement, multiparams, params)
   1095             statement,
   1096             parameters,
-> 1097             statement, parameters
   1098         )
   1099         if self._has_events or self.engine._has_events:

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in _execute_context(self, dialect, constructor, statement, parameters, *args)
   1187                 parameters,
   1188                 cursor,
-> 1189                 context)
   1190 
   1191         if self._has_events or self.engine._has_events:

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in _handle_dbapi_exception(self, e, statement, parameters, cursor, context)
   1384                 if cursor:
   1385                     self._safe_close_cursor(cursor)
-> 1386                 self._autorollback()
   1387 
   1388             if newraise:

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in _autorollback(self)
    822     def _autorollback(self):
    823         if not self._root.in_transaction():
--> 824             self._root._rollback_impl()
    825 
    826     def close(self):

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in _rollback_impl(self)
    701                 self.engine.dialect.do_rollback(self.connection)
    702             except BaseException as e:
--> 703                 self._handle_dbapi_exception(e, None, None, None, None)
    704             finally:
    705                 if not self.__invalid and \

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in _handle_dbapi_exception(self, e, statement, parameters, cursor, context)
   1313                                         self.dialect.dbapi.Error,
   1314                                         dialect=self.dialect),
-> 1315                 exc_info
   1316             )
   1317         self._reentrant_error = True

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/util/compat.pyc in raise_from_cause(exception, exc_info)
    201     exc_type, exc_value, exc_tb = exc_info
    202     cause = exc_value if exc_value is not exception else None
--> 203     reraise(type(exception), exception, tb=exc_tb, cause=cause)
    204 
    205 if py3k:

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/base.pyc in _rollback_impl(self)
    699                 self.engine.logger.info("ROLLBACK")
    700             try:
--> 701                 self.engine.dialect.do_rollback(self.connection)
    702             except BaseException as e:
    703                 self._handle_dbapi_exception(e, None, None, None, None)

/home/developer/.conda/envs/core_runtime/lib/python2.7/site-packages/sqlalchemy/engine/default.pyc in do_rollback(self, dbapi_connection)
    438 
    439     def do_rollback(self, dbapi_connection):
--> 440         dbapi_connection.rollback()
    441 
    442     def do_commit(self, dbapi_connection):

ProgrammingError: (pyodbc.ProgrammingError) ('25000', '[25000] [FreeTDS][SQL Server]The ROLLBACK TRANSACTION request has no corresponding BEGIN TRANSACTION. (3903) (SQLEndTran)')

[Sqlalchemy-tickets] Issue #3945: Imported classes confuses intersphinx (zzzeek/sqlalchemy)

[Charles Bouchard-L. <iss...@bi...>]

New issue 3945: Imported classes confuses intersphinx
https://bitbucket.org/zzzeek/sqlalchemy/issues/3945/imported-classes-confuses-intersphinx

Charles Bouchard-Légaré:

Sphinx (using [intersphinx](http://www.sphinx-doc.org/en/stable/ext/intersphinx.html)) can't find some classes from your objects.inv, resulting in 

See the [objects.inv](http://docs.sqlalchemy.org/en/latest/objects.inv)

```text
WARNING: py:class reference target not found: sqlalchemy.ext.declarative.api.Schema
```

This can be worked around by assigning the `__module__` property of classes imported in any `__init__.py`.  See also http://stackoverflow.com/a/24787344