I quite sure I'm not using it...
Look at that code that reproduce the bug
http://utilitybase.com/paste/11481
The last line (p2 = Parent(p)) leads to a load of the children, so
an autoflush. I have a failure with 0.5.2:
sqlalchemy.exc.IntegrityError: (IntegrityError) parent.dumb may not
be NULL
that's the cascade of p2 being added to the session upon being
associated with the in-session items in p.children. So yes, cascade
rules do add things to the session. setting cascade=None on
Child.parent would prevent it, or alternatively you can turn off
autoflush on the session for
On Jan 28, 10:45 am, GustaV buisson.guilla...@gmail.com wrote:
Hi all,
In the __init__ method of a mapper, the load of a relation may lead to
an autoflush operation. When it happens, the object being instanciated
is already in the session and so INSERTed in the flush, whereas it is
not
I'm using Declarative extension actually.
In the pylons framework, the session initialise like this
def init_model(engine):
Call me before using any of the tables or classes in the
model
sm = orm.sessionmaker(autoflush=True, autocommit=False,
bind=engine)
meta.engine = engine
On Jan 28, 2009, at 11:21 AM, GustaV wrote:
I'm using Declarative extension actually.
In the pylons framework, the session initialise like this
def init_model(engine):
Call me before using any of the tables or classes in the
model
sm = orm.sessionmaker(autoflush=True,
