Sorry if I resume this after two months but I think there's a bug in
cascade deletion of the relationship you suggested me:
mapper(Sensor, sensors,
properties={
'data': relationship(Data, backref='sensor',
foreign_keys=[data.c.id_meas],
primaryjoin=and_(sensors.c.id_meas==data.c.id_meas,
data.c.id_acq==acquisitions.c.id,
acquisitions.c.id_cu==sensors.c.id_cu),
cascade='all, delete-orphan', single_parent=True)
})
since, on a cascade delete of a Sensor sqlalchemy issues this query:
SELECT data.id AS data_id, data.id_acq AS data_id_acq, data.id_meas AS
data_id_meas, data.id_elab AS data_id_elab, data.value AS data_value
FROM data, acquisitions, sensors
WHERE ? = data.id_meas AND data.id_acq = acquisitions.id AND
acquisitions.id_cu = sensors.id_cu
(1,)
DELETE FROM data WHERE data.id = ?
that's going to delete all data with id_meas = 1 while it should be
SELECT data.id AS data_id, data.id_acq AS data_id_acq, data.id_meas AS
data_id_meas, data.id_elab AS data_id_elab, data.value AS data_value
FROM data, acquisitions, sensors
WHERE ? = data.id_meas AND ? = acquisitions.id_cu AND data.id_acq =
acquisitions.id AND acquisitions.id_cu = sensors.id_cu
(1, 3)
DELETE FROM data WHERE data.id = ?
with the `AND ? = acquisitions.id_cu` part added because Sensor has a
composite primary key (id_cu, id_meas).
I know it's a rare situation so I have no problems in removing cascade
and doing deletions on my own but I'd like to be sure it's not a fault
of mine but a bug.
Thanks for your support.
On Dec 30 2010, 5:45 pm, Michael Bayer <[email protected]>
wrote:
> this is again my error messages not telling the whole story, ill see if i can
> get the term "foreign_keys" back in there:
>
> mapper(Sensor, sensors,
> properties={
> 'data': relationship(Data, backref='sensor',
> foreign_keys=[data.c.id_meas],
> primaryjoin=and_(sensors.c.id_meas==data.c.id_meas,
> data.c.id_acq==acquisitions.c.id,
> acquisitions.c.id_cu==sensors.c.id_cu),
> cascade='all, delete-orphan', single_parent=True)
> })
>
> or
>
> mapper(Sensor, sensors,
> properties={
> 'data': relationship(Data, backref='sensor',
> foreign_keys=[sensors.id_meas],
> primaryjoin=and_(sensors.c.id_meas==data.c.id_meas,
> data.c.id_acq==acquisitions.c.id,
> acquisitions.c.id_cu==sensors.c.id_cu),
> cascade='all, delete-orphan', single_parent=True)
> })
>
> depending on if this is one-to-many or many-to-one. A relationship like
> this is really better off as a viewonly=True since populating it is not going
> to add rows to the "acquisitions" table.
>
> On Dec 30, 2010, at 10:15 AM,neurinowrote:
>
>
>
>
>
>
>
> > data = Table('data', metadata,
> > Column('id', Integer, primary_key=True),
> > Column('id_acq', Integer, ForeignKey('acquisitions.id'),
> > nullable=False),
> > Column('id_meas', Integer, nullable=False),
> > Column('value', Float, nullable=True),
> > )
>
> > acquisitions = Table('acquisitions', metadata,
> > Column('id', Integer, primary_key=True),
> > Column('id_cu', Integer, ForeignKey('ctrl_units.id'),
> > nullable=False),
> > Column('datetime', DateTime, nullable=False),
> > )
>
> > sensors = Table('sensors', metadata,
> > Column('id_cu', Integer, ForeignKey('ctrl_units.id'),
> > primary_key=True,
> > autoincrement=False),
> > Column('id_meas', Integer, primary_key=True, autoincrement=False),
> > Column('name', Unicode(20), nullable=False),
> > Column('desc', Unicode(40), nullable=False),
> > )
>
> > ctrl_units = Table('ctrl_units', metadata,
> > Column('id', Integer, primary_key=True, autoincrement=False),
> > Column('desc', Unicode(40), nullable=False)
> > )
>
> > and this mapping:
>
> > ...
> > orm.mapper(Sensor, sensors,
> > properties={
> > 'data': orm.relationship(Data, backref='sensor',
> > primaryjoin=and_(sensors.c.id_meas==data.c.id_meas,
> > data.c.id_acq==acquisitions.c.id,
> > acquisitions.c.id_cu==sensors.c.id_cu),
> > cascade='all, delete-orphan', single_parent=True)
> > })
> > ...
--
You received this message because you are subscribed to the Google Groups
"sqlalchemy" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sqlalchemy?hl=en.
