OK, a misunderstanding, the proof of concept is working as of a specific version (rf41aa3ad1da9) - if you've just checked out the branch then you're seeing a refactoring of code which is in flux, and that stack trace is specifically failing on new code that isn't actually doing anything other than being compared against what the usual relationship() mechanics produce given a particular input. It isn't yet implemented for the "local_remote_side" parameter in particular.
The entire system is not nearly ready for testing in any case, just wanted to report that a strategy has been devised that solves the issue - thanks ! On Feb 8, 2012, at 5:31 AM, Pau Tallada wrote: > Uhm, the test I attached on previous mail fails using this branch: > > $ python test_relationships.py > Traceback (most recent call last): > File "test_relationships (1).py", line 53, in <module> > c1 = Company() > File "<string>", line 2, in __init__ > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/instrumentation.py", > line 309, in _new_state_if_none > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/util/langhelpers.py", > line 485, in __get__ > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/instrumentation.py", > line 157, in _state_constructor > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/event.py", > line 274, in __call__ > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/mapper.py", > line 2325, in _event_on_first_init > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/mapper.py", > line 2250, in configure_mappers > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/mapper.py", > line 1166, in _post_configure_properties > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/interfaces.py", > line 128, in init > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/properties.py", > line 917, in do_init > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/properties.py", > line 945, in _create_new_thing > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/relationships.py", > line 58, in __init__ > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/relationships.py", > line 246, in _parse_joins > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/sql/visitors.py", > line 234, in cloned_traverse > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/sql/visitors.py", > line 227, in clone > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/sql/expression.py", > line 2998, in _copy_internals > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/sql/visitors.py", > line 230, in clone > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/relationships.py", > line 241, in visit_binary > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/relationships.py", > line 165, in _run_w_switch > File > "/home/tallada/Projectes/PIC/multivac/trunk/lib/sqlalchemy_1401.zip/sqlalchemy/orm/relationships.py", > line 222, in go > NotImplementedError > > > 2012/2/8 Pau Tallada <[email protected]> > Thank you very much! > > I'll test it ASAP :) > > > 2012/2/8 Michael Bayer <[email protected]> > I have good news on this front, in that I've nailed down how this will work, > including a patch that gets this basic thing working as a proof of concept. > However, the issue of being able to distinguish "remote" and "foreign" in a > binary expression where a column points to itself can benefit from moving > completely to the newer concept I'm working on, which is different enough > that I think it should be for the next major SQLAlchemy release (currently > it's called 0.8). I'll keep the ticket updated with progress reports. > > > On Feb 1, 2012, at 10:49 AM, Michael Bayer wrote: > >> This is essentially ticket #1401 and I've attached this there as well as >> moved up the priority, however this issue is extremely complicated and would >> require some serious rethinking of the relationship()'s inner workings. >> It would take several days to come up with a general solution so I can't >> give you a fix for this right now. >> >> http://www.sqlalchemy.org/trac/ticket/1401 >> >> >> >> >> On Feb 1, 2012, at 5:40 AM, Pau Tallada wrote: >> >>> Hi! >>> >>> I have a table with a self-reference of two columns that represents a tree >>> structure. >>> I was trying to build an outerjoin to select all the nodes have children >>> but NO grandchildren, but the SQL constructed was incorrect, as it was not >>> aliasing properly one of the columns. >>> >>> note_t Table('node_t', metadata, >>> Column('id', Integer, primary_key=True), >>> Column('project_id', Integer), >>> Column('parent_id', Integer), >>> sa.ForeignKeyConstraint( >>> ['project_id', 'parent_id'], >>> ['node_t.project_id', 'node_t.id])) >>> >>> mapper(Node, node_t, properties= { >>> 'children' : relationship(Node, >>> remote_side=[note_t.c.id, node_t.c.project_id] >>> ) >>> }) >>> >>> print str(session.query(Node).outerjoin(Node, Node.children, aliased=True))) >>> >>> Generated (simplified): >>> SELECT node.id, node.project_id, node.parent_id >>> FROM node >>> LEFT OUTER JOIN node AS parent ON node.parent_id = parent.id >>> AND node.project_id = node.project_id >>> >>> Expected: >>> SELECT node.id, node.project_id, node.parent_id >>> FROM node >>> LEFT OUTER JOIN node AS parent ON node.parent_id = parent.id >>> AND node.project_id = parent.project_id >>> >>> Making the join condition explicit generates the correct SQL >>> Parent = aliased(Node) >>> print str(session.query(Node).outerjoin(Parent, (Node.parent_id == >>> Parent.id) & (Node.project_id == Parent.project_id))) >>> >>> I have attached a small test file (test.py) that shows this behaviour and >>> is based on one of your tests suites (test_relationships). >>> >>> Thanks in advance! >>> >>> Pau. >>> -- >>> ---------------------------------- >>> Pau Tallada Crespí >>> Dep. d'Astrofísica i Cosmologia >>> Port d'Informació Científica (PIC) >>> Tel: +34 93 586 8233 >>> ---------------------------------- >>> >>> >>> >>> -- >>> 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. >>> <test.py> >> >> >> -- >> 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. > > > -- > 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. > > > > -- > ---------------------------------- > Pau Tallada Crespí > Dep. d'Astrofísica i Cosmologia > Port d'Informació Científica (PIC) > Tel: +34 93 586 8233 > ---------------------------------- > > > > > > -- > ---------------------------------- > Pau Tallada Crespí > Dep. d'Astrofísica i Cosmologia > Port d'Informació Científica (PIC) > Tel: +34 93 586 8233 > ---------------------------------- > > > > -- > 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. -- 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.
