[sqlalchemy] Re: Specifying Alternate Join Conditions to relation()

2009-07-15 Thread Mike Conley
Ahh!! Starting to understand it. In this case, the entire expression to primaryjoin must be a string. The explanation about eval'ing the string when mappers are compiled makes sense. Now I see the relation in GHZ's example and in the code I will be writing should look something like this. boston_

[sqlalchemy] Re: Specifying Alternate Join Conditions to relation()

2009-07-15 Thread Michael Bayer
Mike Conley wrote: > I was getting ready to do something similar, and now I am confused. > > How is what GHZ is doing fundamentally different from the documentation > example "Specifying Alternate Join Conditions to relation()"? Doesn't that > example wrap the primaryjoin in the and_() construct a

[sqlalchemy] Re: Specifying Alternate Join Conditions to relation()

2009-07-15 Thread Mike Conley
I was getting ready to do something similar, and now I am confused. How is what GHZ is doing fundamentally different from the documentation example "Specifying Alternate Join Conditions to relation()"? Doesn't that example wrap the primaryjoin in the and_() construct also? http://www.sqlalchemy.o

[sqlalchemy] Re: Specifying Alternate Join Conditions to relation()

2009-07-15 Thread Michael Bayer
GHZ wrote: > > boston_addresses = relation('Address', >primaryjoin = and_ > ('User.user_id==Address.user_id', > > 'Address.city=="Boston"'), >foreign_keys = ['Address.user_id']) the expression sent to primaryjoin must be an expr