there is... u do not want to know if A points B or B points A, u want to know if A and B are related in whatever aspect. That is, A and B are members of some set X denoting that aspect. i.e. moving the belonginess out of A and B alltogether. but this isn't going to make your DB simpler... quite the opposite.
On Friday 05 December 2008 10:40:16 Eric Ongerth wrote: > Thanks for the ideas. I thought of all of the above. The one I've > been using is the accessor which unions together the necessary > things. My question came up when I wondered if there was some even > more fundamental way to handle these forwards-backwards cases. I'm > glad to know I'm already doing all I can. > > On Dec 2, 3:32 pm, Michael Bayer <[EMAIL PROTECTED]> wrote: > > On Dec 2, 2008, at 5:54 PM, Eric Ongerth wrote: > > > Now when I want to find out whether a Foo has a relation to > > > another Foo, I have to check whether there exists any row in > > > foo_relations that has the given Foo as either as "this" OR > > > "that". Also, what if I need backrefs on the foo_relations > > > mapper? The backref from 'this' and the backref from 'that' > > > would both point to something called a foo, but they would have > > > to be given separate labels in order ot not be conflicting > > > property names -- when really, I would not want to know if a > > > foo was the 'that' or the 'this' of some foo relation. > > > > > > So ideally in a case like this, I could set an option that says > > > the m:m relation is bidirectional, and that the backrefs for > > > both foreign keys in the m:m table should really point to the > > > same place (or at least be unioned together). > > > > > > I have a feeling that would violate some part of the RDBMS > > > standards, and I'm perfectly willing to go without or work > > > around. This is more of a philosophical point for learning's > > > sake -- what do other people do in such cases? > > > > you can.... store two rows in the association table, one for each > > direction. or provide an accessor which just unions together > > the forwards and backwards references between Foo objects. or > > make a readonly relation() that does the appropriate "OR" logic. > > I might even try combining both of those techniques somehow. > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "sqlalchemy" group. To post to this group, send email to sqlalchemy@googlegroups.com 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 -~----------~----~----~----~------~----~------~--~---
