maybe somehow make it pseudo-composite using some "hardcoded" db-identifier as discriminator..
here Mike's reply from some days ago on a related question: > [sqlalchemy] Re: table without any primary_keys? > From: Michael Bayer <[EMAIL PROTECTED]> > To: sqlalchemy@googlegroups.com > Date: 2008-05-18 06:54 > > On May 17, 2008, at 9:52 PM, [EMAIL PROTECTED] wrote: > > seems such thing is disallowed, or at least Mapper complains. > > is it SQL requirement or what? > > the mapper needs some set of PK columns defined for mapped classes > so that it can identify objects. they dont need to be actual PK > columns in the database. > > in theory there just needs to be a function that can extract a > primary key from a row - it doesnt even necessarily have to be "use > these columns". though such a feature would require some API > changes. HTH svil On Thursday 22 May 2008 14:10:38 Geoff wrote: > Hi, > > I have split up my database horizontally and am using UUIDs to > uniquely identify a row across databases. Using UUIDs as a primary > key is slow (InnoDB) so I wanted to use the common trick of having > a INT primary key using auto_increment on each database. This is > all fine, until sqlalchemy checks its cache of objects after a > query to see if the object has already been retrieved. This breaks > because the primary key is not unique across databases when I use > an auto_incremented INT. > > I reckon the solution is going to have to be manually setting the > field used by sqlalchemy to make the cache decision. Is there any > way of doing this already, or am I going to have to put it in > myself? > > Thanks! > --~--~---------~--~----~------------~-------~--~----~ 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 -~----------~----~----~----~------~----~------~--~---