Michael Bayer wrote: >> Does this changeset make it so that ActiveMapper doesn't have to >> jump through a bunch of hoops to make sure that tables/mappers are >> processed in a particular order for relationship dependencies? > > probably. it would be very helpful for you to get the latest trunk and > try it out. its a pretty major change to the mapper internals.
Downloaded and tested. It seems to not break anything if I don't make any changes to ActiveMapper. Also, it seems that I can remove about 20 lines of code from the 'process_relationships' function that were doing some checking for foreign keys and the existence of the related classes. I am still having to do a lot more checking than I would like because the 'relation' constructor takes in a class or mapper, and I have to wait for the class or mapper to *exist* before creating it. Any chance of being able to pass a class name there instead? Then I could remove this code entirely, and create relations as they are detected in the metaclass. This would *greatly* simplify ActiveMapper. At this point, it still eliminates a little chunk of code, and probably does a much better job of detecting strange orders :) -- Jonathan LaCour http://cleverdevil.org _______________________________________________ Sqlalchemy-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/sqlalchemy-users
