the two below mentioned issues are fixed in rev 1629. however, while the extra column in your many-to-many table will no longer cause SA's ORM to get confused when deleting rows, you still probably want to use the "association object" pattern so that you can map an object attribute to that extra column.
On Jun 15, 2006, at 10:51 PM, Michael Bayer wrote: > if youre using an association table that has other state in it, you > should probably use the "association object" pattern: > > http://www.sqlalchemy.org/docs/ > datamapping.myt#datamapping_association > > however, ive just noticed the latest trunk (which I dont think youre > using) has an issue with setting the blank list, so im fixing that, > and also ill look and see why the default many-to-many operation > seems to track every column in the table instead of just the ones > that relate. > > On Jun 15, 2006, at 12:49 PM, Steve Zatz wrote: > >> I may be making a mistake but in the following situation in which >> there is a many-to-many relationship if the table containing the >> relationships has columns in addition to the Foreign Key columns then >> trying to delete relationships by doing: >> >> item.keywords = [] >> >> does not work. I don't know SQL well enough to know whether you >> should >> not have any other columns in the relationship table but this doesn't >> seem like a reasonable constraint. The delete is issued with the >> additional column values having values of None but if there is >> data in >> those columns, the delete does not take place. Perhaps this is an >> expected feature but it doesn't quite make sense to me. >> >> The example is below: >> >> --------------------------------------------------------------------- >> - >> ------------------------------------------------ >> >> >> item_table = Table('item',metadata, >> Column('id', Integer, primary_key=True), >> Column('name',String(150)), >> ) >> >> keyword_table = Table('keyword', metadata, >> Column('id', Integer, primary_key=True), >> Column('name', String(25), nullable=False), >> ) >> >> itemkeyword_table = Table('item_keyword', metadata, >> Column('item_id', Integer,ForeignKey >> ('item.id'), >> nullable=False), >> Column('keyword_id', Integer, >> ForeignKey('keyword.id'), nullable=False), >> Column('flag', Boolean, default=True), #<-- >> additional column >> ) >> >> >> class Item_a(object): >> def __init__(self, name): >> self.name = name >> >> >> class Keyword_a(object): >> def __init__(self, name): >> self.name = name >> >> mapper(Keyword, keyword_table, properties = dict(items = >> relation(Item, secondary=itemkeyword_table, lazy=False),)) >> mapper(Item, item_table, properties = dict(keywords = >> relation(Keyword, secondary=itemkeyword_table, lazy=False),)) >> >> item = Item("test) >> keyword = Keyword("key_for_test") >> item.keywords.append(keyword) >> item.keywords = [] >> #the above will not remove the corresponding rows in >> itemkeyword_table because >> #there is non-null data in some of the columns >> >> >> _______________________________________________ >> Sqlalchemy-users mailing list >> Sqlalchemy-users@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/sqlalchemy-users > > > > _______________________________________________ > Sqlalchemy-users mailing list > Sqlalchemy-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/sqlalchemy-users _______________________________________________ Sqlalchemy-users mailing list Sqlalchemy-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sqlalchemy-users
