That could be true for the 'parent' but I am not sure about the child in a one-to-many relationship. Think of the scenario(also the example in SQLObject).
A person object can have multiple address. The nice thing about SQLObject is that I can defined a MultiJoin so when I access person.addresses, it is a iterables. If I don't access person.addresses, it may not issue the select statement, but if I do, even just getting the id, the whole result set would be retrieved. I don't think SQLObject would use a seperate SQL if I just want the IDs. Lee McFadden wrote: > I'm pretty sure that the SQL is only executed as needed. The easiest > way to test this would be to duplicate your model.py as a standalone > and standard SQLObjects file, load it up in a python shell (probably > using exec(file)) and then using Role._connection.debug = True. That > will churn out every query that gets executed so you can see exactly > what's going on. > > I tried doing this in the tg-admin shell to no avail. I could set the > object's _connection.debug but I never got any debug info from > SQLObject. Is there any reason why this is? > > "Role > On 17/10/05, [EMAIL PROTECTED] <[EMAIL PROTECTED]> wrote: > > > > I think the question would be, is this "Role" instance a full > > representation of the underlying tuple(row). If it does, it means the > > data has been retrieved thus the performance concern. If it is lazily > > evaluated(only retrieved when accessed), that would be the solution for > > the initial question. > > > > gps wrote: > > > "Two attributes will generally be created, role, which returns a Role > > > instance, and roleID, which returns an integer ID for the related > > > role." - SQLObject 0.7 documentation when talking about what ForeignKey > > > does. > > > >
