[storm] data separation

[Gustavo Niemeyer]

> Your problem is here: you need to have AutoReload by default, ie:

What Johan is describing is a problem in the schema rather than
something AutoReload was designed to solve.   He's printing a
reference rather than the id attribute, and he's printing it even
before the object is added to the database at all, which means that
loading the attribute from the database doesn't help.

The real problem in this case is that the proposed table organization
is problematic.  There are three different tables with
autoincrementing keys, and they should supposedly all share the same
sequence, but that's not being told or done anywhere.

Think about the following scenario: foo_private is flushed, then
foo_state is flushed, and now we have to set foo's primary key.
There's absolutely nothing in your schema enforcing foo_private and
foo_state to have the same key.

There are other problems there too, but this should be enough to pass
the main issue across.

-- 
Gustavo Niemeyer
http://niemeyer.net