Geert Bevin wrote:
the right way to do things, that'd be super. In particular I tried
and failed to make the Account object use the RIFE user ID as its
identifier, which I feel is the right data model since it's a
one-to-one mapping.
I think this should be possible by adding a 'sparse' constraint to
your identifier property. I haven't tried it out yet in this
situation, but that's what its intention is: allowing you to manage
the ID values yourself. I'll try it out once I get back from
Barcelona, next week.
That is only half the issue, unfortunately. The other half is that RIFE
user IDs are 64-bit integers, but as far as I can tell there's no way to
use a "long" field as an identifier with the GenericQueryManager. It
wants to use "int" IDs and complains if I don't. (I'm assuming this is a
well-known limitation, so I won't bother providing example errors, but I
can do so if you like.)
Actually, the lack of "long" IDs is kind of a bad thing in other
respects too; I've worked on apps in the past that have had to deal with
IDs greater than 2 billion, in some cases because there really were that
many rows, and in some because IDs weren't sequentially assigned for
various reasons, so we needed more than 32 bits even though we had fewer
than 2 billion rows.
Is there some way around this?
I also didn't see a more elegant way to do the many-to-one constraint
between Account and RoleUser, but it's entirely possible there is one
I'm not aware of. At the very least I imagine there's a more elegant
way to look up the RoleUser table name.
Best regards,
Geert
Oops, there was one more question there. :) Is there in fact an
official, clean way to look up the RoleUser table name?
-Steve
_______________________________________________
Rife-users mailing list
[email protected]
http://lists.uwyn.com/mailman/listinfo/rife-users
