Using models for multiple one-to-many table relationships

Question

Hi there!
I have a database with tables that have one-to-many relationships. They
are laid out with a standard join tables:
A     A_B     B     B_C     C
a_id     a_id     b_id     b_id     c_id
b_id     c_id
and has 1 A for n of B. In some cases, the B table also joins with
another table for yet more information.
My question: what technique should I use to organize a database of this
type into JATO Models?
I could create an AModel, for example, that would have a set of query
field descriptors for the information relevant from tables A and B. But
then how do I query table C when that information is necessary? Or,
more to the point, create information in C when a new entry in B is created.
Do the models use each other, or is that left up to the view bean (seems
incorrect for data isolation).
Any pointers appreciated,
Byron

Answer 1

Byron,
You can create a separate "modifying" model for each table and do
inserts/deletes/updates in a transaction across the tables with the
three models. That's really the easy part.
The somewhat more tedious part is the "join" models, but not hard to do.
You just have to decide what you need from the different tables and
quite possibly create multiple join models for different select scenarios.
So let's just run through a simple two table join model.
Create a SQL model. I usually name them based on the table joins. Let's
use a somewhat more real example:
table = customer
columns = custid, firstname, lastname, email
table = address
columns = addrid (seqid), custid (fk), type
('w'ork,'h'ome,'b'illing,'s'hipping,'o'ther), address1, address2, city,
state, zip
Suppose we want to select a customer and show a list of all his/her
addresses.
I would create a SQL model class (extending QueryModelBase of course)
called CustomerAddressJoinModel (this naming convention gets tough with
three or more tables, so you have to be creative).
In that model, I would create the fields just like any single table
model. You need only include the columns you need. Maybe you create one
join model that only has custid, lastname, firstname, type, city, zip.
OK, so we only cut out 2 columns, but you get the point. If you had a
table with lots of columns with data you rarely need to display and/or
update, then you can exclude them. Probably a slick way to exclude
dynamically, but this is straight forward.
Another join model could have all the columns of both tables.
So the trick is to set the static where criteria (using the
setStaticWhereCriteria method) in the model's constructor. I think the
sql model template has this in it.
setStaticWhereCriteria(customer.custid=address.custid);
Now you just set your criteria like any other model, and execute it.
For a 3 table join, you just set the static criteria like this:
tableA.aid=tableB.aid AND tableB.bid=tableC.bid
does that make sense?
You can also set the "modifying table" (setModifyingTable(table) - i
think that's right) in the model and inserts/updates/deletes will only
operate on that table within the model. Problem is maybe in the the
field binding in your views. You may want to select from one field, and
modify using another field. Can be done though, just not directly supported.
In the next release of JATO, you will be able to set a display fields
binding for read and write using different fields and even different models.
craig
Byron Servies wrote:
Hi there!
I have a database with tables that have one-to-many relationships. They
are laid out with a standard join tables:
     A     A_B     B     B_C     C
     a_id     a_id     b_id     b_id     c_id
     b_id     c_id
and has 1 A for n of B. In some cases, the B table also joins with
another table for yet more information.
My question: what technique should I use to organize a database of this
type into JATO Models?
I could create an AModel, for example, that would have a set of query
field descriptors for the information relevant from tables A and B. But
then how do I query table C when that information is necessary? Or,
more to the point, create information in C when a new entry in B is created.
Do the models use each other, or is that left up to the view bean (seems
incorrect for data isolation).
Any pointers appreciated,
Byron
To download the latest version of JATO, please visit:
http://www.sun.com/software/download/developer/5102.html
For more information about JATO, please visit:
http://developer.iplanet.com/tech/appserver/framework/index.jsp