Hi Gautam,
Virgo Smart wrote:
Hello,
I tried out the problem again by making changes to the class-descriptors and schema
as per Armin's suggestions and the "Advanced O/R Mapping Technique" tutorial
(Apache OJB site). However the persistence broker still returns instances of Thing
class instead of the Book and Toy in the stuffInDrawer collection of the Drawer class.
If there are any aspects that I might have over-looked, please point them out. Note
that Thing is a concrete class in my case.
You need OJB 1.0.4 or the latest version from OJB_1_0_RELEASE branch.
In OJB test-suite you can find a test called
InheritanceMultipleTableTest#testInheritancedObjectsInCollectionReferences
which tests exactly your described scenario.
http://svn.apache.org/viewvc/db/ojb/branches/OJB_1_0_RELEASE/src/test/org/apache/ojb/broker/InheritanceMultipleTableTest.java?view=markup
This test pass with all databases I use for testing (hsql, mysql, maxDB,
OracleXE).
I am thinking about modifying the OJB source code to fix this extent-awareness problem
satisfactorily. However I would need some guidance in helping me understand how OJB
creates SQLs in context of repository.xml (descriptors) and how RowReader and
RsIterator is used in creating instances of concrete classes.
Could any please help me understand the concepts related to SQL building, RowReader and
RsIterator and how the current implementation behaves ? Does anyone have a
sequence-diagram / activity diagram to illustrate the logic involved ?
I am thinking on the lines of using the class-descriptor to look out of extent-class
definitions for the class whose data in being read and in the process of being
materialized. If I find an extent-class, I would then look in to the table mapped to
this extent class for row-entry (using PK or PK/FK as the search criteria) and then
decide which object instance to materialize.
I am not sure if this is the right way to go. Any advise or help on this is highly
appreciated.
I think there is no "extent-awareness problem" referring your
Drawer-test. In InheritanceMultipleTableTest you can find many tests
testing TPS-inheritance - only one specific scenario fail.
regards,
Armin
Thank you for all the support.
Regards,
Gautam.
On Wed, 22 Nov 2006 Virgo Smart wrote :
Hello Armin,
Apologies for posting a repeat of a similar problem as mentioned in an earlier
post. I had already posted this one before I recieved the reply to the earlier
one.
Looking at the modified class descriptor, I see that the extent-class and anonymous
fields are being dropped. I think this is ok. However I am not sure if the
"super" reference descriptor can point to the primary key field descriptor of
the parent class.
Wouldn't this mean that the PK in the BOOKS / TOYS table would also need to be
a FK pointing to the PK of the THINGS table ? Wouldn't this be illegal in SQL ?
Thanks and Regards,
Gautam
On Wed, 22 Nov 2006 Armin Waibel wrote :
Hi Gautam,
as said in my previous post in thread "OJB: Identity equals method impl." you should use the
"normal" TPS inheritance without a FK field and without "extent-class" declaration within
TPS (and don't override fields in the sub-classes):
<class-descriptor
class="Thing"
table="THINGS">
<field-descriptor
name="id"
column="ID"
jdbc-type="INTEGER"
primarykey="true"
autoincrement="true"
/>
<field-descriptor
name="name"
column="NAME"
jdbc-type="VARCHAR"
size="255"
/>
<field-descriptor
name="drawerId"
column="DRAWER_ID"
jdbc-type="INTEGER"
access="anonymous"
/>
</class-descriptor>
<class-descriptor
class="Toy"
table="TOYS">
<field-descriptor
name="id"
column="ID"
jdbc-type="INTEGER"
primarykey="true"
autoincrement="false"
/>
<field-descriptor
name="category"
column="CATEGORY"
jdbc-type="VARCHAR"
size="255"
/>
</field-descriptor>
<reference-descriptor
name="super"
class-ref="Thing"
>
<foreignkey field-ref="id"/>
</reference-descriptor>
</class-descriptor>
<class-descriptor
class="Book"
table="BOOKS">
<field-descriptor
name="id"
column="ID"
jdbc-type="INTEGER"
primarykey="true"
autoincrement="false"
/>
<field-descriptor
name="author"
column="AUTHOR"
jdbc-type="VARCHAR"
size="255"
/>
<field-descriptor
name="isbnCode"
column="ISBN_CODE"
jdbc-type="VARCHAR"
size="255"
/>
</field-descriptor>
<reference-descriptor
name="super"
class-ref="Thing"
>
<foreignkey field-ref="id"/>
</reference-descriptor>
</class-descriptor>
<class-descriptor
class="Drawer"
table="DRAWER">
<field-descriptor
name="id"
column="id"
jdbc-type="INTEGER"
primarykey="true"
autoincrement="true"
/>
<collection-descriptor
name="stuffInDrawer"
element-class-ref="Thing"
>
<inverse-foreignkey field-ref="drawerId"/>
</collection-descriptor>
</class-descriptor>
The FK field "drawerId" in class Thing (of the 1:n reference in Drawer) is
declared as anonymous. Please read carefully the section about how anonymous keys work to
avoid problems
http://db.apache.org/ojb/docu/guides/advanced-technique.html#How+do
regards,
Armin
Virgo Smart wrote:
Hello,
I am facing an issue when trying to retrieve the Drawer object.
Instead of retrieving a collection of Toy and Book instances in the
stuffInDrawer attribute, collection containing Thing instances is returned.
Following is the class descriptor that is used.
<class-descriptor
class="Thing"
table="THINGS">
<extent-class class-ref="Toy"/>
<extent-class class-ref="Book"/>
<field-descriptor
name="id"
column="ID"
jdbc-type="INTEGER"
primarykey="true"
autoincrement="true"
/>
<field-descriptor
name="name"
column="NAME"
jdbc-type="VARCHAR"
size="255"
/>
<field-descriptor
name="drawerId"
column="DRAWER_ID"
jdbc-type="INTEGER"
access="anonymous"
/>
</class-descriptor>
<class-descriptor
class="Toy"
table="TOYS">
<field-descriptor
name="id"
column="ID"
jdbc-type="INTEGER"
primarykey="true"
autoincrement="true"
/>
<field-descriptor
name="category"
column="CATEGORY"
jdbc-type="VARCHAR"
size="255"
/>
<field-descriptor
name="parentThingId"
column="THING_ID"
jdbc-type="BIGINT"
access="anonymous"
>
</field-descriptor>
<reference-descriptor
name="super"
class-ref="Thing"
>
<foreignkey field-ref="parentThingId"/>
</reference-descriptor>
</class-descriptor>
<class-descriptor
class="Book"
table="BOOKS">
<field-descriptor
name="id"
column="ID"
jdbc-type="INTEGER"
primarykey="true"
autoincrement="true"
/>
<field-descriptor
name="author"
column="AUTHOR"
jdbc-type="VARCHAR"
size="255"
/>
<field-descriptor
name="isbnCode"
column="ISBN_CODE"
jdbc-type="VARCHAR"
size="255"
/>
<field-descriptor
name="parentThingId"
column="THING_ID"
jdbc-type="BIGINT"
access="anonymous"
>
</field-descriptor>
<reference-descriptor
name="super"
class-ref="Thing"
>
<foreignkey field-ref="parentThingId"/>
</reference-descriptor>
</class-descriptor>
<class-descriptor
class="Drawer"
table="DRAWER">
<field-descriptor
name="id"
column="id"
jdbc-type="INTEGER"
primarykey="true"
autoincrement="true"
/>
<collection-descriptor
name="stuffInDrawer"
element-class-ref="Thing"
>
<inverse-foreignkey field-ref="drawerId"/>
</collection-descriptor>
</class-descriptor>
I have used the table per subclass hierarchy strategy.
I have used SequenceManagerHighLowImpl as the sequence manager. The ids (ID
column) in the TOYS and BOOKS tables are unique when the Drawer object is
stored. In the sense, when a Drawer containing 3 things (2 books and 1 toy) is
stored, the ids stored in THINGS table are 1, 2, and 3, where as the ids stored
in BOOKS table are 1 and 2 and id stored in the TOYS table is 3.
I hope I have provided sufficient details to debug the issue. Could anyone
please point me as to where am I going wrong or if there is any way to
circumvent the problem ?
Thanks and Regards,
Gautam.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
