Re: Problem with reading records having a PK with value 0 (not null)
Hi Danilo,
The problem lies in the support for primitive data types, since a
primitive type cannot be null, the value 0 is interpreted as null.
It seems that non primitive data types also react the same way (even
if they shouldn't, so from my point of view it is a bug)
I checked in two tests reproduce the described problem
(PersistenceBrokerTest in test suite) and a bug fix for
BrokerHelper in CVS.
Many thanks for your detailed description and sample code.
regards,
Armin
Danilo Tommasina wrote:
Hi Armin,
here a short example.
create TABLE FOO ( TAB_ID INTEGER PRIMARY KEY )
and put an entry into the table with ID = 0 (from SQL not from OJB)
/*/
repoistory.xml
class-descriptor class=org.apache.ojb.test.Foo table=FOO
field-descriptor name=tabId column=TAB_ID jdbc-type=DECIMAL
nullable=false primarykey=true autoincrement=true /
class-descriptor/
//
public class Foo {
public Integer tabId;
//Add getter and setter
}
//
some test code
(...)
Criteria c = new Criteria();
c.addEqualTo( tabId, new Integer( 0 ) );
Query q = new QueryByCriteria( Foo.class, c );
Foo myFoo = (Foo) pb.getObjectByQuery( q );
pb.beginTransaction();
pb.store( myFoo );
pb.commitTransaction();
(...)
and now you have two entries in your database (Integer PK with value 0
was interpreded as null and a new clone has been inserted)
This is actually a minor issue, since it is not possible through OJB to
add a DB entry with a PK = 0, in this case the error will never occur.
However if you are using data inserted by other applications it is
possible to have entries with ID = 0 to work with.
sorry I didn't tested the code, I hope it works...
bye
Danilo
Hi Danilo,
Danilo Tommasina wrote:
Hi,
I reported the same issue some months ago...
The problem lies in the support for primitive data types, since a
primitive type cannot be null, the value 0 is interpreted as null.
It seems that non primitive data types also react the same way (even
if they shouldn't, so from my point of view it is a bug)
if so, I agree sounds like a bug. Can you post a test case to
reproduce the problem or describe what to do with pseudo code?
Fetching an object with ID = 0 works, however OJB will handle the 0
value as null and put the next sequence value into the fetched object
on next store or when calling a
PeristenceBroker.retrieveAllReferencies( Object obj )
For OJB 1.1 a complete refactoring of the primitive data types
handling is scheduled.
Here you can post your proposals for the upcoming 1.1 version
http://nagoya.apache.org/wiki/apachewiki.cgi?OJBProjectPages/OJBOnePointOne
The workaround is to never use IDs = 0 and this is actually what I am
doing ;)
or don't use primitive data types for PK fields in your pc objects (or
do both ;-)).
regards,
Armin
bye
Danilo
Hi,
It looks like OJB cannot fetch correctly records which have a BigInt
primary
key with value 0; at least in the following conditions;
- I'm using RC5
- Have a table (here Locations) with a primary key Id with a
auto-increment
sequence
- I'm using Oracle 9.2
- The table has a 1-N collection using a proxy.
For instance, I have the following mapping:
class-descriptor class=Locations table=LOCATIONS
field-descriptor name=id column=ID
jdbc-type=BIGINT primarykey=true
autoincrement=true
sequence-name=locations_id_seq/
field-descriptor name=latitude column=LATITUDE
jdbc-type=BIGINT/
field-descriptor name=longitude column=LONGITUDE
jdbc-type=BIGINT/
...
collection-descriptor name=collElements
element-class-ref=Elements proxy=true
inverse-foreignkey field-ref=locationid/
/collection-descriptor
/class-descriptor
Now in the database, there is a location with Id=0.
When I use OJB to fetch this record (with a criteria on Id=0), the
resulting
instance is correct except for the Id field which has the next sequence
value.
Tracing the code, what happens is when the query is executed:
- RsIterator.getObjectFromResultSet reteive the simple fields
correctly
- As there is a collection, retrieveCollections is called
- The query to retrieve the collection is built: Query
fkQuery =
getFKQuery(obj, cld, cds);
- This retrieves the PK fields values (getFKQuery1toN ) which
indirectly calls getValuesForObject
- When the value for the Id is retrieved
(getAutoIncrementValue),
it is considered as Null (BrokerHelper.isNull) and thus the sequence is
called.
Can this be considered a bug?
Are there any workarounds?
Thanks,
-Alex
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands,
Re: Problem with reading records having a PK with value 0 (not null)
Hi, I reported the same issue some months ago... The problem lies in the support for primitive data types, since a primitive type cannot be null, the value 0 is interpreted as null. It seems that non primitive data types also react the same way (even if they shouldn't, so from my point of view it is a bug) Fetching an object with ID = 0 works, however OJB will handle the 0 value as null and put the next sequence value into the fetched object on next store or when calling a PeristenceBroker.retrieveAllReferencies( Object obj ) The workaround is to never use IDs = 0 and this is actually what I am doing ;) bye Danilo Hi, It looks like OJB cannot fetch correctly records which have a BigInt primary key with value 0; at least in the following conditions; - I'm using RC5 - Have a table (here Locations) with a primary key Id with a auto-increment sequence - I'm using Oracle 9.2 - The table has a 1-N collection using a proxy. For instance, I have the following mapping: class-descriptor class=Locations table=LOCATIONS field-descriptor name=id column=ID jdbc-type=BIGINT primarykey=true autoincrement=true sequence-name=locations_id_seq/ field-descriptor name=latitude column=LATITUDE jdbc-type=BIGINT/ field-descriptor name=longitude column=LONGITUDE jdbc-type=BIGINT/ ... collection-descriptor name=collElements element-class-ref=Elements proxy=true inverse-foreignkey field-ref=locationid/ /collection-descriptor /class-descriptor Now in the database, there is a location with Id=0. When I use OJB to fetch this record (with a criteria on Id=0), the resulting instance is correct except for the Id field which has the next sequence value. Tracing the code, what happens is when the query is executed: - RsIterator.getObjectFromResultSet reteive the simple fields correctly - As there is a collection, retrieveCollections is called - The query to retrieve the collection is built: Query fkQuery = getFKQuery(obj, cld, cds); - This retrieves the PK fields values (getFKQuery1toN ) which indirectly calls getValuesForObject - When the value for the Id is retrieved (getAutoIncrementValue), it is considered as Null (BrokerHelper.isNull) and thus the sequence is called. Can this be considered a bug? Are there any workarounds? Thanks, -Alex - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: Problem with reading records having a PK with value 0 (not null)
Hi Danilo, Danilo Tommasina wrote: Hi, I reported the same issue some months ago... The problem lies in the support for primitive data types, since a primitive type cannot be null, the value 0 is interpreted as null. It seems that non primitive data types also react the same way (even if they shouldn't, so from my point of view it is a bug) if so, I agree sounds like a bug. Can you post a test case to reproduce the problem or describe what to do with pseudo code? Fetching an object with ID = 0 works, however OJB will handle the 0 value as null and put the next sequence value into the fetched object on next store or when calling a PeristenceBroker.retrieveAllReferencies( Object obj ) For OJB 1.1 a complete refactoring of the primitive data types handling is scheduled. Here you can post your proposals for the upcoming 1.1 version http://nagoya.apache.org/wiki/apachewiki.cgi?OJBProjectPages/OJBOnePointOne The workaround is to never use IDs = 0 and this is actually what I am doing ;) or don't use primitive data types for PK fields in your pc objects (or do both ;-)). regards, Armin bye Danilo Hi, It looks like OJB cannot fetch correctly records which have a BigInt primary key with value 0; at least in the following conditions; - I'm using RC5 - Have a table (here Locations) with a primary key Id with a auto-increment sequence - I'm using Oracle 9.2 - The table has a 1-N collection using a proxy. For instance, I have the following mapping: class-descriptor class=Locations table=LOCATIONS field-descriptor name=id column=ID jdbc-type=BIGINT primarykey=true autoincrement=true sequence-name=locations_id_seq/ field-descriptor name=latitude column=LATITUDE jdbc-type=BIGINT/ field-descriptor name=longitude column=LONGITUDE jdbc-type=BIGINT/ ... collection-descriptor name=collElements element-class-ref=Elements proxy=true inverse-foreignkey field-ref=locationid/ /collection-descriptor /class-descriptor Now in the database, there is a location with Id=0. When I use OJB to fetch this record (with a criteria on Id=0), the resulting instance is correct except for the Id field which has the next sequence value. Tracing the code, what happens is when the query is executed: - RsIterator.getObjectFromResultSet reteive the simple fields correctly - As there is a collection, retrieveCollections is called - The query to retrieve the collection is built: Query fkQuery = getFKQuery(obj, cld, cds); - This retrieves the PK fields values (getFKQuery1toN ) which indirectly calls getValuesForObject - When the value for the Id is retrieved (getAutoIncrementValue), it is considered as Null (BrokerHelper.isNull) and thus the sequence is called. Can this be considered a bug? Are there any workarounds? Thanks, -Alex - 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]
Re: Problem with reading records having a PK with value 0 (not null)
Hi Armin,
here a short example.
create TABLE FOO ( TAB_ID INTEGER PRIMARY KEY )
and put an entry into the table with ID = 0 (from SQL not from OJB)
/*/
repoistory.xml
class-descriptor class=org.apache.ojb.test.Foo table=FOO
field-descriptor name=tabId column=TAB_ID jdbc-type=DECIMAL
nullable=false primarykey=true autoincrement=true /
class-descriptor/
//
public class Foo {
public Integer tabId;
//Add getter and setter
}
//
some test code
(...)
Criteria c = new Criteria();
c.addEqualTo( tabId, new Integer( 0 ) );
Query q = new QueryByCriteria( Foo.class, c );
Foo myFoo = (Foo) pb.getObjectByQuery( q );
pb.beginTransaction();
pb.store( myFoo );
pb.commitTransaction();
(...)
and now you have two entries in your database (Integer PK with value 0
was interpreded as null and a new clone has been inserted)
This is actually a minor issue, since it is not possible through OJB to
add a DB entry with a PK = 0, in this case the error will never occur.
However if you are using data inserted by other applications it is
possible to have entries with ID = 0 to work with.
sorry I didn't tested the code, I hope it works...
bye
Danilo
Hi Danilo,
Danilo Tommasina wrote:
Hi,
I reported the same issue some months ago...
The problem lies in the support for primitive data types, since a
primitive type cannot be null, the value 0 is interpreted as null.
It seems that non primitive data types also react the same way (even
if they shouldn't, so from my point of view it is a bug)
if so, I agree sounds like a bug. Can you post a test case to reproduce
the problem or describe what to do with pseudo code?
Fetching an object with ID = 0 works, however OJB will handle the 0
value as null and put the next sequence value into the fetched object
on next store or when calling a
PeristenceBroker.retrieveAllReferencies( Object obj )
For OJB 1.1 a complete refactoring of the primitive data types handling
is scheduled.
Here you can post your proposals for the upcoming 1.1 version
http://nagoya.apache.org/wiki/apachewiki.cgi?OJBProjectPages/OJBOnePointOne
The workaround is to never use IDs = 0 and this is actually what I am
doing ;)
or don't use primitive data types for PK fields in your pc objects (or
do both ;-)).
regards,
Armin
bye
Danilo
Hi,
It looks like OJB cannot fetch correctly records which have a BigInt
primary
key with value 0; at least in the following conditions;
- I'm using RC5
- Have a table (here Locations) with a primary key Id with a
auto-increment
sequence
- I'm using Oracle 9.2
- The table has a 1-N collection using a proxy.
For instance, I have the following mapping:
class-descriptor class=Locations table=LOCATIONS
field-descriptor name=id column=ID
jdbc-type=BIGINT primarykey=true
autoincrement=true
sequence-name=locations_id_seq/
field-descriptor name=latitude column=LATITUDE
jdbc-type=BIGINT/
field-descriptor name=longitude column=LONGITUDE
jdbc-type=BIGINT/
...
collection-descriptor name=collElements
element-class-ref=Elements proxy=true
inverse-foreignkey field-ref=locationid/
/collection-descriptor
/class-descriptor
Now in the database, there is a location with Id=0.
When I use OJB to fetch this record (with a criteria on Id=0), the
resulting
instance is correct except for the Id field which has the next sequence
value.
Tracing the code, what happens is when the query is executed:
- RsIterator.getObjectFromResultSet reteive the simple fields
correctly
- As there is a collection, retrieveCollections is called
- The query to retrieve the collection is built: Query
fkQuery =
getFKQuery(obj, cld, cds);
- This retrieves the PK fields values (getFKQuery1toN ) which
indirectly calls getValuesForObject
- When the value for the Id is retrieved
(getAutoIncrementValue),
it is considered as Null (BrokerHelper.isNull) and thus the sequence is
called.
Can this be considered a bug?
Are there any workarounds?
Thanks,
-Alex
-
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]
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
