Need help with Hibernate? Read this first:
http://www.hibernate.org/ForumMailingli ... AskForHelp

Hibernate version: 3.1.x

Mapping documents:

Code between sessionFactory.openSession() and session.close():

Full stack trace of any exception that occurs:
1.java.lang.ClassCastException: java.lang.String
at org.hibernate.type.TimestampType.toString(TimestampType.java:55)

2.
org.hibernate.PropertyAccessException: IllegalArgumentException occurred calling getter of com.pa.asset.orm.base.BaseAgent.UserName
at org.hibernate.property.BasicPropertyAccessor$BasicGetter.get(BasicPropertyAccessor.java:171)
at org.hibernate.tuple.AbstractEntityTuplizer.getIdentifier(AbstractEntityTuplizer.java:176)
at org.hibernate.persister.entity.AbstractEntityPersister.getIdentifier(AbstractEntityPersister.java:3257)
at org.hibernate.persister.entity.AbstractEntityPersister.isTransient(AbstractEntityPersister.java:2983)
at org.hibernate.engine.ForeignKeys.isTransient(ForeignKeys.java:181)
The generated SQL (show_sql=true):

Debug level Hibernate log excerpt:



I am facing these issues migrating from 3.0 to 3.1.x for any operation (selects or update) .Most of these are simple select operations and work fine in 3.0 .
The first execption occurs always ,presumably beacuse most of the hiberanate mapping files have a date / timestamp field .
The second exception occurs possibly beacuse the primary key in the table is a character filed and not a numeric field..
These exceptions

I have been using hibernate synchronizer (an eclipse plugin ) to generate these mapping files ..When will i be able to use hibernate 3.1 afterall

Can you post your HBM.xml for class "com.pa.asset.orm.base.BaseAgent" ?

Can you post your table description for the table, in mysql "DESCRIBE myTable" ?

Can you post the POJO or at least the getter/setters, constructor, field variables ?

<class
name="Agent"
table="Agent" >
<id
name="UserName"
type="string"
column="UserName">

<generator class="assigned"/>
</id>


<property
name="FullName"
column="FullName"
type="string"
not-null="false"
length="50"
/>

<property
name="Created"
column="Created"
type="timestamp"
not-null="false"
length="23"
/>

<property
name="Grace"
column="Grace"
type="integer"
not-null="false"
length="10"
/>



<set name="MemberOfGroups" inverse="true">
<key column="UserName"/>
<one-to-many class="MemberOfGroup" />
</set>

<many-to-one
name="Person"
column="Person_ID"
class="Person"
not-null="false">

</many-to-one>


</class>


CREATE TABLE Agent (
UserName NVARCHAR(10) NOT NULL,
FullName NVARCHAR(50),
Expires DATETIME,
Person_ID INTEGER,
Grace INTEGER,
Created DATETIME
)
Go

CREATE INDEX PK_Agent ON Agent (UserName ASC)
Go

ALTER TABLE Agent ADD CONSTRAINT PK_Agent PRIMARY KEY (UserName)
Go

ALTER TABLE Agent ADD CONSTRAINT FK_Agent_PERSON FOREIGN KEY (Person_ID)
REFERENCES PERSON (ID)
ON DELETE RESTRICT
ON UPDATE RESTRICT
Go




// primary key
private java.lang.String userName;

// fields
private java.lang.String fullName;
private java.util.Date expires;
private java.lang.Integer grace;
private java.util.Date created;

// many to one
private com.pa.asset.orm.Person person;

// collections
private java.util.Set<com.pa.asset.orm.MemberOfGroup> memberOfGroups;
private java.util.Set<com.pa.asset.orm.Person> agentPerson;

Here are some of the different type of fields used in this mapping
Agent class is inherited from the class BaseAgent..and I am using hibernate with Spring 1.2.6 ,tomcat 5.5 and SQL server 2000

there is not much work needed in migrating to version 3.1 .
http://www.hibernate.org/357.html
i think the PropertyAccesException is due to wrong type of value passed as id .

But this code works in hibernate 3.0 and its not just one place .I get these errors every where ...More over these are not update operations mere select operations ....I am wondering if there is any jar file that is causing this problem

I use java.sqlTimestamp in my POJOs due to having problem using java.lang.Date, never had a problem since.

Given the error is related to a Timestamp hibernate type is it possible to try that avenue out ?

I managed to find the cause for the bugs

First one related to the timestamp :
1. In my hibernate query I was doing a query something like
List list = getHibernateTemplate().find(" from table t where t.somedate = ?",args);

where args was defined as
Object[] args = {dateinstringform};

I should have been passing the parameter as an object of type Date()
like

Object[] args = {dateindateform};

I guess hibernate 3.0 automatically did the type conversion from String to Date ..
-------------------------------
2.An even more amazing reason for the second bug ..This one was a shocker....

My query was something like

List list = getHibernateTemplate().find("select grp from Person p,Group grp where p.type = grp.class and grp.Person = ?",args);

where args was defined as
Object[] args = {personinstringform};

It should have been

Object[] args = {new Person(personinstringform)};

But I guess since the primary key was in string form, Hibernate 3.0 automagically created a new type using the constructor Person(String name) ....

I wonder if these features were removed on purpose in hibernate 3.1 ... I feel it may be a good thing to have this type of automatic type conversion atleast in the first case...Would appreciate some kind of thoughts about this from the guys at hibernate too..How about some kind of polling
____________________________________________________
Quote:That which hits the fan tends to get flung in all directions.Get out of the way before u throw

My 2 cents.

Object[] usualy means you have to be type specific if type matters; this is common to "the java way of using this language technique". Since 99% of all objects have a String representation I suppose its not possible to guess the type 100% if all you provide is a String.

Granted HQL can attempt to work out the datatype from the context, but aren't you asking it to do a lot for every execution of the statement and leaving the door open to easy programming errors that probably wont show up in normal testing. All that processing work would be unecessary when I think that for every situation the programmer already knows the type when he constructs the HQL.

My 2 cents.

But hibernate 3.0 did recognized it right ..I wonder why this stopped working in 3.1 ( and what such things have stopped working so I can trace where such operations have been performed)..


Yes it can lead to lazy & error prone programming if used unwisely ..but it can be useful for simple operations like converting from String to a numeric type or date type..where in the type of data is usually validated in the front end or in the Struts layer