inverse = “true” example and explanation
There are many Hibernate articles try to explain the “inverse” with many Hibernate “official” jargon, which is very hard to understand (at least to me). In few articles, they even suggested that just forget about what is “inverse”, and always put inverse=”true” in the collection variable.
This statement is always true – “put inverse=true in collection variable”, but do not blindfold on it, try to understand the reason behind is essential to optimal your Hibernate performance.
What is “inverse” ?
This is the most confusing keyword in Hibernate, at least i took quite a long time to understand it. The “inverse” keyword is always declare in one-to-many and many-to-many relationship (many-to-one doesn’t has inverse keyword), it means which side is responsible to take care of the relationship.
“inverse”, should change to “relationship owner”?
In Hibernate, only the “relationship owner” should maintain the relationship, and the “inverse” keyword is created to defines which side is the owner to maintain the relationship. However the “inverse” keyword itself is not verbose enough, I would suggest change the keyword to “relationship_owner“.
In short, inverse=”true” means this is the relationship owner, and inverse=”false” (default) means it’s not.
1. One to many Relationship
This is a one-to-many relationship table design, a STOCK table has many occurrences in STOCK_DAILY_RECORD table.
2. Hibernate Implementation
See the Hibernate implementation in XML mapping files.
File : Stock.java
public class Stock implements java.io.Serializable {
...
private Set<StockDailyRecord> stockDailyRecords =
new HashSet<StockDailyRecord>(0);
...
File : StockDailyRecord.java
public class StockDailyRecord implements java.io.Serializable {
...
private Stock stock;
...
File : Stock.hbm.xml
<hibernate-mapping>
<class name="com.mkyong.common.Stock" table="stock" ...>
...
<set name="stockDailyRecords" table="stock_daily_record" fetch="select">
<key>
<column name="STOCK_ID" not-null="true" />
</key>
<one-to-many class="com.mkyong.common.StockDailyRecord" />
</set>
...
File : StockDailyRecord.hbm.xml
<hibernate-mapping>
<class name="com.mkyong.common.StockDailyRecord" table="stock_daily_record" ...>
...
<many-to-one name="stock" class="com.mkyong.common.Stock">
<column name="STOCK_ID" not-null="true" />
</many-to-one>
...
3. inverse = true / false
Inverse keyword is applied in one to many relationship. Here’s the question, if save or update operation perform in “Stock” object, should it update the “stockDailyRecords” relationship?
File : Stock.hbm.xml
<class name="com.mkyong.common.Stock" table="stock" ...>
...
<set name="stockDailyRecords" table="stock_daily_record" inverse="{true/false}" fetch="select">
<key>
<column name="STOCK_ID" not-null="true" />
</key>
<one-to-many class="com.mkyong.common.StockDailyRecord" />
</set>
...
1. inverse=”true”
If inverse=”true” in the set variable, it means “stock_daily_record” is the relationship owner, so Stock will NOT UPDATE the relationship.
<class name="com.mkyong.common.Stock" table="stock" ...>
...
<set name="stockDailyRecords" table="stock_daily_record" inverse="true" >
2. inverse=”false”
If inverse=”false” (default) in the set variable, it means “stock” is the relationship owner, and Stock will UPDATE the relationship.
<class name="com.mkyong.common.Stock" table="stock" ...>
...
<set name="stockDailyRecords" table="stock_daily_record" inverse="false" >
See more examples below :
4. inverse=”false” Example
If keyword “inverse” is not define, the inverse = “false” will be used, which is
<!--Stock.hbm.xml-->
<class name="com.mkyong.common.Stock" table="stock" ...>
...
<set name="stockDailyRecords" table="stock_daily_record" inverse="false">
It means “stock” is the relationship owner, and it will maintains the relationship.
Insert example …
When a “Stock” object is saved, Hibernate will generated three SQL statements, two inserts and one update.
session.beginTransaction();
Stock stock = new Stock();
stock.setStockCode("7052");
stock.setStockName("PADINI");
StockDailyRecord stockDailyRecords = new StockDailyRecord();
stockDailyRecords.setPriceOpen(new Float("1.2"));
stockDailyRecords.setPriceClose(new Float("1.1"));
stockDailyRecords.setPriceChange(new Float("10.0"));
stockDailyRecords.setVolume(3000000L);
stockDailyRecords.setDate(new Date());
stockDailyRecords.setStock(stock);
stock.getStockDailyRecords().add(stockDailyRecords);
session.save(stock);
session.save(stockDailyRecords);
session.getTransaction().commit();
Output…
Hibernate:
insert
into
mkyongdb.stock
(STOCK_CODE, STOCK_NAME)
values
(?, ?)
Hibernate:
insert
into
mkyongdb.stock_daily_record
(STOCK_ID, PRICE_OPEN, PRICE_CLOSE, PRICE_CHANGE, VOLUME, DATE)
values
(?, ?, ?, ?, ?, ?)
Hibernate:
update
mkyongdb.stock_daily_record
set
STOCK_ID=?
where
RECORD_ID=?
Stock will update the “stock_daily_record.STOCK_ID” through Set variable (stockDailyRecords), because Stock is the relationship owner.
The third statement is really NOT necessary.
Update example …
When a “Stock” object is updated, Hibernate will generated two SQL statements, one inserts and one update.
session.beginTransaction();
Stock stock = (Stock)session.get(Stock.class, 57);
StockDailyRecord stockDailyRecords = new StockDailyRecord();
stockDailyRecords.setPriceOpen(new Float("1.2"));
stockDailyRecords.setPriceClose(new Float("1.1"));
stockDailyRecords.setPriceChange(new Float("10.0"));
stockDailyRecords.setVolume(3000000L);
stockDailyRecords.setDate(new Date());
stockDailyRecords.setStock(stock);
stock.getStockDailyRecords().add(stockDailyRecords);
session.save(stockDailyRecords);
session.update(stock);
session.getTransaction().commit();
Output…
Hibernate:
insert
into
mkyongdb.stock_daily_record
(STOCK_ID, PRICE_OPEN, PRICE_CLOSE, PRICE_CHANGE, VOLUME, DATE)
values
(?, ?, ?, ?, ?, ?)
Hibernate:
update
mkyongdb.stock_daily_record
set
STOCK_ID=?
where
RECORD_ID=?
Again, the third statement is NOT necessary.
5. inverse=”true” Example
If keyword “inverse=true” is defined :
<!--Stock.hbm.xml-->
<class name="com.mkyong.common.Stock" table="stock" ...>
...
<set name="stockDailyRecords" table="stock_daily_record" inverse="true">
Now, it means “stockDailyRecords” is the relationship owner, and “stock” will not maintains the relationship.
Insert example …
When a “Stock” object is saved, Hibernate will generated two SQL insert statements.
session.beginTransaction();
Stock stock = new Stock();
stock.setStockCode("7052");
stock.setStockName("PADINI");
StockDailyRecord stockDailyRecords = new StockDailyRecord();
stockDailyRecords.setPriceOpen(new Float("1.2"));
stockDailyRecords.setPriceClose(new Float("1.1"));
stockDailyRecords.setPriceChange(new Float("10.0"));
stockDailyRecords.setVolume(3000000L);
stockDailyRecords.setDate(new Date());
stockDailyRecords.setStock(stock);
stock.getStockDailyRecords().add(stockDailyRecords);
session.save(stock);
session.save(stockDailyRecords);
session.getTransaction().commit();
Output …
Hibernate:
insert
into
mkyongdb.stock
(STOCK_CODE, STOCK_NAME)
values
(?, ?)
Hibernate:
insert
into
mkyongdb.stock_daily_record
(STOCK_ID, PRICE_OPEN, PRICE_CLOSE, PRICE_CHANGE, VOLUME, DATE)
values
(?, ?, ?, ?, ?, ?)
Update example …
When a “Stock” object is updated, Hibernate will generated one SQL statement.
session.beginTransaction();
Stock stock = (Stock)session.get(Stock.class, 57);
StockDailyRecord stockDailyRecords = new StockDailyRecord();
stockDailyRecords.setPriceOpen(new Float("1.2"));
stockDailyRecords.setPriceClose(new Float("1.1"));
stockDailyRecords.setPriceChange(new Float("10.0"));
stockDailyRecords.setVolume(3000000L);
stockDailyRecords.setDate(new Date());
stockDailyRecords.setStock(stock);
stock.getStockDailyRecords().add(stockDailyRecords);
session.save(stockDailyRecords);
session.update(stock);
session.getTransaction().commit();
Output…
Hibernate:
insert
into
mkyongdb.stock_daily_record
(STOCK_ID, PRICE_OPEN, PRICE_CLOSE, PRICE_CHANGE, VOLUME, DATE)
values
(?, ?, ?, ?, ?, ?)
Many people like to compare between inverse and cascade, but both are totally different notions, see the differential here.
Conclusion
Understanding the “inverse” is essential to optimize your Hibernate code, it helps to avoid many unnecessary update statements, like “insert and update example for inverse=false” above. At last, try to remember the inverse=”true” mean this is the relationship owner to handle the relationship.
Probably it’s me, but I found this article very confusing. I appreciate your effort expetially for the SQL generated query part but IMHO your article can be improved.
The main source of confusion comes from the part when you say that inverse should be renamed “relationship_owner” while it is clear that it is rather the opposite. It should be renamed “relationship_NOT_owner”. In one of your example you say that having “inverse=true” makes the Stock object not to update the other side of the relationship. Thus at the end of the day I find that “Inverse” is the correct name for the flag, It means that THIS side is the INVERSE side of a relationship that means the other side is the owner.
Another way of improving your article IMHO would be to clearly explain what a relationship owner actually is and what it is expected to do.
Actually it’s from the Set’s perspective
…
The set stockDailyRecords is the owner
confusing
Thank you. Very helpful !!
clearest explanation I read.
Can you please explain, if inverse=false, and if stock_id inside stock_daily_record is a not null record then how will this work.
Do we not get an error while executing insert into stock_daily_record?
Hi, where is the update query for stock object in both update examples.
thanks for the post
ya it is awsome explation…but can we used inverse with list tag..m trying but seq column return null value
I don’t see link to download source code. Could you please make it available?
Awesome! What a confusing topic “inverse”. Eventually, I got tired of figuring it out myself, and tried this and that. You my man, provided a great example.
Good work.
For your case 4. inverse=”false” Example
When a “Stock” object is saved, Hibernate will generated three SQL statements, two inserts and one update.
Then, you mention:
Note
Again, the third statement is NOT necessary.
stockDailyRecords.setStock(stock);
//stock.getStockDailyRecords().add(stockDailyRecords); // COMMENTED OUT THIS LINE
session.save(stock);
session.save(stockDailyRecords);
Then I was able to prevent that update which is really not necessary. And i got the resultsimilar to that of case 5. So, what is the difference then?
lot of thanx
As you say someone is relationship owner and it will take responsibility to maintain relationship.. Let me tell you this line becomes a most confusing one. In example you save both the objects STOCK as well as Stock_Daily_Record.
In inverse=true, Stock_Daily_Record is relationship owner OR inverse=false where STOCK is relationship owner. In both of these case why do you save both objects ?
Because if you see code wise both looks similar.
Hi Mkyong,
Thanks for writing such a good article. Your article really makes it easy to understand.
But my problem is I am using inverse=”true” i.e. stockDailyRecords is the owner.
In first step I inserted stockDailyRecords record so it will generate two queries
1) insert in stockDailyRecords table and
2) insert in stock table
In second step I am updating stockDailyRecords so it will generate two queries..
1) update stockDailyRecords table and
2) insert in stock table
but in this case I am getting unique constraint violated because it is trying to insert in the same record in stock twoice.
Is there any solution?
Use not not-null= false both side.
Thank you. Much clear now.
Once again clear my confusion.
Awesome, lucid and straight-forward explanation. Thanks a lot.
You saved my day! Thanks!
However, I have one query. I saw this statement in this post:
“In few articles, they even suggested that just forget about what is “inverse”, and always put inverse=”true” in the collection variable.”
Are we not supposed to use inverse=”false”. Why?
My implementation is like this:
I am maintaining only one-to-many association in XML, not many-to-one. In java, I am not giving reference of parent to the child. And, by keeping inverse=”false”, the foreign key is getting updated through the update query.
Is this practice ok? Or should I switch to more conventional two way relationship style?
Very nice explaination…It’s a big confusion portion of NHibernate,you resolved it very nicely…
Thanks a Ton…
MKyoung,
First of all, let me congratulate you for this awesome job, your website has been a life saver lately. Keep going.
It’d be nice if you could make some reference to inverse control using annotations too. It worthwhile to say that mappedBy attribute does the same thing when you are using an biderection relationship between @OneToMany(mappedBy=”attributeyouwanttomap”) and @ManyToOne()
Thanks once more for you great job!
if inverse=”false”
and
When Stock object is saved,
Why we need stockDailyRecord to be saved?
“session.save(stockDailyRecords);”
saving the stockDailyRecord is based on Cascade .. not on inverse … inverse just make sure who is the owner of the relationship
very well written & explained.
very clear . thanks
Thanks Mkyong for the nice artice.
In the section:[ inverse=”false” Example], I am not getting what you meant by the below:
“Note
Again, the third statement is NOT necessary.”
Which third statement are you referring to? Please explain.
In the Update Example of the (Inverse = false) , Actually there was not any third statement ,it was by mistake written that “Again, the third statement is NOT necessary.” , so Instead of third statement he is referring to the second update statement actually.
Hence it was “Again, the second statement is NOT necessary.”
Hi MkYong,
Its a really nice article.
Can you please post your code for download purpose.
It will definitely make our understanding lot better.Thanks in advance.
Abhishek J
very good artical with practical example.Very clear.Thanks a lot for making me understand.
This is a link very short definition of when to use inverse=true/false
http://tadtech.blogspot.com/2007/02/hibernate-when-is-inversetrue-and-when.html
I used to work for RH and I asked the same thing
Why does Gavin use variable-names that have nothing to do with the function the variable performs.
inverse means nothing..he should change it to relationship-owner
I think Gavin coded this before Collection came into play..so from what i see
1)Hibernates extension of Java collection classes makes no sense with new generic impl’s
2)I worked at one installation on the Pacific Coast that ripped Hibernate out of their code because the CTO said it is an ill-named,undocumented,resource-hogging piece of !!!!
Martin-
Confusion and Obfuscation are the main tactics used in battling adversaries..Sun Tsu
Gr8 Example!
As clear as ever. Thanks mkyong.