db4o Developer Community

Hi,

 Can anybody plz tell me,  why this simple query is not optimized? I have an Order-class with an Id attribute and want to find an order with a specific Id. but the checkOptimization-method shows that this query is UNOPTIMIZED.

 

    public void getOrderByID(ObjectContainer db, final String id) {
        checkOptimization(db);
        ObjectSet<Order> results = db.query(new Predicate<Order>() {
            public boolean match(Order candidate) {
                return candidate.getReferenceId().equals(id);
            }
        });
        for (Order order : results) {
            LOGGER.info(order);
        }
    }

 ////////////   checkOptimization method ////////////

    public static void checkOptimization(ObjectContainer db){
        ((ObjectContainerBase)db).getNativeQueryHandler().addListener(new Db4oQueryExecutionListener() {
            public void notifyQueryExecuted(NQOptimizationInfo info) {
                System.err.println(info);
            }
        });
    }

 

thanks.

Have you tried == instead of .equals()?

Brett 

Yes, but it does not help. I can not even find my object with "==" and the query is furthermore  not optimized and a full-search will be done.

Hi!

Can you check against the field instead of get method? It seems the optimizer doesn't handle it.

Goran

Goran:

Can you check against the field instead of get method? It seems the optimizer doesn't handle it.

Hi Goran,

just to make things clear: if bean-conventions are used, it should work with accessors, too... at least I think so...;-)

So: if the field is called referenceId, getReferenceId() should be resolved by the NQ-Parser to fetch the appropriate field. But if the field is called id, but the accessor is still called getReferenceId(), the query cannot be optimized. Maybe this is the problem.

And just another one: If I remember correctly, the accessor needs to be a "real simple" accessor, which consists simply of a "return referenceId"... if there's more code involved in the accessor, the NQ-Parser does a good job if it doesn't optimize the query...

Cheers, Maik

---

http://db4o.blogspot.com/

thanks 4 replies.

@  Gora:  I have checked against the field, but it is still unoptimized.

@ Maik: I have the bean-conventions in my code:

 

public  abstract class Order {
   
    private String referenceId;

    public String getReferenceId() {
        return referenceId;
    } ...

I have about 100.000 Objects, and you can imagine how long does it take to run a full-search. I have no Idea.

 

 

Just two ideas:

• Do you have bloat.jar and db4o-nq.jar in your classpath?
• Please change your class definition to: public abstract class Order {}. I'm not totally sure, but if I remember correctly there were some issues with abstract classes and nq-optimazation.

Cheers, Maik

---

http://db4o.blogspot.com/

thanks Maik for your help, but unfortunately it did not help. I have had the both jars in my classpath. With or without the abstract definition, the result is the same.

cheers, Mehdi

Hi,

the following simple test runs optimized. Can you provide a similiar test which shows the problem?

Cheers, Maik

 

import java.io.File;

import com.db4o.Db4o;
import com.db4o.ObjectContainer;
import com.db4o.internal.ObjectContainerBase;
import com.db4o.internal.query.Db4oQueryExecutionListener;
import com.db4o.internal.query.NQOptimizationInfo;
import com.db4o.query.Predicate;

public class NativeQueryByStringTest {

    private String id;

    public NativeQueryByStringTest(String id) {
        this.id = id;
    }

    public String getId() {
        return id;
    }

    public static void main(String[] args) {
        new File("test.yap").delete();
        
        ObjectContainer db4o = Db4o.openFile("test.yap");
        for (int i = 0; i < 100000; i++) {
            db4o.store(new NativeQueryByStringTest("" + i));
        }
        db4o.commit();
        db4o.close();

        db4o = Db4o.openFile("test.yap");
        ((ObjectContainerBase) db4o).getNativeQueryHandler().addListener(
                new Db4oQueryExecutionListener() {
                    public void notifyQueryExecuted(NQOptimizationInfo info) {
                        System.err.println(info);
                    }
                });
        
        final String id = "12345";
        System.out.println(db4o.query(new Predicate<NativeQueryByStringTest>() {
            public boolean match(NativeQueryByStringTest candidate) {
                return candidate.getId().equals(id);
            }        
        }));        
        db4o.close();
    }
}

---

http://db4o.blogspot.com/

Hi Maik, thanks for ur post. I have the same code and after I have run your code the output was:

 UNOPTIMIZED/null
[api.models.com.db4o.performance.NativeQueryByStringTest@139f2f6]

 

Can you plz post your classpath file? mine is as follows:

<?xml version="1.0" encoding="UTF-8"?>
<classpath>
    <classpathentry kind="src" path="src/main/java"/>
    <classpathentry kind="con" path="org.eclipse.jdt.launching.JRE_CONTAINER"/>
    <classpathentry kind="lib" path="lib/log4j-1.2.11.jar"/>
    <classpathentry kind="lib" path="lib/bloat-1.0.jar"/>
    <classpathentry kind="lib" path="lib/db4o-7.2.39.10644-nqopt.jar"/>
    <classpathentry kind="lib" path="lib/db4o-7.2.39.10644-java5.jar"/>
    <classpathentry kind="output" path="bin"/>
</classpath>
 

 

cheers, Mehdi

My output is:

DYNOPTIMIZED/CANDIDATE.id == "12345"
[NativeQueryByStringTest@1e4fc14]

.classpath is nearly the same, at least no essential difference... 

<?xml version="1.0" encoding="UTF-8"?>
<classpath>
    <classpathentry kind="src" path="src"/>
    <classpathentry kind="con" path="org.eclipse.jdt.launching.JRE_CONTAINER"/>
    <classpathentry kind="lib" path="/home/mjablonski/db4o-7.2/lib/bloat-1.0.jar"/>
    <classpathentry kind="lib" path="/home/mjablonski/db4o-7.2/lib/db4o-7.2.39.10644-java5.jar"/>
    <classpathentry kind="lib" path="/home/mjablonski/db4o-7.2/lib/db4o-7.2.39.10644-nqopt.jar"/>
    <classpathentry kind="output" path="bin"/>
</classpath>

Very strange... I'm running the test from within Eclipse using JDK 1.6.07. Do you use any "unusual" settings for your project?

Cheers, Maik

---

http://db4o.blogspot.com/

yes, it is strange. I m using Eclipse too. My JDK is 1.5, but that could not be the reason. thanks anyway Maik.

Mehdi

I am having the same kind of problem as this thread. Using db40 7.2.10988, JDK 5.

 

I found the solution to my particular problem.

My Data bean has private fields that begin with '_', like:

    private String _author;

And the accessor methods do not use the underbar:

    public String getAuthor() { return _author; }

 

When I change my private member to plain 'author', the queries start being optimized.