Yea man. i am availabe for stimulating j2ee discussions. expecially in the use of frameworks and patterns.

@subcareer

Finally, here is my solution for exercise two, but i strongly believe that the code can be better,
I will just post as per the requirements


[b]

// PawnTest class

public class PawnTest extends junit.framework.TestCase {

public void testCase() {

String firstPawnColor = "White";
Pawn whitePawn = new Pawn(firstPawnColor);
//String whitePawn = firstPawn.getColor();
assertEquals(firstPawnColor, whitePawn.getColor());

String secondPawnColor = "black";
Pawn blackPawn = new Pawn(secondPawnColor);
assertEquals(secondPawnColor, blackPawn.getColor());

//Pawn noClolor = new Pawn();


}
}

// Pawn class

public class Pawn {

private final static String BLACK_PAWN = "black";
private final static String WHITE_COLOR = "white";
private String color;

public Pawn() {
this.color = WHITE_COLOR;
}

public Pawn(String color){
this.color = color;

}


public String getColor(){

return color;
}

public void setColor(String color){
this.color = color;
}

}

//BoardTest class

import junit.framework.TestCase;
import java.util.*;




public class BoardTest extends TestCase {


private Board board;
private Pawn pawn;

public void setUp(){

pawn = new Pawn();
board = new Board(pawn);


}

public void testCase(){


// Pawn pawn = new Pawn();
//board = new Board();
assertEquals(0,board.getNumberOfPieces());
assertEquals("white",board.getPawnColor());

}

public void testAddPawn(){


pawn = new Pawn("black";
board = new Board(pawn);
board.addPawn(pawn);
assertEquals("black",board.getPawnColor());
assertEquals(1,board.getNumberOfPieces());


Pawn pawn2 = new Pawn("white";
board.addPawn(pawn2);
assertEquals("white", pawn2.getColor());
assertEquals(2, board.getNumberOfPieces());


assertEquals("black",pawn.getColor());
assertEquals("white",pawn2.getColor());

}
}


//Board class
import java.util.*;

public class Board {

//create an ArrayList of pawns to be added to the board
private ArrayList<Pawn> board = new ArrayList<Pawn>();

private Pawn pawn;

//Board constructor, this can be used to add new pawns
public Board(Pawn pawn){
this.pawn = pawn;
}


//public Board() {
// pawn = new Pawn();
//}


public void addPawn(Pawn pawn){
board.add(pawn);
}

public int getNumberOfPieces(){
return board.size();
}

public String getPawnColor(){
return pawn.getColor();
}


}

//AllTest.java

import junit.framework.*;

public class AllTest {

public static TestSuite suite() {
TestSuite suite = new TestSuite();

suite.addTestSuite(PawnTest.class);
suite.addTestSuite(BoardTest.class);

return suite;
}


}
[/b]

@subcareer

Thanks for the encouragement, these are my final codes after refactoring

[b]
// PawnTest class
package pieces;

public class PawnTest extends junit.framework.TestCase {

public void testCase() {

String firstPawnColor = Pawn.WHITE_PAWN;
Pawn whitePawn = new Pawn(firstPawnColor);
assertEquals(firstPawnColor, whitePawn.getColor());

String secondPawnColor = Pawn.BLACK_PAWN;
Pawn blackPawn = new Pawn(secondPawnColor);
assertEquals(secondPawnColor, blackPawn.getColor());

}
}

// Pawn class

package pieces;

public class Pawn {

public final static String BLACK_PAWN = "black";
public final static String WHITE_PAWN = "white";
private String color;


public Pawn(String color){
this.color = color;

}


public String getColor(){

return color;
}

}

//BoardTest class

import junit.framework.TestCase;
import java.util.*;
import pieces.*;

public class BoardTest extends TestCase {


private Board board;


public void setUp(){


board = new Board();


}

public void testCase(){

//Just testing to make sure, I have no pawn on the board
assertEquals(0,board.getNumberOfPieces());


}

public void testAddPawn(){


//create a piece of pawn and add it to the board

Pawn pawn1 = new Pawn(Pawn.BLACK_PAWN);
board.addPawn(pawn1);

//Check numbers of pawns on board
assertEquals(1,board.getNumberOfPieces());

//Test if the pawn is the first pawn added to the board,then test for the colour
assertEquals(pawn1, board.get(0));
assertEquals("black",board.getPawnColor(0));

//same test as for pawn1
Pawn pawn2 = new Pawn(Pawn.WHITE_PAWN);
board.addPawn(pawn2);
assertEquals(2, board.getNumberOfPieces());
assertEquals(pawn2, board.get(1));
assertEquals("white", board.getPawnColor(1));

//Testing with the Pawn getColor() to ensure the colour is as per expected.
assertEquals("black",pawn1.getColor());
assertEquals("white",pawn2.getColor());

}
}

//Board class
import java.util.*;
import pieces.Pawn;



public class Board {

//create an ArrayList of pawns to be added to the board
private List<Pawn> pawns = new ArrayList<Pawn>();


//Board constructor, this can be used to add new pawns
public Board(Pawn pawn){
pawn = new Pawn(Pawn.WHITE_PAWN);
}

public Board(){}


public void addPawn(Pawn pawn){
pawns.add(pawn);
}

public int getNumberOfPieces(){
return pawns.size();
}


public Pawn get(int index){

return pawns.get(index);
}


//Not too comfortable with this method,think there should be a better way
//to handle. Just getting a Pawn object from get(index) and calling it's
//getColor() to get the colour.

public String getPawnColor(int index){
return get(index).getColor();
}


}

[/b]

Hope to hear from you, just started chapter three today. Once more thanks

nice to see you guys are into test driven development as well. nice one.

however junit is only good for testing ur business logic ,if you want to test your database layer or web tier. use. DBUNIT - for database testing and Jwebunit or httpunit for web tier testing. these testing frameworks are essentially built on top to junit

have fun

@subcareer n candylips

Thanks for the support, had to migrate to ubuntu linux,
took most of my time getting all my java stuff to run on it. Had to look for a way
getting a chm reader for linux. Now am back on track and ready to move on.

I must commend you for your enthusiasm.I am highly interested.if only you are willing to teach,i'm willing to learn.The problem here is,i am an experienced VB programmer and i want to make that transition to J2EE.Already started with J2se but i am not too grounded are you guys willing to take me on the whole hog.i'll deeply appreciate it.

Please, subcareer - could you tell me where I can get good computer books in Lagos or if you buy from Amazon.com, how do you do about the shipping. Please, I will appreciate anybody's good response on this. Thanks in advance.

THANKS FOR THE INFORMATION[/color][flash=200,200][/flash][/b]. [/b]

EJB, Spring, Hibernate, Test Driven Development, Frameworks, Patterns. Quite Interesting.

If you guys dont mind, I'll share my perspective on Enterprise Java Development. My Java history went from Java SE/ AWT/Swing to Servlets then JSP then MVC then Struts then Webwork until I eventually chose Tapestry as web framework in preference to JSF. On the business logic side, I studied EJB for a short spell intending to use it wholeheartedly until I kept noticing serious criticisms of its design especially on theserverside.com.

Right now, I use Tapestry as web layer, pure java classes for business layer and Hibernate for persistence. Talking ioC/ Dependency injection I use Hivemind and not Spring. Personally, I'm yet to find a solution that cannot be solved using a simple servlet framework. So, my question is: is your use of EJB really justified? Dont you notice that you can do an equal or even better and simpler job by using a web framework with pure java and hibernate? I look forward to your responses.

For unit testing, try testNG, it builds on JUnit but addresses its shortfalls. A lot of gurus already advocate it. Cheers.

ok bioye i have gat a couple of months Tapestry experience behind my belt and i think its really awesome.
I have used Struts and Spring MVC extensively you just can't fault the joy you derive when you use Tapestry.
Well i still use Spring for the IOC and use Hivemind only to inject Tapestry services like State and Engine services. While hivemind is good for simple IOC , Spring is more roboust especially when you are working with transaction management , Caching and Aspect Oriented Programming.
AlsoWhat Hivemind doesnt give you which spring does is forcing you to programm to interfaces.


ok so you normally use Tapestry(Pages) - Java classes- Hibernate

Ques: How do you handle Hibernate Lazy Loading Exceptions in your Tapestry pages 
I have heard about the Tapestry DataSqueezer but i havent used it . but if you were using Spring you have access
to the spring open sessionin view filter free of charge 



i have played around with TestNG and its cute good but i havent get a chance to use it on any project am currently involved in. In most cases Junit is just appropraite but you can also use DBunit for database testing and if you work in a team eviroment whereby your code depends on someone else code you can use Mocks a good framework that does this is JMock

@ sbucareer

You posts is so hard to read with that font

The fact about EJB is that a lot of enterprise apps dont really require the full stack of J2EE. That is why the light weight frameworks are more popular in use these days

EJB is most suitable if you want to build a full fledge distributed application . Most apps dont require these except the very complex ones

@candylips,

Yes, I'm the one from juglagos.  I'm trying out to see if I can remember you but I seem not to have enough clues!  About how I manage Hibernate's lazy loading,  I guess it might be due to a limitation of ognl/tapestry/hibernate integration.  Have you tried Tapernate? for me, I 'manually' load the object in the Page class before accessing it on the Page template.  This is a workaround and I still have to find time to find a proper solution.  If you do let me know.

@sbucareer,

From my little understanding,  J2EE consists of many technologies apart from EJB.  EJB is only used to handle business logic and business model.  This is what I'm replacing POJO for.  The other things you've been talking about are services, which can be managed for me by, say, Spring, Hibernate and the J2EE API's (minus EJB). 

Hence, I don't even need an Application Server.  All I need is a web server like tomcat or jetty.

Services like clustering/load-balancing, transactions/JTA, JMS, JMX, security, JNDI, connection pooling can be used alongside Tomcat, Spring, Hibernate and POJO's and you wont miss EJB at all.  In fact, it does NOT make sense to use EJB unless you are doing distributed objects or declarative security.

My points are justified when you realise that everything in the latest EJB/ J2EE spec were copied verbatim from Spring, Hibernate and Xdoclet - the tools I have been using for years instead of EJB/App Servers.  So, right now there is a more compelling reason not to use the new EJB because they probably will end up copying my open source tools 2 years from now!

if I'm wrong, i'm willing to learn

@bioye

Ok Tapernate should do the trick for lazy loading of objects. Checking it out i can see that you can configure open session in vew in the hivemind.xml file. i havent used it though cos i use Tapestry within spring and spring has its own implementaion of this pattern.

Well if you are manually loading your objects i guess you are working with a very simple object model or  you are probably not really utilizing the full power of Hibernate.

candylips:

if u remember the guy that used to send code to the list and tried to propose a project thats me.

I'm guessing you are Femi, the guy that did not like UML then. Do you like it now?

candylips:

Well if you are manually loading your objects i guess you are working with a very simple object model or you are probably not really utilizing the full power of Hibernate.

Wont say so. More like I'm forcing the loading of the object from Java code since ognl seems to fumble. Sample code here:


Session session = HibernateUtil.currentSession();
Query query = session.createQuery("from Employer e where e.status = :status ";
query.setString("status",ApplicationConstant.APROVAL_PENDING);
List list = query.list();
if(list != null)
{
setEmployers(list);
}


@sbucareer
first, my name is bioye. I know about CORBA and the history.

I know the difference between a web server and an app server. And I agree with you that EJB or use of an App Server is not appropriate for some projects. I was just asking if people are not over-engineering in certain circumstance just like candylips stated. Infact, if it works well for you and you have no complaints, please use it!


I stress - You only need EJB or an App Server when you need distributed components across multiple servers. So, my question is do you need this currently?

I differentiate between needing and using distributed components here. Some people probably use them even when they don't need them.

Now, be careful not to mix up EJB, J2EE and Application Server features here. EJB is just one feature of J2EE. The only reason you need an App Server is if you are using EJB. Every other container service can be gotten from a container like tomcat or common third-party tools.

Further, I address your laundry list of J2EE container services,

1. Persistence - I don't have to use EJB or app server here because I can do everything with Hibernate and Tomcat
2. Security - Actually, a lot of J2EE security is implemented in the servlet container. Role based and declarative security is available in Tomcat, JAAS is available in J2SE. I can apply servlet filters too and you can use Acegi with Spring for even more options. Spring allows declarative security at the method level using annotations.
3. Database Transanctions - You can use the JTA API from Tomcat and Hibernate comes with advanced Transaction management. Spring even does better including declarative transactions.
4. Thread Pooling - Tomcat has this via Apache Commons Pool
5. Messaging - Though, MDBs is the easiest way to implement messaging, it is the least flexible. I can do my messaging in a web app by calling JMS API directly (OpenJMS) and deploying in Tomcat. It gives me the most flexibility too. JavaSpaces is also an option.
6. Distributed Databases - Hibernate supports this via JTA and Transaction Managers. Similar to 3 [/b]above
7. [b] Maintenance - Not clear here but if you are talking JMX, you can manage components via Tomcat 100%! Also, good OO programming helps.
8. 2pc/ xa - Spring will handle this well along with Hibernate + XA driver. Never heard of 3pc commit tho. . Similar to 3 and 6[/b]above
9. [b]Thick client support - 3rd party tools like Caucho's Hessian and Burlap or GLUE; and I can use RMI with Tomcat

At least, Juergen Hoeller, one of the guys that wrote Spring would probably know better. Please, read his 3 posts on this discussion at the serverside and come back and comment:
http://www.theserverside.com/news/thread.tss?thread_id=15228#58871

EJB and App Servers are great but for me, the alternatives are easier, safer, more innovative and most important, free!

And one last thing, don't you think it's a little impolite when you use a word like [b]mourn [/b]when referring to my comments?

if I'm wrong, i'm willing to learn

@bioye

No thats not me . .

Looking at your code i have a few observations. firstly its not very good if you have your database code mixed with your  controller code. What you should have done is to abstract out the database specific code into a DAO layer, inject the DAO into your Tapestry pages and then call the required method before then setting the page method.

Looking at your code well. it should work if you are just loading the values in a single table . but when you have a foreign key relationship with many tables e.g lets say your employer table has a 1 - many relationship with employee table , and maybe a 1 - 1 relationship with Department table.

When you run that query in your code Hibernate should automatically look for all the related tables and populate your object graph. This normally fails if you have lazy loading setup and u do not have a kind of Open-session-in -view pattern and tells  hibernate to load the next object graph when it is required most of the time at the view-layer.

This is because the Hibernate session with the database would have been closed by then. But i guess this shouldnt be a problem if you use Tapernate. i just had a look at it yesterday and you can set up the OIV pattern in Hivemind.
This is the same logic i use the only difference is that i manage my Hibernate sessions in Spring

@sbucareer

To add to what bioye has said. and to repeat what i have said earlier. EJB is good but using it would limit you to an EJB  way of doing things. EJB is only good for doing a truly distributed application but majority of projects do not really require this .
So why should you  use EJB in a project when you are not building a distributed applicaition that absolutely requires the sending of objects across multiple platforms ,.
Just to let you know,  Spring for example has a whole lot more features out of the box which EJB can nto even dream of.

i think i have said these in my earlier posts but the Dependecy Injection provided by EJB 3.0 is vey basic and crap compared to what you get  with Spring.

Also, To do EJB 3.0 you need Java 5.0 and unforntuately it is not every body that has the luxury of Java 5.0 in the production environment. However Spring is compatible with 5.0 and earlier versions.

Based on these two facts alone a lot of developers would definately go for something else other that EJB 3.0

Also for your information the fact that you use EJB doesnt mean you will earn more money. I have been in the UK job market for a while now and worked on a lot of contract Jobs with many companies and i can tell you with absolute fact that most Organizations  dont really care less about EJB.

Even Investment banks that create real-time distribtued trading software use a combination of C++ using Extensive CORBA with Just pure core java. Spring is also used extensively to help structure application code.

always try to get your facts right before you make comments please

@Candylips,

Looking at your code i have a few observations. firstly its not very good if you have your database code mixed with your controller code. What you should have done is to abstract out the database specific code into a DAO layer, inject the DAO into your Tapestry pages and then call the required method before then setting the page method.

I agree with you that DAO's are neater. But you really derive the most value from them when you are on very large projects and you do a lot of maintenance. Following my KISS principles, I would only refactor during maintenance when modularity will ease headaches. However, don't worry, I'll soon move totally to the DAO approach. Meanwhile, for curiosity's sake, can you post an example of how you accessed an object graph from a tapestry page class via a DAO?

Looking at your code well. it should work if you are just loading the values in a single table.

Like I told you, it's just a workaround. I actually have relationships that span graphs of objects in my code. I'm not even sure if it works so well. Even when the tapestry page fails to load the object, a refresh solves the problem. However, if you get a good solution to the problem, please let me have it. I must agree with you that some of my code needs refactoring though. I'm guilty of that and there is a reason why it's so - a steep deadline!

candylips:

@bioye

Yea thats me. But i don't think i ever gave the impression that i didnt like UML. Basically you guys never gave the project a chance so it died a natural death.

Common. I don't think you should blame 'we guys'. I posted a UML model of the HR solution and you outrightly disagreed with my approach saying we should go straight to code. I responded that you should post some code to get us started since I couldnt figure how we could collaborate without a proper design to refer to. I But you never posted any code! We let you take the lead but you dissappointed us. And I couldn't just continue because you seriously disagreed with my approach.

Anyways, that's in the past and no hard feelings. We can forget about all that now.

bioye:

@Candylips,

I agree with you that DAO's are neater. But you really derive the most value from them when you are on very large projects and you do a lot of maintenance. Following my KISS principles, I would only refactor during maintenance when modularity will ease headaches. However, don't worry, I'll soon move totally to the DAO approach. Meanwhile, for curiosity's sake, can you post an example of how you accessed an object graph from a tapestry page class via a DAO?

Well thats easy, you need to define your DAO in Hivemind or Spring and if you want to access the object in your Tapestry page you add something like this in the Page specification.

<inject property="employeeDAO" type="spring" object="employeeDAO"/>

your page class will have

public abstract IEmployeeDAO getEmployeeDAO();

or if you are using annotations

@InjectObject("employeeDAO"
public abstract IEmployeeDAO getEmployeeDAO();

Well my example uses spring i guess if you are configuring your DAO in Hivemind the type attibute should be object instead of spring

Thats it ! you can now access your DAO methods in your Pages .

bioye:

Like I told you, it's just a workaround. I actually have relationships that span graphs of objects in my code. I'm not even sure if it works so well. Even when the tapestry page fails to load the object, a refresh solves the problem. However, if you get a good solution to the problem, please let me have it. I must agree with you that some of my code needs refactoring though. I'm guilty of that and there is a reason why it's so - a steep deadline!

Yea you would be better of refactoring you database logic into a DAO layer but a steep deadline is not an excuse 

bump

Bump