I.T. Go Home ....

SQL - optimize peformance by limiting table full scans

2016-01-27T00:44:00.001-08:00

There's an excellent article here about using the PIVOT/UNPIVOT keywords in Oracle to organize and report data:
https://dzone.com/articles/impress-your-coworkers-by-using-sql-unpivot

However, I think the most interesting message here is how to limit your full table scans and thus enhance the DB access performance. So, read on ! Thanks Lukas Eder for the great article !

Could you be .... the most beautiful bug in the world ....

2014-07-05T08:17:00.002-07:00

That's right. I've spent almost 3 weeks to fix a bug which has been haunting me all this time, but now it looks like the most "beautiful" I've ever encountered.

The context

We have an nUnit test for a medical streaming application that basically reads volumes (depicting body parts) from a disk, loads them in memory and them streams them to the rendering pipeline to be displayed on the screen.
The test consists of two parts:
1. initialization - where, among others, we force garbage collection, by calling GC.Collect(2)
2. the streaming itself, during which we perform some measurements; one of these is the number of generation 2 collections, which should stay on 0 for the whole duration of the streaming (we tolerate only generation 0 and 1 collections, but since generation 2 collections are expensive, we absolutely want to avoid them occurring during our streaming); let's call this measurement GC2; so we expect GC2 = 0

The issue I faced was that suddenly GC2 was 1, so the test was failing because of that.

The approach

First I used .NET Memory Profiler 4.5 to analyze my nUnit test. But before doing that, I needed to run nUnit in process, because otherwise it was starting the nunit-agent.exe process and then I was analyzing the wrong process. In order to run nUnit (version 2.6.2) in process, I changed the nunit-x86.exe.config like this:

<?xml version="1.0" encoding="utf-8"?>
<configuration>
<startup useLegacyV2RuntimeActivationPolicy="true">
<supportedRuntime version="v4.0.30319" />
</startup>

<runtime>
<legacyUnhandledExceptionPolicy enabled="1" />

<loadFromRemoteSources enabled="true" />

<assemblyBinding xmlns="urn:schemas-microsoft-com:asm.v1">
<probing privatePath="lib;addins" />
</assemblyBinding>
</runtime>
</configuration>

With this in place, I could analyze the nunit-x86 process and noticed the garbage collection pattern as in the figure below:

As I said, I knew that during the initialization phase we triggered several generation 2 collections and on top of those there was another one causing the test to fail. How to identify it ?
And then I've got this simple idea: I measure the number of collections of gen 2 at the beginning of the streaming (after all initialization has taken place) and then, for each sample streamed during the test, I count the number of gen 2 collections again and if it increased, I produce a beep:


g2 = GC.CollectionCount(2);
....
if (GC.CollectionCount(2) - g2 > 0) {
    Console.Beep(500, 200);
}

Yes, a simple beep would tell me when the extra gen 2 collection was produced.
With this simple approach, I analyzed the test by taking snapshots of memory immediately before and after the beep. Namely, I've used the following tools:
- .NET Memory Profiler 4.5
- CLRProfiler
- ANTS
- WinDbg
- PerfView (with this one I could inspect memory dumps that I generated from the code)

All these tools showed me the same thing: there was no suspicious garbage that could have triggered a new gen 2 collection. So what was causing this extra collection after all ?

The other approach - back to basics

After many hours of investigating where could the garbage come from and not being able to find anything, I decided to get back to basics: I started the good, old perfmon and then it struck me - I noticed the "# Induced GC" counter under ".NET CLR Memory" category and I though let's give it a try.
I ran nunit under perfmon and when I heard the "beep" (see above), I also saw the "Induced GC" growing by one. So that was it ! The generation 2 collection was induced, it wasn't due to any extra garbage.
The next step was to find out who induced it. For this I used JetBrains dotTrace Performance 5.3.1 and this is what I've discovered:

Thus, the induced gen 2 collection is called from Microsoft code, namely from a GDI Bitmap method.
This discovery finally led me to the solution for my bug, but I'll spare you the details ....

Conclusion

From the very beginning, I've been put on the wrong foot. I was trying to look for any suspicious garbage that would trigger my gen 2 collection, but I couldn't find any. I used many tools and I spent a lot of time to analyze this, without finding anything. Because there was nothing to find. My gen 2 collection was induced, i.e. forced by Microsoft code.
Getting back to basics proved to lead me to the solution: simple tools like perfmon can't be ignored. Actually, you should start with this, be it for the fact that you might find some hints. Armed with this knowledge, you can then use any other more complex tools to analyze it further.

A chat application with WebSockets, Java and jQuery

2014-06-05T14:18:00.001-07:00

I'm not going to describe the WebSockets protocol here - you can find zillions of tutorials online. In this short article I show how to build a simple chat application based on WebSockets.

The good news is that you don't have to worry about the nitty-gritty details of parsing WebSockets headers and packets, that's already implemented by open source libraries. On top of those you can write a server in Java, C#, Python, etc.
I use Tyrus, which makes it so easy to implement the server side of my chat application.
First I need to create a host for my endpoint:

package server;

import java.io.BufferedReader;
import java.io.InputStreamReader;
 
import org.glassfish.tyrus.server.Server;
 
public class WebSocketServer {
 
    public static void main(String[] args) {
        runServer();
    }
 
    public static void runServer() {
        Server server = new Server("localhost", 8025, "/chat", ChatEndpoint.class);
 
        try {
            server.start();
            BufferedReader reader = new BufferedReader(new InputStreamReader(System.in));
            System.out.print("Please press a key to stop the server.");
            reader.readLine();
        } catch (Exception e) {
            throw new RuntimeException(e);
        } finally {
            server.stop();
        }
    }
}

Look at the line where I create the Server object: I'll deploy ChatEndpoint to localhost:8025/chat.
Now I need to create the server itself - this ChatEndpoint type:

package server;

import java.io.IOException;
import java.util.Queue;
import java.util.concurrent.ConcurrentLinkedQueue;
import java.util.logging.Level;
import java.util.logging.Logger;
 
import javax.websocket.CloseReason;
import javax.websocket.OnClose;
import javax.websocket.OnMessage;
import javax.websocket.OnOpen;
import javax.websocket.Session;
import javax.websocket.CloseReason.CloseCodes;
import javax.websocket.server.ServerEndpoint;
 
@ServerEndpoint(value = "/test")
public class ChatEndpoint {
 
    private Logger logger = Logger.getLogger(this.getClass().getName());
    // keep all open WebSocket sessions (from all users)
    private static Queue queue = new ConcurrentLinkedQueue<>();

    @OnOpen
    public void onOpen(Session session) {
        logger.info("Connect with session: " + session.getId());
        queue.add(session);
        logger.log(Level.INFO, "Connected with " +  session.getId());
    }
 
    @OnMessage
    public void onMessage(String message, Session session) {
        logger.log(Level.INFO, "Mesage received " + message);
        try {
            // broadcast message to all open WebSocket sessions
            for (Session s : queue) {
             // include the original sender
             s.getBasicRemote().sendText(message);
             logger.log(Level.INFO, "Message sent: " + message);
            }
         } catch (IOException e) {
            logger.log(Level.INFO, e.toString());
         }
        // we could have a return here, in which case the returned string
        // would be the one sent from server to client,
        // as if doing s.getBasicRemote().sendText(message)
        //return message + "TEST";
    }
 
    @OnClose
    public void onClose(Session session, CloseReason closeReason) {
        logger.info(String.format("Session %s closed because of %s", session.getId(), closeReason));
        queue.remove(session);
        logger.log(Level.INFO, "Connection closed with " + session.getId());
    }
}

Tyrus makes things easy for us. Look at the annotation @ServerEndpoint(value = "/test") - here I'm saying that my server is accessible under the relative path "/test" under the URL mentioned before, thus the whole URL becomes: localhost:8025/chat/test.

We have three more annotations:
- @OnOpen to mark the method for opening the connection from a client to this server; here session uniquely identifies the connecting client
- @OnClose to mark the callback triggered when the connection with a particular client closes
- @OnMessage for the method that does the most important job: whenever a message is received from a client, this method broadcasts the message to all clients (including the sender)

That's all on the server side.
So now the client is a simple web page with some jQuery embellishments:

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html>
<head>
<script type='text/javascript' src='jquery.js'></script>
<script type='text/javascript' src='jquery-ui-1.8.23.custom.min.js'></script>
<script type='text/javascript'>
var wsocket;
function connect() {
wsocket = new WebSocket("ws://localhost:8025/chat/test");
wsocket.onmessage = onMessage;
alert("Connect");
}

function onMessage(evt) {
$("#chatText").append("\n" + evt.data);
}

$(document).ready(
function() {
$("#connect").click(
function() {
wsocket = new WebSocket("ws://localhost:8025/chat/test");
wsocket.onmessage = onMessage;
}
);

$("#send").click(
function() {
wsocket.send($("#username").val() + ":" + $("#tosend").val());
$("#tosend").val("");
}
);
}
);
</script>

<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
<title>Chat</title>
</head>
<body>
Username: <input type="text" id="username"><br>
<input type="submit" value="Connect" id="connect">
<input type="submit" value="Disconnect"><br><br>

<textarea rows="30" cols="50" id="chatText">
</textarea><br>
<textarea rows="4" cols="50" id="tosend">
</textarea>
<br>
<input type="submit" value="Send" id="send">

</body>
</html>

Note how we open the connection with the server: new WebSocket("ws://localhost:8025/chat/test").
Sending messages from the client to the server is simple too: wsocket.send($("#username").val() + ":" + $("#tosend").val());

Incredibly easy, isn't it ?
Oh, by the way, this is the list of jars I need for the application to compile (you can use Maven to simplify the deployment a bit):

Maven tips and tricks

2014-01-05T13:35:00.002-08:00

Maven is cool. Reading the official tutorial on the Maven website, you get a pretty good idea what it does and how it works. The tutorial fails to tell you though the most important thing: follow the Maven conventions and it will work fine; if you don't follow them, Maven won't help, better forget about it.
One of the Maven conventions is the standard directory layout (read here). As long as your project's directory structure looks like in the following figure, using Maven is ok:

Note that under the src/main you need to have your "java" directory for the java source code, "webapp" for the JSP/JSF pages (the eventual web application) and "resources" for some important files (read on, I'll give an example).

However, if you happen to develop a web application under Eclipse, your directory structure will be quite different:

Trying to use Maven now to build and deploy this Eclipse project is doomed to fail. The simplest solution to produce the war you need is to use Export from File menu under Eclipse and choose Web -> "WAR file".

For my own purposes I needed to use Maven, so I first moved the Eclipse project around in order to mimic the directory structure desired by Maven (as shown in the first figure above). From that point on, I learned some simple lessons:

1. as you've noticed, whenever you run a Maven phase/goal, Maven downloads all libraries necessary and places them under your local C:\Users\myUserName\.m2\repository; this is very powerful, as you can use Maven out of the box with a "tabula rasa" computer and end up with a neatly deployed project with all references solved

2. I wanted to add to my Maven project a dependency to an existing jar, so here it's what I've done:

mvn install:install-file -DgroupId=com.mycompany.test -DartifactId=testid -Dversion=1.0 -Dpackaging=jar -Dfile=/path/to/jarfile/f.jar

This resulted in copying the desired dependency, f.jar, somewhere under .m2/repository.

I added then the following to my pom.xml:
<dependency>
<groupId>com.mycompany.test</groupId>
<artifactId>testid</artifactId>
<version>1.0</version>
</dependency>

3. by the way, while running mvn package you'll most likely get error messages referring to missing classes; to fix this, just look for the missing class online, e.g.:

http://www.findjar.com/class/org/apache/openejb/jee/FacesConverter.html

Containing JAR files:

[MAVEN2] openejb-jee-3.1.1.jar
[MAVEN2] openejb-jee-3.1.jar

As you can see, findjar.com gives you links to Maven2 repositories. Following these links will tell you what artifactId and version you'll need to include in your pom.xml in order to fix this dependency.

Or even better, just go to the Central Maven Repository and look there for the missing class. Finally you'll get to the page that shows you exactly what you need to include in your pom.xml (you can copy-paste it):

4. if you're using Hibernate (high likelihood I'd say), add your hibernate.cfg.xml to /src/main/resources; moreover, all other Hibernate configuration files, e.g. Patient.hbm.xml, must be added to the corresponding /resources subdirectory; I needed to add all my .hbm.xml files to /src/main/resources/model; after you run mvn package, they end up in the war file under WEB-INF\classes\model

5. since you're packaging a web application, add the following to your pom.xml:

<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-war-plugin</artifactId>
<version>2.4</version>
</plugin>
</plugins>
</build>

------------------

The following is the pom.xml file I'm using for my project:

 <project xmlns="http://maven.apache.org/POM/4.0.0"

xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0
http://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0</modelVersion>
<groupId>com.mycompany.app</groupId>
<artifactId>app</artifactId>
<version>1.0-SNAPSHOT</version>
<packaging>war</packaging>
<dependencies>
<dependency>
<groupId>junit</groupId>
<artifactId>junit</artifactId>
<version>3.8.1</version>
<scope>test</scope>
</dependency>
<dependency>
<groupId>javax.servlet</groupId>
<artifactId>javax.servlet-api</artifactId>
<version>3.0.1</version>
<scope>provided</scope>
</dependency>
<dependency>
<groupId>org.hibernate</groupId>
<artifactId>hibernate-core</artifactId>
<version>3.6.3.Final</version>
</dependency>
<dependency>
<groupId>javax.faces</groupId>
<artifactId>jsf-api</artifactId>
<version>1.2_02</version>
</dependency>
<dependency>
<groupId>org.apache.openejb</groupId>
<artifactId>openejb-jee</artifactId>
<version>3.1.1</version>
</dependency>
<dependency>
<groupId>commons-logging</groupId>
<artifactId>commons-logging</artifactId>
<version>1.1.3</version>
</dependency>
<dependency>
<groupId>com.mycompany.test</groupId>
<artifactId>testid</artifactId>
<version>1.0</version>
</dependency>
</dependencies>
<build>
<plugins>
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-war-plugin</artifactId>
<version>2.4</version>
</plugin>
</plugins>
</build>
</project>

Enjoy Maven, folks !

Write Once Run Anywhere .... if you can

2013-10-27T06:42:00.001-07:00

Let's see. A few weeks back my hard-drive crashed and I needed to replace it. Luckily, I had saved my data previously, among which a Java-based application with 3 tiers: JSF, jQuery, servlets, and an open-source database. I wanted thus to resurrect this application. "Piece of cake" I told to myself, Java is after all WORA (Write Once Run Anywhere).
So this is what I've done:

downloaded and installed Eclipse Juno (which didn't start on double-clicking the exe, but rather using this command line: C:\eclipse\eclipse.exe -vm "C:\Program Files\Java\jre7\bin")

I want to run this application on Apache Tomcat and Glassfish application servers, so after downloading and installing both, I needed to download server adapters, as shown in the following figure (see "Download additional server adapters" in the figure below)

Note that I tried first with Glassfish 4.0.

3. I created a Dynamic Web Project in Eclipse and imported the sources of my project (saved from the previous hard-drive)

4. Then, as I was building the project, I noticed missing required jar files, so I started looking for them (either jarfinder.com or findjar.com) and adding them to \WEB-INF\lib

At this point in time I thought everything should work fine.

But first I encountered a strange run-time exception, as shown below:

5. After a bit of searching online, it turned out I needed to configure Project Facets to use Dynamic Web Project 3.0, as shown below:

Note: it's very important to select the right Project Facets, because that decides what jars will be used. Fro example, if you don't choose the JSF facet, then you need to add the JSF jars yourself; sometimes there's a good reason to do just that, e.g. the default provided jars won't do the job, so you want to use other jars.

6. As you can see, Eclipse complained about the Java version, so I needed to use the Java 7 compiler, as shown in the following figures:

7. After all this hassle, my application worked, but only with Glassfish 3.1; on 4.0 it throws an NPE; on Tomcat, no way to make it work, even though the same application previously worked ONLY on Tomcat and not on Glassfish ....

By the way, in the past I also had an issue with an application working fine on the former Oracle OAS and not working at all on Tomcat. It turned out that replacing the standard Oracle JSF libs with the ones from Mojarra helped. But you never know ....

I need to investigate this further, but now I'm happy at least I made it work on Glassfish 3.1.

Greetings from the jungle,

Sorin

P.S. A great article about JSF 2.0 with Glassfish and Eclipse is here: http://balusc.blogspot.nl/2011/01/jsf-20-tutorial-with-eclipse-and.html.

Sweat, blood and tears with JSF

2013-06-09T14:13:00.000-07:00

I'll start with the conclusion: if you're building standard, textbook web applications with Java Server Faces, you're going to love the technology. But the moment you start developing things that are a bit off the beaten track, the fun begins .... Let's take for example the calendar I presented recently. I want to add it some more functionality: when clicking on a certain day, you're redirected to a new page where you can create a new appointment. Once you confirm it, you'll be returned to the calendar page. So, this is how I extended the jQuery calendar:



dayClick: function(date, allDay, jsEvent, view) {
     if (allDay) {
        alert('Clicked on the entire day: ' + date);
        var url = "http://localhost:8080/Patients/appointments.jsf?day=" + date.getDate() + "&month=" + (date.getMonth() + 1) + "&year=" + date.getFullYear();    
        $(location).attr('href',url);  
     } else {
        //alert('Clicked on the slot: ' + date);
     }

}

Note that I've hardcoded the path to the appointment page ("appointments.jsp"). More importantly, I'm passing the date (day, month, year) as URL request parameters, e.g.: http://localhost:8080/Patients/appointments.jsf?day=5&month=7&year=2013

On the appointments.jsp page, one of the things I do is to print the chosen date:

<%
out.println("Data: " + request.getParameter("day") + "-" + request.getParameter("month") + "-" + request.getParameter("year"));
%>

Then, I've added a text input and a button, so that the user can fill in a patient name (or part of it) and click the button in order to search (in a DB) for all patients with the name containing the given text. The patients are supposed to be displayed in a selectOneListBox, as shown below:

<h:inputText binding="#{PatientsSearch.inputText1}" valueChangeListener="#{PatientsSearch.lookForName}" immediate="true" id="inputText1"/>
<h:commandButton value="Patient search:" action="#{PatientsSearch.getPatientsByName}">
</h:commandButton>

<h:selectOneListbox id="patients" value="#{PatientsSearch.selectedPatient}">
<f:selectItems value="#{PatientsSearch.patients}" var="p"
       itemLabel="#{p.name}:#{p.phoneMobile}" itemValue="#{p.name}" />
</h:selectOneListbox>

Almost everything works fine, i.e. if you click the "Patient search" button, the listbox will indeed be filled in with the corresponding patients' names and thus you can select on of the names from the list. I'm saying "almost", because if you change your mind and look up another patient name, clicking the "Patient search" button doesn't produce any result (actually I debugged it and noticed that the handler PatientsSearch.getPatientsByName is indeed called, but somehow the listbox is not refreshed with the new names). If you don't select any name in the list and perform a new patient search, this will work (so clicking the button produces the desired result). It looks as if the listbox is not just a "select one", but "use only once": the moment you select an item from the list, you can't refresh it anymore ....
Anyway, I actually didn't like the list, so I've replaced it with a table:

<h:dataTable var="patient" value="#{PatientsSearch.patients}" style="width:300px;height:150px" rowClasses="oddRow, evenRow">
  <h:column id="name">
    <f:facet name="header">
    <h:outputText value="Name" />
    </f:facet>
    <h:outputText value="#{patient.name}" />
  </h:column>
  <h:column id="phone">
    <f:facet name="header">
    <h:outputText value="Phone" />
    </f:facet>
    <h:outputText value="#{patient.phoneMobile}" />
  </h:column>
  <h:column>
    <h:inputHidden id="patientId" value="#{patient.id}"/>
  </h:column>
</h:dataTable>

I want to show the patient's name in the first column and the patient's phone number in the second. Wait a minute, there are three columns .... The third column contains an inputHidden field, whose value is the patient's ID - patient.id. So, what's the deal with this one ? Well, when I click a row in this table, I want that patient to be selected. Since the patient's ID uniquely identifies each patient, I can't simply use the name (it might not be unique). So I'm going to keep the patient's ID in this hidden field and when the user selects a patient, this ID will be sent to the managed bean that takes care of creating the new appointment using the patient's ID. Compare this with the selectOneListbox above: when I was selecting a patient in the listbox, the selected Patient instance (as a Java type) was passed to the managed bean, as the field PatientsSearch.selectedPatient. So I'm losing this elegance when working with the table, but ok ....

Next, let's do what I've just said: when the user selects a patient from the list, the patient's ID needs to be sent to a backing bean (i.e. we need to preserve it in order to create the appointment later). Again, look how simple was to do that with the selectOneListbox: just add value="#{PatientsSearch.selectedPatient}" and you're done. This is the intrinsic functionality of the listbox. The table however is meant rather for listing data, such as the patients in our case, not necessarily for selecting it. After googling around for a while, I've arrived to the conclusion that all I need to do is:
1. use some jQuery/JavaScript functionality to select a row in the table (and highlight the selection)
2. use a JSF inputHidden field to keep the patient's ID of the selected row

These two steps are shown below. Note that I first reset the background of all "tr" (except for the first, which is the header of the table and thus excluded from any selection) and then I set the background of the selected row to orange.

After that I do step 2 - this single line of code cost me many hours of searches online: $("#j_id_id13\\:testId").val($(this).find('td:last').find('input').attr("value"));
Note that the id of the inputHidden is "testId", but JSF prefixes it with "j_id_id13" when generating the corresponding HTML input field - you can see that by showing the page source in the browser. "\\" are needed to escape ":". Then, $(this).find('td:last') means that I want to go to the last column (td) in the current row (this). Finally, I look for "input" (the HTML field generated) and read its "value" attribute - this is the selected patient's ID.
Another important note: I've displayed the value of the just set hidden field in an "alert", since viewing the browser source doesn't show the hidden field being updated.


$("tr").not(':first').click(
    function () {
      $(this).parent().find("tr").css("background", "");
      $(this).css("background","orange");
      $("#j_id_id13\\:testId").val($(this).find('td:last').find('input').attr("value"));
      alert($("#j_id_id13\\:testId").val());
    }
);

<h:inputHidden id="testId"/>

3. once I have selected the patient (a row in the table) and I click the button "Create appointment", I can retrieve the patient's ID from the hidden field with the following handler in the backing bean:

FacesContext context = FacesContext.getCurrentInstance();  
Map requestMap = context.getExternalContext().getRequestParameterMap();  
String value = (String)requestMap.get("j_id_id13:testId");
        
Patient patient;
if (value != null) {
    patient = getPatientById(value);
}
return "toCalendar";

Now I'm almost done. "Almost" again. Remember I'm on the page "appointments.jsp" and here I click "Patient search". What happens is that I'm basically sent back to the same page in the browser, but then the date line I've shown at the beginning will only display nulls:

<%
out.println("Data: " + request.getParameter("day") + "-" + request.getParameter("month") + "-" + request.getParameter("year"));
%>

That happens because when clicking the button, I'm actually creating a new request for this page, losing thus the URL parameters encoded from the calendar page. So, here's what I can do to preserve the date:
1. write the date into a cookie
2. use again hidden fields to pass the date around
3. do URL rewriting with an onclick JavaScript handler when clicking the "Patient search" button
4. use some form of browser storage (see HTML5, etc.)

And the ideas could just flow like crazy hadn't I found the simplest method:

<h:commandButton value ="Patient search:" action="#{PatientsSearch.getPatientsByName}">
    <f:param name="day" value="#{request.getParameter('day')}" />
    <f:param name="month" value="#{request.getParameter('month')}" />
    <f:param name="year" value="#{request.getParameter('year')}" />  
</h:commandButton>

Note here the EL expressions used to transmit the request parameters to the f:param and thus keep them in the request scope.
That's all folks ! Pffffffffff ....

P.S. I know you'll say there are many JSF plugins which can implement this easier, but sometimes I'm just too conservative ....

Calendar with data from database: jQuery + servlets + Ajax

2013-03-03T13:07:00.000-08:00

I have a Java application that reads appointment data from a database and then shows them in a Java Swing UI. Now I want to publish this application online.
After a bit of research, I've found this convenient jQuery calendar and decided to use it. The main question is then how to use my existing Java code that reads appointments and expose them to this calendar. So, what I've done:

1. created a Java servlet that reads the database

2. created the jQuery fullcalendar that invokes this servlet

Note here the "url: agenda", which basically means I'm calling the servlet deployed at /agenda.

3. then I realized that I wanted to be able to react to month/week/day changes: whenever I click the previous and next buttons, I should pass the new time period to the servlet, retrieve the new appointments from the DB and show them in the same calendar; this is how I've achieved this:

Unique in the world

2013-01-19T13:14:00.001-08:00

I remember when I started programming how nothing seemed to be impossible. I was completely ignorant to the fact that there were plenty of experienced developers out there who could do the job faster and better than me. I behaved as if I was the only programmer left on Earth.
The years passed and the lesson I've learned is that the freshness of the spirit should never get worn out. Never lose it. Do your work as if you are unique in the world. Don't think about the others, they don't exist - it's just you, and you can do whatever you want to do.

Another Romanian success story

2013-01-06T03:23:00.000-08:00

Hi all and a happy new year !

No, this post is not about Dacia (daciacars.co.uk), but the initial intention was to praise AllView (www.allview.ro), a Brasov-based company, which produces several types of tablets and smart phones. I thought it was great that a Romanian company was a market leader in Romania (according to some studies 40% of tablets sold in Romania are Romanian-made) and it is pushing its products to Germany, Italy, the Netherlands, etc.

However, according to this blog, the only Romanian-made thing inside these AllView tablets is the BitDefender antivirus (www.bitdefender.com), a true Romaian success story. The rest is "made in China".
By the way, a few years ago Microsoft bought GeCAD, another Romanian company which produces the popular RAV Antivirus.
And by the way again, do you know that the second most spoken language at Microsoft is Romanian ? That's what I call a success story.

Smart phone ? Yeah, right ....

2012-05-29T14:08:00.002-07:00

Last year I finally bought my smartphone - a shiny LG Optimus 2, dual core. I was positively impressed by the flashy look and feel, I loved the applications I could download for free from the Android store - albeit they were mostly games - and of course, I spent late nights playing Angry Birds ....

After a few days of over-enthusiasm, I came back with my feet on the ground. The first "ouch" was when I spent hours to find a simple driver which allowed me to connect my new toy with the laptop. After finally managing that, I got lost trying to copy some files from the laptop to the phone and the other way around. Finally, I got that easier by enabling the Bluetooth between the two devices.

Using GPS was a pain in the a$$ .... At the best I ended up spending huge amounts of money for the mobile Internet. Downloading and using the maps offline was not working properly.

Frustrated, I tried to use the camera - the photos were such a poor quality (bad contrast, ugly colors) that I simply needed to remove them without any regrets.

I've got to the conclusion my phone is too smart, 'cause I've never got to figure out its algorithm for sorting the SMSs. I'm still not able to find the latest one where it should be: on the top of the stack. I went so far that when I was receiving an SMS, I was removing the SIM-card from the smart a$$ and putting it in my plain old ugly Nokia 1110i, to be able to read that SMS.

The battery is another "great success story": just a few calls and SMSs each day and in two days the battery of my dear smart a$$ is gone .... The same usage pattern with Nokia 1110i takes close to eight days without recharge !

In short:

	*LG Optimus 2, dual core* *(bought in May 2011)*	*Nokia* *(bought in January 2007)*
*Battery*	2 days (if you’re lucky)	8 days
*GPS*	Useless	-
*Startup time*	~ 60 seconds	3 seconds
*Camera*	Useless	-
*SMS*	Mixes up SMSs, can’t find the latest	Good, logical sorting
*Price*	480 Euros	20 Euros

The latest experience was something reminding an opera buffa. I moved from the Netherlands to Romania and I adjusted the time (it's one hour later in Romania). It was the night of daylight saving time, so I knew the time would be set one extra hour ahead. I did that manually and I specifically instructed the smart a$$ to not automatically adjust the time settings. After that I set the alarm clock at 7 am and I went to bed. I woke up with the alarm ringing loud, I stopped it and - after the usual swears - I rushed out. But I got a hunch: why was it so dark outside ? I turned on the TV to notice it was 6 am. So my smart .... whatever .... managed to set the time again one hour ahead, even though I had specifically instructed it not to do that ....

Did I mention than when I change the countries, my smart .... thing .... changes the language automatically ? When I move among FR, DE, NL, UK, IT, ES, PT, RO, etc. this is usually not a problem, but what am I gonna do in China or Israel or some Arabic country ? I wouldn't even be able to set the language back to English.
Anybody needs a smartphone ?

Update 1: Some people would say "get an iPhone, man". Possibly, for what you pay it should be a class above the other smart a$$es.

Update 2: Smartphones are wrecking the planet ....

Java Pet Store 2.0 - the architecture (part 1)

2011-12-11T09:50:00.000-08:00

To my surprise, the Java Pet Store 2.0 page, even though recently updated, still doesn't mention much about the design of the application.
So here we go, below you find some diagrams and explanations over how it's built and how it works. I'll continue the story in future posts, here I start with the overview and show how the catalog gets displayed.

First, check out this part of the web.xml file:

<servlet>
        <display-name>ControllerServlet</display-name>
        <servlet-name>ControllerServlet</servlet-name>
        <servlet-class>com.sun.javaee.blueprints.petstore.controller.ControllerServlet</servlet-class>
    </servlet>
    <servlet-mapping>
        <servlet-name>ControllerServlet</servlet-name>
        <url-pattern>/catalog</url-pattern>
    </servlet-mapping>

It tells that the ControllerServlet needs to be used for URLs that match the pattern "/catalog".

For example, when following the link "Dogs" on the main page, we navigate to:

http://localhost:8087/petstore/faces/catalog.jsp?catid=Dogs
(I have Glassfish running on localhost:8087)

Since this matches the "/catalog" pattern, ControllerServlet.service() is triggered, as shown in Figure 2.

The ControllerServlet implements the FrontController J2EE pattern and delegates to various actions (see Figure 1 and Figure 2), depending on the servlet path. In this case the servlet path is "/catalog" and it maps to CatalogXmlAction. The latter reads the "command" parameter from the request and, because this one is "categories", it calls the CatalogFacade to get the categories via JPA (note in Figure 1 that "Category" is a JPA Entity).

Figure 1: The main entities involved in fetching the catalog

Figure 2: Displaying the catalog

The CatalogXml action, once it has the categories, it writes them to the response (HttpServletResponse) in JSON format (see below).

So how come, given this JSON format, the categories and the pets are so nicely displayed in the page (see Figure 3) still ? Well, this is done with a bit of Dojo/JavaScript magic.

[{"id":"CATS","catid":"CATS","name":"Cats","description":"Loving and finicky friends","imageURL":"cats_icon.gif",
"products": [{"id":"feline01","catid":"CATS","name":"Hairy Cat","description":"Great for reducing mouse populations",
"imageURL":"cat1.gif"},{"id":"feline02","catid":"CATS","name":"Groomed Cat","description":"Friendly house cat keeps you 
away from the vacuum","imageURL":"cat2.gif"}]},
etc.
]

Figure 3: The Pet Store application - after choosing the categories

Take a look at catalog.js, which is used in catalog.jsp to format the page:

function loadAccordion () {
        // go out and get the categories
        // this should be made more geric
        var bindArgs = {
            url:  applicationContextRoot + "/catalog?command=categories&format=json",
            mimetype: "text/json",
            load: function(type,json) {
               ac.load(json);
               processURLParameters();
             },
             error: ajaxBindError
        };
        dojo.io.bind(bindArgs);
    }

This method is called from initCatalog() in catalog.js, which in turn is called in catalog.jsp:

<script type="text/javascript">
    dojo.event.connect(window, "onload", function(){initCatalog();});
</script>

Finally, loadAccordion() calls ac.load(json) and ac.showCategory(params.catid), where "ac" is AccordionMenu() (the menu with categories in Figure 3), which again, sends us to:

this.load = function(lcategories) {
        categories = lcategories;
        // create all the rows
        for (var l=0; l &lt; categories.length; l++) {
            var row = createRow(l,"accordionRow", ITEM_HEIGHT);
            createLinks(row.div, categories[l].name, l, "accordionLink");
            divs.push(row);
        }
    }

Here, createLinks() makes the links on the left-hand side menu ("Cats", "Dogs", etc.), using categories[l].name as parameter. "Categories" come from the JSON shown above, retrieved in loadAccordion(): url: applicationContextRoot + "/catalog?command=categories&format=json".

this.showCategory = function(catid) {
        for (var l=0; l < categories.length; l++) {
            if (catid == categories[l].name) {
                // now tell the scroller to load the first product
                initiateExpansion(l);
                if (categories[l].products[0]) {
                    dojo.event.topic.publish("/catalog", {type:"showProducts", productId:categories[l].products[0].id});
                } 
                break;
            }
        }
    }

Notice here "type" and "productId" labels. Back in catalog.js, we do:

dojo.event.topic.subscribe("/catalog", this, handleEvent);

So here we subscribe to the "/catalog" events with the callback "handleEvent", which does:

else if (args.type == "showProducts") {
          is.reset();
          populateItems(args.productId, 0, 0, true);
      }

Here, args.type is the label "type" we published above and args.productId is the label "productId". With the argument args.productId, "populateItems()" displays the images, etc. In order to do that, it uses DOM fields from catalog.jsp, e.g.:

var targetElement = document.getElementById("bodySpace");

To be continued ....

The computer and the cow

2011-07-25T04:36:00.000-07:00

Isn't it ironic that nowadays we try to specialize and super-specialize to learn how to program computers and send people into the space, while we should be learning how to milk the cow and grow tomatoes ?

MySQL Scalability

2011-05-22T02:28:00.000-07:00

When it comes to scaling MySQL (and in fact scaling DBs in general), Cal Henderson in his excellent book "Building Scalable Web Sites" makes a concise summary:

1. For limited nonlinear read capacity scaling, we can use MySQL's replication system

2. For architectural linear scaling, we can go:

a. vertical, and partition our table space into clusters, or

b. horizontal, and partition our data into shards

A bit more details follow.

1. With replication, basically data (e.g. the whole database) is replicated among multiple machines. This is done to have multiple sources to read from, i.e. it's a method used to enhance scalability of reads. For a typical web application, where the number of reads is 10-100 times larger than the number of writes, this might be good enough.

2. If we want good scalability of writes and reads, we need to split data into pieces, such that several write operations are possible in parallel (e.g. each write happening on a piece of data, independent from the other writes). We can split the data in two ways:

a. Clustering (vertical partitioning): each chunk of split data contains a subset of your tables; this chunk is called cluster and it can be organized as you wish: a single machine or again, data can be replicated (see 1) on multiple machines with master-slave or master-master relationships.

b. Federation (horizontal partitioning): each chunk of data contains a part of a table; for example suppose we split our table USERS by the username column: users go into different chunks based on the first letter of the username. Such a chunk of data and the machines behind them are called shards.

Pitfall: queries that need data in several shards

To avoid it: make sure that all records you need to fetch reside on the same shard. In some cases this is difficult, because one query will be able to use data on a single shard, but another query might involve the same data as the first, plus some other data on a different shard. A solution to this is denormalization: we store data on both shards (so we have duplication) in existing, normalized tables and in new, denormalized tables.

Of course another issue here is to synchronize the duplicated data, in order to avoid inconsistencies. But that's yet another chapter in the long history of designing for performance.

Eclipse tips and tricks (episode)

2011-04-02T11:24:00.000-07:00

Whenever you have a JSP page opened in Eclipse, the standard editor, called JSP Editor is not very fancy. However, do mouse right-click on the file name and choose Web Page Editor. The result is much more interesting, since it shows you the look and feel (Design tab) of the page as well as the source code. In the top-right corner of the window you have the Palette, which enables fast design by drag-and-dropping controls to the page. More to the right you have the Outline window, which allows you to select any tag of the page and add attributes to it.
This is similar to the powerful Oracle JDeveloper's editor.

Next figure illustrates the Web Page Editor in Eclipse.

Java Forever

2011-03-31T14:40:00.001-07:00

What else can I add ?

Threads and Web Applications

2010-12-09T14:34:00.000-08:00

Say you have a standalone (non-web) application where you want to show images. Suppose the images don't have dependencies among each other, so you can load and show each image in a separate thread.
Now say you want to design the same thing as a web application. To make things more concrete, suppose you want to build an application that shows a map of a region as a set (matrix) of tiles, each tile being displayed in a certain, predetermined part of the viewer. Let's call this application randomly 'google maps'.
So we'll have a servlet - say we build this with Java technology - that talks to a certain service from which it gets the corresponding tiles and then generates the page showing these tiles in their proper positions. It is tempting to take the same approach as in the standalone application and let the servlet run threads, each of them dealing with a tile. However, we know running threads from servlets calls for trouble (I'll write about that in separate posts).

So, what to do ? Well, the idea is to run the requests for the tiles in parallel as Ajax requests from - for example, a JavaScript piece of code residing in a JSP page. Each such Ajax request invokes the servlet to retrieve the desired tile. Thus, the servlet remains "pure", unpolluted with dangerous threads and we leave the parallelism and all the other optimizations in the hands of the web container.

Debugging Java PetStore with Glassfish and Eclipse

2010-11-21T13:56:00.000-08:00

Once I made PetStore work in Glassfish, I wanted to debug it in Eclipse.

1. I ran asadmin in my application server path:
C:\J2EE\Sun\AppServer\bin>asadmin start-domain

2. In the browser, I ran the application server Admin Console, available in my case at localhost:4848

3. I went to Application Server menu (top left), I chose JVM Settings and I enabled Debug

4. I restarted the Application Server (after enabling Debug and saving this configuration, I noticed the link "Restart needed" top left; I clicked on it, after that I ran asadmin start-domain from the console)

5. I also ran asadmin start-database to start the database for PetStore

6. I started Eclipse (I use version 3.4.1); under Run menu, I chose "Debug configurations ...."; then I chose the PetStore project (which I had created previously in Eclipse) and port 9009 (the default port where Glassfish is listening for debuggers); I also enabled "Allow termination of remote VM"

7. Just for the test, I put a breakpoint in ImageAction.java, method service; then I started the debug configuration created at bullet 6; finally, I navigated in a browser to
http://localhost:8087/petstore/faces/catalog.jsp
and I clicked a dog under Pets; I could see my breakpoint being hit.

Motivate me !

2010-10-20T13:56:00.000-07:00

"Give me a place to stand on, and I will move the Earth." (Greek: δῶς μοι πᾶ στῶ καὶ τὰν γᾶν κινάσω)
Archimedes

Let's take this scenario: you get hired by a seemingly good company, you get an apparently good job and you are very enthusistic about it. Slowly but surely the job becomes a demotivating and boring routine (pleonasm maybe, but I want to make it clear), and your enthusiasm drops. Sounds familiar ?

Think about it: we are all different, each of us has a special talent, some desires, some ideals. Wouldn't it be good to get a job in line with those ? A typical IT company employs people of different ages, backgrounds, nationalities, etc. Each of them is unique and can bring unique added value to the organization by employing her/his own, unique skills.
To me the ideal IT company shouldn't hire people to work on a given range of products. It should be the other way around: the products of the company should be tailored to the employees' desires, talents, affinities. Motivate the employees and surely you'll have a win-win situation. As a company you have to create products and services that your employees love to build/develop/provide. Give them motivation and they will move the Earth.

I have yet to find such a company. Maybe it hasn't been invented yet, but if you know of any which gets close to my dream, please let me know.

Making Java PetStore work

2010-09-05T13:18:00.000-07:00

Recently I've been struggling with making the Pet Store 2.0 application work.
So, shortly, what I've done:
- downloaded the jar from https://blueprints.dev.java.net/petstore/
- downloaded Glassfish v2.1.1 from https://glassfish.dev.java.net/

Then I tried to deploy and run the Pet Store by running the following commands:
- in the application server directory /bin:
asadmin start-domain
asadmin start-database
- in the Pet Store demo directory C:\J2EE\PetStore\latest\javapetstore-2.0-ea5:
ant setup
ant run

Even though all commands were successful, the application won't start. More specifically I got the error "The requested resource () is not available" when navigating to http://localhost:8087/petstore.

So, without further ado, these are the steps I've performed to make it work.

1. in C:\J2EE\PetStore\latest\javapetstore-2.0-ea5\bp-project\build.properties I corrected the following variables:
a. javaee.home to point to my application server: c:/J2EE/Sun/AppServer
b. javaee.server.username to point to my username (the one I used when I installed Glassfish)
c. javaee.server.passwordfile to point to ${javaee.home}/samples/passwordfile

2. I created the file "passwordfile" under c:/J2EE/Sun/AppServer/samples containing a single line:
AS_ADMIN_PASSWORD=myPassword (the password I chose when I installed Glassfish)

3. I also updated the file app-server.properties in C:\J2EE\PetStore\latest\javapetstore-2.0-ea5\bp-project with javaee.home=c:/J2EE/Sun/AppServer

At this point, I got the Pet Store demo running on the Glassfish server (after running again "ant setup" and "ant run"). However, only the first page was properly shown (http://localhost:8087/petstore/) and as soon as I navigated to "Enter the Store" (http://localhost:8087/petstore/faces/index.jsp), I saw a blank page.

Note: instead of running "ant run", you can also deploy the petstore.war from the /dist directory (or from /build) by using the Glassfish admin console:

By looking at the logging of the "ant setup" command I noticed that there was some issue with the username/password when trying to install the connection pool. I could also see this by inspecting the Glassfish log files (in the admin console, at http://localhost:4848/). So, I've done the following:

4. instead of relying on the ant script to install the connection pool and the data source (i.e. the ant setup command), I installed them by hand using the Glassfish admin console; see below the general properties of the connection pool:

Also, note the "Additional Properties":

Here I had to add the following properties:
- Password with the value APP (I know this from the file app-server.properties, see 3),
- DatabaseName with the value petstore (I found this by googling the web)

At this point if you click Ping under the General tab, you should get the message "Ping succeeded".

5. I also installed the data source (using the Glassfish admin console), which I called jdbc/PetstoreDB (I know this from the web.xml file):

Finally, I could navigate to http://localhost:8087/petstore/faces/index.jsp and got the Pet Store demo working properly ! (of course, don't forget to start the DB server and the application server by running:
asadmin start-domain
asadmin start-database )

The evil debugger

2010-08-20T14:14:00.000-07:00

Let's take a look at the following piece of C# code:

1:  using System; 
2:  using System.Collections.Generic; 
3:  using System.Linq; 
4:  using System.Text;


5:  namespace ConsoleApplication1 { 
6:    class Program { 
7:      private string myString = null; 
8:      public string MyString { 
9:        get { 
10:          if (myString == null) { 
11:            myString = "Bad string"; 
12:          } 
13:          return myString; 
14:        } 
15:        set { 
16:          if (myString == null) { 
17:            myString = value; 
18:          } 
19:        } 
20:      } 
21:      public override string ToString() { 
22:        return MyString; 
23:      } 
24:      static void Main(string[] args) { 
25:        Program program = new Program(); 
26:        program.MyString = "Good string"; 
27:        Console.WriteLine(program); 
28:      } 
29:    } 
30:  }

The result, as you would expect should be "Good string" written to the console. That is, when you run it from the command line.
Suppose I run it in Visual Studio .NET 2008 and I put a breakpoint at line 27 in the debugger. Then I debug the program, I reach line 27, I move to line 28 and the result is of course "Good string".
Let's put a breakpoint at line 26. You'd say "what does the breakpoint have to do with the result ?" Well, it does, because in this case, "Bad string" will be displayed. This is like Heisenberg principle (or the observer effect), isn't it ? When we debug, we actually influence the result. This is possible because when the breakpoint is set at line 26, MyString getter is executed and in effect myString is set to "Bad string". After that trying to set myString to "Good string" can't succeed because in the setter the test (myString == null) is false. At line 27 ToString() is called, displaying of course "Bad string".

The default behavior of the debugger in Visual Studio .NET 2008 is to evaluate the properties when the debugger hits them. Also, the ToString() method is called by default whenever you hover with the mouse pointer over an instance (like "program" in line 26).
You can disable these behaviors under Tools -> Options -> Debugging -> General.

It's good to know this, because as you can see, .NET properties and ToString() evaluation in VS.NET can give you lots of headaches.

Hibernate: OpenSessionInView reloaded

2010-08-15T07:51:00.000-07:00

In the previous post regarding Hibernate, I've shown how the OpenSessionInView pattern can be implemented.
In short, the idea was that the transactional statements (CRUD operations with the database) happen in between the following lines:

sf.getCurrentSession().beginTransaction();
....
sf.getCurrentSession().getTransaction().commit();

These lines are executed by a servlet filter, making thus sure that the Hibernate session is kept open until the view is rendered, and thus allowing us to use managed entities, as opposed to detached.

However, there is still something about which we haven't talked yet: what happens in the real case, with several simultaneous requests to the servlet ? We know that servlets are multithreaded objects, i.e. multiple threads may call the servlet's methods at the same time. We also know that servlets are not thread safe, so we have to pay attention to instance or class variables, which can be shared by several threads.

In the case of the OpenSessionPerView pattern, note that sf.getCurrentSession().beginTransaction() may be called by several threads. This means that the result of sf.getCurrentSession(), which is the current Session object, is shared among threads, which is dangerous since a Session is not a thread safe object. This asks for trouble, since the instance of Session is used to create queries and perform CRUD operations with the database.

So how can we ensure a thread-safe behavior of the OpenSessionPerView pattern ? ThreadLocal pattern to the rescue.
In a few words, what we need is a Session object per thread, which can be easily implemented using a ThreadLocal variable, as in the code below:

public class HibernateUtil {
    private static final SessionFactory sessionFactory;
    private static final ThreadLocal threadSession =
    new ThreadLocal();

    static {
      // initialize SessionFactory
      sessionFactory = new Configuration().configure().
      buildSessionFactory();
    }

    public static Session getSession() {
      Session s = (Session) threadSession.get();
      // open a new Session, if this thread has none yet
      try {
        if (s == null) {
          s = sessionFactory.openSession();
          threadSession.set(s);
        }
      } catch (HibernateException ex) {
        throw new InfrastructureException(ex);
      }
      return s;
    }

    public static void closeSession() {
      try {
        Session s = (Session) threadSession.get();
        threadSession.set(null);
        if (s != null && s.isOpen()) s.close();
      } catch (HibernateException ex) {
        throw new InfrastructureException(ex);
      }
    }
}

You get the picture. We would call HibernateUtil.getSession().getTransaction() in our servlet filter, instead of sf.getCurrentSession().beginTransaction(). Now each thread running through the servlet (and the filter) gets its own session object, so it performs the CRUD operations to the database without interfering with the other threads.

Invoke code on a certain thread in C# / .NET

2010-06-27T11:19:00.000-07:00

Virtually all GUI frameworks and the languages that come with them give you the possibility to invoke a certain piece of code on the UI thread. This is necessary because the UI components aren't usually thread safe (making all their methods thread safe would be a big blow to the performance of the application), so any update of the UI can't be done from another thread. That piece of code needs to be marshalled to the main thread.
.NET goes a step further and offers us the possibility to invoke a piece of code on any thread. The example bellow shows how this can be done.
ThreadTask will be run in a new thread, whose name is "Alpha thread". We create a Control on this thread (new Control()) and then we create its handle: ctrl.CreateControl().
Finally, we do Application.Run(), which makes our thread wait for messages in the message loop.
Then we use our control to invoke the method MyTestMethod on our thread.

Note that control handle creation is essential; without it, MyTestMethod would be invoked on the main thread. Also, attaching the thread to the message loop by doing Application.Run() ensures that the thread waits for invokes from other threads.

using System;
using System.Windows.Forms;
using System.Threading;

namespace TestThreads2 {
    public class TestThread {
        private Control ctrl = null;

        // This method that will be called when the thread
        // is started
        public void ThreadTask() {
            if (ctrl == null) {
                // create the control
                ctrl = new Control();
                // create the handle
                ctrl.CreateControl();
            }
            Console.WriteLine("ThreadTask is running on thread: " +
                Thread.CurrentThread.Name);
            // ensure the message loop is attached to this thread
            Application.Run();
        }

        public Control Ctrl {
            get { return ctrl; }
            set { ctrl = value; }
        }
    }

    public class Program {
        private static void MyTestMethod() {
            Console.WriteLine("TestMethod invoked on thread: " +
                Thread.CurrentThread.Name);
        }

        static void Main(string[] args) {
            //Control.CheckForIllegalCrossThreadCalls = true;
            TestThread myThread = new TestThread();
            Thread thread =
                new Thread(
                    new ThreadStart(myThread.ThreadTask)
                );
            thread.Name = "Alpha thread";
            thread.Start();
            Thread.Sleep(1000);
            if (myThread.Ctrl.InvokeRequired) {
                myThread.Ctrl.BeginInvoke(
                    new MethodInvoker(MyTestMethod)
                );
            } else {
                MyTestMethod();
            }
        }
    }
}

Check out here how to do the same thing in Java.

Invoke code on a certain thread in Java

2010-06-19T07:04:00.000-07:00

I think Java should be a bit jealous on the .NET's interesting feature of being able to invoke a piece of code on any given thread. In .NET that's possible by creating a Control (say myControl) on a certain thread and then using myControl.Invoke or myControl.BeginInvoke having as parameter the piece of code you want to invoke on that thread. The parameter is a delegate, another feature missing in Java, but let's ignore this for now.
In Java we have SwingUtilities.invokeAndWait and SwingUtilities.invokeLater to invoke code on the event dispatching thread, but not on any given thread. They take a Runnable as parameter.

So how would we implement such a feature in Java ? We'll need a thread that has nothing better to do but try to dequeue items from its queue and invoke them (the items would be the pieces of code that we need invoked on that thread). Something like this:

package threading;


import java.util.concurrent.BlockingQueue;
import java.util.concurrent.LinkedBlockingQueue;


public class Invoker {
    
    private static Thread myThread;
    private static BlockingQueue<Task> queue;
    
    private Invoker() {
    }
    
    public static void BeginInvoke(Task task) {
        if (myThread == null) {
            queue = new LinkedBlockingQueue<Task>();
            myThread = new MyThread(queue);
            myThread.start();
        }
        if (Thread.currentThread() != myThread) {
            try {
                queue.put(task);
            } catch(InterruptedException e) {
                e.printStackTrace();
                Thread.currentThread().interrupt();
            }
        } else {
            task.perform();
        }
    }


}

And the actual thread (I simulate the .NET delegate by using Java anonymous classes):


package threading;


import java.util.concurrent.BlockingQueue;


public class MyThread extends Thread {
    private final BlockingQueue<Task> taskQueue;
    
    public MyThread(BlockingQueue<Task> queue) {
        taskQueue = queue;
    }
    
    public void run() {
        while(true) {
            try {
                Task currentTask = taskQueue.take();
                currentTask.perform();
            } catch (InterruptedException e) {
                e.printStackTrace();
                // Restore the interrupted status
                Thread.currentThread().interrupt();
            }
        }
    }


}

We need an interface for our task - either an existing one (Swing uses Runnable), or a new one, like:

package threading;


public interface Task {


    void perform();
}

Finally, the test class with the main method:


package threading;


public class Test {


    public static void main(String[] args) {
        System.out.println("main thread: " + Thread.currentThread().getId() + " " + Thread.currentThread().getName());
        Invoker.BeginInvoke(new Task() {
            public void perform() {
                System.out.println("perform task on thread: " + Thread.currentThread().getId() + " " + Thread.currentThread().getName());
                
            }
        });
        Invoker.BeginInvoke(new Task() {
            public void perform() {
                System.out.println("perform task again on thread: " + Thread.currentThread().getId() + " " + Thread.currentThread().getName());
                
            }
        });
    }
}

It should come as no surprise the output:

main thread: 1 main
perform task on thread: 7 Thread-0
perform task again on thread: 7 Thread-0

Mashup with Google Maps or How to Call Java from JavaScript

2010-06-07T13:20:00.000-07:00

You can simply embed Java code with the usual notation and even expressions (< % = expression % >) in JavaScript used in JSP.

Below is an example of how to do that. The code creates some Java arrays (between <% and %>) and uses them in Javascript in order to show markers on a Google map. You can see it in action here (choose trip Customizer and play with it).

The code talks for itself, so here it is:

<HEAD>

   <title>Customize you trip !</title>
     <script src="http://maps.google.com/maps?file=api&amp;v=2&amp;key=ABQIAAAAyBigJTywiIxNM2cWUWhcFxR-sC-25ycSkOaRk1LgFwDltfgq9BRvhFBXWyducVZRhIvJw2DLcsPUdg"
      type="text/javascript"></script>
    <script type="text/javascript">


    //<![CDATA[


    function load() {
      if (GBrowserIsCompatible()) {
        var map = new GMap2(document.getElementById("map"));
        map.setCenter(new GLatLng(45.29, 25.56), 6);
        var coords = new Array();


<%
    int[] coordsX = (int[])(request.getSession().getAttribute("coordsX"));
    int[] coordsY = (int[])(request.getSession().getAttribute("coordsY"));
    String[] locations = (String[])(request.getSession().getAttribute("locations"));
    if ((coordsX == null) || (coordsX.length == 0)) {
        coordsX = new int[1];
        coordsX[0] = 0;
        coordsY = new int[1];
        coordsY[0] = 0;
    }
    for (int i = 0; i < coordsX.length; i++) {
        if (
            (coordsY[i] < 2967) && 
            (coordsY[i] > 2021) && 
            (coordsX[i] < 4823) &&
            (coordsX[i] > 4362)
        ) {
%>
          var coord = new GLatLng(
                <%= (float)coordsX[i]/100 %>, 
                <%= (float)coordsY[i]/100 %>);
          var marker = new GMarker(coord);
          GEvent.addListener(marker, "click", function() {
            marker.openInfoWindowHtml("<b><%= locations[i] %></b>");
          });
          map.addOverlay(marker);
        coords[<%= i %>] = coord;
<%
        }
    }
%>
        var polyline = new GPolyline(coords, "#ff0000", 10);
    map.addOverlay(polyline);




        // define the crosshair tile layer and its required functions
        var crossLayer = new GTileLayer(new GCopyrightCollection(""), 0, 15);
        crossLayer.getTileUrl =  function(tile, zoom) {
        return "./include/tile_crosshairs.png";
        }
        crossLayer.isPng = function() {return true;}


        // Create a new map type incorporating the tile layer
        var layerTerCross = [ G_PHYSICAL_MAP.getTileLayers()[0], crossLayer ];
        var mtTerCross = new GMapType(
                                      layerTerCross,
                                      G_PHYSICAL_MAP.getProjection(), 
                                      "Ter+");
        map.addMapType(G_PHYSICAL_MAP);
        map.addMapType(mtTerCross);
        map.addControl(new GLargeMapControl())


        var mapControl = new GHierarchicalMapTypeControl();
        // Set up map type menu relationships
        mapControl.clearRelationships();
        mapControl.addRelationship(G_SATELLITE_MAP, G_HYBRID_MAP, "Labels", false);
        mapControl.addRelationship(G_PHYSICAL_MAP, mtTerCross, "Crosshairs");
        // Add control after you've specified the relationships
        map.addControl(mapControl);
      }
    }
    //]]>
    </script>
</HEAD>


In the code above, "map" is actually a div:






<div id="map" style="width: 700px; height: 400px"></div>






In case you're wondering how this mix of JSP, JavaScript and Java actually works, let's start with the JSP: as we know, JSP runs on the server side and gets translated into a servlet. Thus, the code between < % and % >, including the one inside JavaScript, is textually copied into the servlet code,  since it is just Java code. The resulting servlet will generate HTML, so it will also dynamically generate the JavaScript code (which you can see in your browser by viewing the page source). The generated JavaScript code will have the for loop expanded with actual values for the coordinates and it will execute on the client side, in the browser.

That's all, folks.

Hibernate: OpenSessionInView explained

2010-06-01T13:02:00.000-07:00

When you enter the Hibernate world, you hear a lot about sessions, transactions, session-per-request, etc., to the point you start mixing the concepts. Let's face it:

Fact 1: Hibernate is big

Fact 2: Hibernate is easy if you understand the fundamentals; everything else builds on top of those

I'll try to explain these fundamental concepts and then I'll cover the famous OpenSessionInView pattern with an example.

So, what's a session ? If I tell you "a session is a unit of work bla bla bla", you'll say "I've read that in about 1,200,000 hits returned by Google and yet I don't really understand the concept". Let me put it this way: from the developer's perspective (what can I do, I usually incarnate this perspective), a session is actually an API that abstracts the notion of persistence service. Thus, whenever you want to do something with the objects (representing the O in the ORM), like synchronize them with the DB, persist them to the DB, etc., you need a session instance.

However, as we all know, each operation to the DB - be it a read or a write - needs a transaction. Thus, because you use the session in order to perform DB operations, it means whenever we're using the session we need a transaction (with a noteworthy exception, but we'll cover that later). If you're trying to run an operation on the session without being inside a transaction, you'll get a runtime exception of the type org.hibernate.TransactionException: Transaction not successfully started.

In other words, it's quite logical that the scope of a session (i.e. from the moment you create/open it to the moment you close it) and the scope of a transaction coincide. Without further ado, this is the session-per-request pattern.

Now let's take a painfully real example. Suppose I have two tables in my DB: a Patients table and a PatientPhones table (basically each patient from the first table has a number of telephone numbers recorded in the second table). They are linked by PATIENT_ID, a foreign key in PatientPhones table and primary key in the Patients table.

The Hibernate file that describes these tables is:

<?xml version="1.0"?>

<!DOCTYPE hibernate-mapping PUBLIC

"-//Hibernate/Hibernate Mapping DTD 3.0//EN"

"http://hibernate.sourceforge.net/hibernate-mapping-3.0.dtd">

<hibernate-mapping package="model">

</id>

</bag>

</class>

</hibernate-mapping>

Notice the lazy="true" attribute above ? This means that actually the phone numbers will be retrieved from the DB lazily, i.e. when we request them. For example, suppose I do this:


Session session = HibernateUtil.getSessionFactory().getCurrentSession();
session.beginTransaction();

// note that in this query we use the name of the class, Patient2,
// rather than the name of the table (Patients)
ArrayList patients =
new ArrayList(session.createQuery("from Patient2").list());
session.getTransaction().commit();
HibernateUtil.getSessionFactory().close();

Then, after the session has been closed, I do this:


for (Patient2 p : patients) {
System.out.println("Patient " + p.getName() + " " + p.getId());
}

This will work, because the patients list has been eagerly populated with the DB data inside the transaction above.

But if I try:





for (Patient2 p : patients) {
  if ((p.getPhoneNumbers() != null) && (p.getPhoneNumbers().size() > 0)) {
    System.out.print(" * phones :");
    Iterator iter = p.getPhoneNumbers().iterator();
    while (iter.hasNext()) {
        System.out.print(" " + iter.next());
    }
    System.out.println();
  }
}

This fails with the error LazyInitializationException (or something like this) because in the xml file above I asked the phone numbers to be lazily retrieved, i.e. when I do p.getPhoneNumbers(). However, at this moment I am outside the session, so I'm working with the detached object p. In order to be able to retrieve the phones, I have 2 choices:

1. either do lazy = "false" in the xml above

2. or reattach each object p to which I'm asking getPhoneNumbers(), like this:




Session session = HibernateUtil.getSessionFactory().getCurrentSession(); 
session.beginTransaction(); 
for (Patient2 p : patients) {     
 System.out.println("Patient " + p.getName() + " " + p.getId());     
 session.lock(p, LockMode.NONE);     
 if ((p.getPhoneNumbers() != null) && (p.getPhoneNumbers().size() > 0)) {         
  System.out.print(" * phones :");         
  Iterator iter = p.getPhoneNumbers().iterator();         
  while (iter.hasNext()) {             
   System.out.print(" " + iter.next());         
  }         
  System.out.println();     
 } 
} 
session.getTransaction().commit(); 
HibernateUtil.getSessionFactory().close();

This situation is extremely relevant, because in a typical web application you'd do exactly this:

whenever you show the results of a DB query (via Hibernate) in your view (a jsp page for example), your objects are detached (the session is closed, the transaction committed), so you can get the exception LazyInitializationException.

The solution is the Open Session in View pattern, which simply states that the session should remain open and the transaction uncommitted until the results have been processed in the view. This can be easily done with a filter, e.g.:

public class HibernateSessionRequestFilter implements Filter {

private SessionFactory sf;

public void doFilter(ServletRequest request, ServletResponse response,
FilterChain chain) throws IOException, ServletException {

try {

sf.getCurrentSession().beginTransaction();
// Call the next filter (continue request processing)
chain.doFilter(request, response);
 // Commit and cleanup
 sf.getCurrentSession().getTransaction().commit();
} catch (StaleObjectStateException staleEx) {
 throw staleEx;
} catch (Throwable ex) {
 // Rollback only
 try {
  if (sf.getCurrentSession().getTransaction().isActive()) {
  sf.getCurrentSession().getTransaction().rollback();
  }
 } catch (Throwable rbEx) {
  }
  // Let others handle it... maybe another interceptor for exceptions?
  throw new ServletException(ex);
}
}

public void init(FilterConfig filterConfig) throws ServletException {
sf = HibernateSession.getSessionFactory();
}

(NOTE: this pattern only works when the jsp and Hibernate code run on the same JVM, i.e. when the presentation and business logic tiers run on the same machine)

I will register this filter in my web.xml:

<filter-name>HibernateFilter</filter-name>

<filter-class>view.backing.HibernateSessionRequestFilter</filter-class>

</filter>

<filter-mapping>

<filter-name>HibernateFilter</filter-name>

<url-pattern>/*</url-pattern>

</filter-mapping>

Now, in one of my JSF backing beans, PatientsSearch.java, I could have:





public void getAllPatients() {
Session session = null;
try {
 session = HibernateSession.getSessionFactory().getCurrentSession();
 Query query = session.createQuery("from Patient2 p order by p.name asc");
 int pageNumber = 1;
 int pageSize = rowsPerPage;
 query.setFirstResult((pageNumber - 1) * pageSize);
 query.setMaxResults(pageSize);
 patients = new ArrayList(query.list());
 totalRows = count();
} catch (Exception e) {
 if (session != null) session.getTransaction().rollback();
 e.printStackTrace();
}
}

And in the view, patientSearch.jsp, I could have:

<h:dataTable value="#{PatientsSearch.patients}" var="patient"


binding="#{PatientsSearch.dataTable1}" id="dataTable1"
border="2" bgcolor="White" frame="border"
">
<h:column>
<f:facet name="header">
<h:outputText value="Name"/>
</f:facet>
<h:commandLink action="#{PatientsSearch.edit}">
<h:outputText value="#{patient.name}" style="font-family:Garamond;"/>
</h:commandLink>
</h:column>
<h:column>
<f:facet name="header">
<h:outputText value="ID"/>
</f:facet>
<h:outputText value="#{patient.id}"/>
</h:column>
<h:column>
<f:facet name="header">
<h:outputText value="Address"/>
</f:facet>
<h:outputText value="#{patient.address}"/>
</h:column>
<h:column>
<f:facet name="header">
<h:outputText value="Fix phone"/>
</f:facet>
<h:outputText value="#{patient.phoneFix}"/>
</h:column>
<h:column>
<f:facet name="header">
<h:outputText value="Mobile phone"/>
</f:facet>
<h:outputText value="#{patient.phoneMobile}"/>
</h:column>
</h:dataTable>

<h:commandButton value="All patients" id="commandButton1"
action="#{PatientsSearch.getAllPatients}"/>

This works just as described above: in the view I don't need to worry about the session; even if lazy = "true", I know that the calls patient.phoneMobile and patient.phoneFix will retrieve the data, because everything happens between the two important statements in the filter:

sf.getCurrentSession().beginTransaction();
....
sf.getCurrentSession().getTransaction().commit();

By the way, in Patient2.java, I have:

public String getPhoneFix() {

    if (phoneNumbers.size() > 0) return (String)phoneNumbers.get(0);

return null;

}

Simple, isn't it ?

However, there is potentially something wrong with this approach. Check out the next article to see what and how we can correct it.