Java

GWT with Cypal Studio

GWT 1.5 (official version is 1.5.2) is now finally released and not longer a release candidate. GWT is very nice as AJAX toolkit for a java programmer. You code your application in pure Java and all the JavaScript stuff is done behind the scenes. Important is to understand the build process. The Java Source Code is translated by the GWT Compiler to JavaScript Code. This GWT Compiler understands Java 1.5 with limited API. Eclipse has it’s own built-in Java compiler corresponding to the compiler level of the Java Project. Usage of unsupported Java functions will only be shown by the GWT compiler. Eclipse can only limit the access by setting access rules on the used JRE. In the project preferences under Java Build Path -> Libraries is the JRE. Under this Tree element is the access rules item. In the edit dialog can e.g. the usage limited to java.lang.* by setting “java/lang/*” as accessible. All other classes will not be available and cause compile errors in eclipse.

GWT ships with command line tools to create a new eclipse project including batch files to run the GWT compiler and launch files to start the hosted mode of GWT. Hosted mode starts a embedded tomcat and special GWT browser. The refresh button of this browser recompile the Java Code to JavaScript. Changes in the Java Code will directly shown in the browser after refresh. No restart is required.

GWT has two main concepts for a GWT application. The module defines the main class for a application and the AsyncCallbacks the service layer for accessing server side logic. The module name must be defined for the applicationCreator gwt command line tool as a java package name like com.mypackage.myapp.Main. Inside the source folder of the created eclipse project is the xml file with the gwt project definition and a Java class extending EntryPoint. The EntryPoint onModuleLoad method creates the UI of our GWT module. In the GWT docs online you will find a nice gallery with available ui elements called widgets. Remember that all code inside this method will be translated into JavaScript code.

Accessing business logic requires to define a interface extending RemoteService and a @RemoteServiceRelativePath annotation containing the name of the Service. This service must be implemented by class extending RemoteServiceServlet and implementing your interface. Additionally is a interface needed called <yourInterfaceName>Asyn, which contains the same signatures with a added AsyncCallback callback parameter e.g.

 
void sayHello(String text, AsyncCallback<String> callback);
final MyServiceAsync service = (MyServiceAsync) GWT.create(MyService.class); service.sayHello(textbox.getText(), new AsyncCallback<String>() { public void onFailure(Throwable caught) { Window.alert(caught.getMessage()); } public void onSuccess(String result) { Window.alert(result); } }); } });

So many steps has to be made by hand to code a GWT app in this way.

Cypal Studio is a open source eclipse plugin to extend web tools platform of eclipse with GWT. Just extract the install zip file to the plugins subfolder of eclipse 3.3. With Eclipse 3.4 (aka Ganymede) you have to use the dropins/<pluginname>/plugins folder. Cypal integrates as a facet for dynamic web projects. Best environment is a JRe 1.5, GWT 1.5 and a Tomcat 5.5 with this JRE as run time. Call the New Project wizard of Eclipse and choose Dynamic web project. Choose the Tomcat 5.5 as target run time, set web modules version to 2.4 and choose “Cypal Studio for GWT” as configuration. Change either the workspace or project JRE to 1.5. If not done before you must now define the GWT home folder. Give here the path to the extracted GWT 1.5.2 archive. Select under File->New->Other..->Cypal Studio->GWT Module. Give the Module a Java Package name and and a name. The EntryPoint class, the html and the gwt xml file will be created. Under Run->Run configurations you will find “GWT hosted mode application”. Click on the new button, select the Project and click on the run button. The application will be compiled by GWT and the internal browser of GWT started with your Module. Remote Services can be created with File->New->Other..->Cypal Studio->GWT Remote Service. Enter here a name and a uri. Advised is to use the same value for Name and service URI e.g. MyService. This creates a public interface, a async interface, a implementation and change your gwt xml file to include this service as a servlet. Remember to use gwt serializable objects as parameters and return value for your service methods.

Linux

Upgrading VMWare Server 1.0.6 to 2.0 beta 2

Seems to be an easy step to upgrade my local vmware server to the current version. VMWare server 2 changed the administration tool to a web interface like the one from the GSX or ESX server. First step is to uninstall the old vmware server and reboot. The download took a bit longer then before. Nearly 600mb comparing to 150mb for a 1.0.7 server install package for windows. You can change the ports of the administration web interface during installation. I leave this settings to default but changed that the server will not automatically started during windows boot. In the next screen is the possibility to change the default folder of vmware for images. I changed this to my existing folder of vmware server 1.0.6. Like before you have after installation a shortcut to the vmware server in the quicklaunch and on the desktop.

Firefox was a really a pain in the ass as browser for the web interface of beta 1 of vmware server 2. The download page now mention for Beta 2:

Firefox 3 as a supported browser for the management interface.

The icon open https://<hostname>:8333/ui/ as default url. Firefox 3 complains that the ssl certificate is self signed. You must click the link at the bottom to “add the exception” and the “add exception” button. The upcoming dialog request to get the certificate and let you after this download “confirm the security exception”.

Now the login screens ask for the credentials. Good question. The old native admin tool simply ask to login locally without any credentials. During installation where no user created or asked to define. Vmware use the local user database of windows as authentication provider. So define a new local user with admin rights or use a existing one to login.

My main vm image is a ubuntu server 8.04 with just a text console. Vmware use own linux modules for fast networking, folder sharing,… So i tried to upgrade my existing vmware-tools. Just start the vm and click the link to “install vmware tools”. It connects the virtual cdrom to readonly device with the install packages. Enter “mount /dev/cdrom /mnt/cdrom” inside the vm to copy install package to the tmp folder. You have to copy the *.tar.gz install package because ubuntu doesn’t work with RPMs like OpenSuse. The extracted folder contains the vmware-tools-install.pl script. Starting the install as root with ./vmware-tools-instal.pl or sudo ./vmware-tools-install.pl produce a error. VMware tools recognise the old kernel modules and stop the installation. The error message prints the name of the modules like vmxnet. So use “modprobe -v vmxnet” to get the path to the *.ko file. Delete it with rm and do this for all mentioned modules. Delete afterwards the old vmware settings by calling “rm -rf /etc/vmware*”. Now you can call the vmware-tools-install script to install successfully the kernel modules. Reboot the vm to use the new modules. Now you can see in the web interface e.g. the ip of the running vm.

Linux

Damn Small Screen

I switched recently on my home server from OpenSuse 10.3 to Ubuntu Hardy Heron 8.0.4. Main reason was the package management with zypper as backend. Each Distro use several Software Repositories with RPMs where you can search for new software to install on your machine. Normally between four to eight repo’s are configured independent which distro you use. OpenSuse become more and more a pain in my ass. Main reason was the more and more increasing time which yast needs to show me the available software packages. For a test drive is VMWare Server really nice. Just setup a new VM, install a distro inside and see if it fits your needs.

Under Ubuntu you must first decide if you want the desktop or server edition. On my notebook was the primary requirement to have a back end for application server like Archiva or Nexus. Both very comfortable maven proxies with a web frontend. So i decided to install the server edition. The server edition have no window manager and starts a linux just with a console login. So far so good. A little bit hidden is the possibility to change the default screen solution from 800×600 to a larger one. Scrolling log or config files is not that pretty under such conditions.

The solution is in the /etc/grub/menu.1st file. In the end of the file is a list of linux versions. Each runnable linux configuration has a title, root, kernel and initrd line. For a 1024×768 you can append to the kernel line

vga=0x317

according to this howto. Ubuntu has in the last weeks published a new service pack called 8.04.1. Normal updated are installed with a “apt-get update” followed by “apt-get upgrade” to install updates for installed packages. I saw messages about retained packages after the availibity of this new service pack. A “apt-get dist-upgrade” forces to install the retained packages including a new kernel. Remember the line with the kernel paramater in the grub menu file. The update updates this file as well. Grateful it notice my changes and ask me what to do. My first intention was to say ok overwrite my changes and i add that vga parameter as well. That works but it can be handled much smarter. Linux kernel updates can be happen more often.

The solution is to add the vga parameter to defoptions line instead of the kernel line. This line is normally uncommented with a leading # character. So add this line

defoptions=vga=0x317

This 0x317 is the hex definition. Also possible is to write vga=719 with decimal notation.

Java

dbunit and null values

For testing database based applications with more or less fixed schema dbunit is nice for creating test data. There are several ways of defining the data in xml and loading it into the db. With the FlatXmlDataSet it looks like

The first row defines the schema for dbunit with this format. When you have self referencing data with columns linked to the pk column it gets more complicated. Think of a growing company with locations all over the world. The hq will be the primary location and all other belong to their regional hq which belongs to the global hq. In the table is a column parent_location as fk to the id column. So the global hq has no parent and all other rows have parents. With the above sample data all rows has null as value in the parent column. DBUnit’s FlatXmlDataSet ignores all other column data in subsequential rows if they are not mentioned in the first row of a table. To avoid this behavior you must provide a dtd inline or external to define the columns:




]>

Now the first row has a null parent value and all others the value defined in the XML file.

Java

plugins disappear…

Eclipse 3.3 runs with Java 1.4 but recommended is Java 5. Every customer is different for us. Not every time is Java 5 or the current Java 6 the JRE for development / production. These technical requirements are not fix for all time. So i adjust my laptop settings when i start to work for a new customer. Under Windows is the PATH environment variable responsible to define which JRE is executed when you run a java application like the eclipse ide or ANT. It contains the bin folder of the JDK/JRE. If not defined then the java.exe in %SYTEMROOT%\system32 (e.g. C:\windows\system32) check the windows registry:

HKEY_LOCAL_MACHINE\Software\JavaSoft\Java Runtime Environment -> CurrentVersion (e.g. value 1.6)

For a quick check i open a dos shell via entering “cmd” in the execute textfield of windows and type “java -version”:

java version “1.4.2_16”

Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_16-b05)

Java HotSpot(TM) Client VM (build 1.4.2_16-b05, mixed mode)

These recommendation for Java 5 as runtime are valid as well for the plugins. Eclipse plugins define in the META-INF/MANIFEST file which runtime is minimum required:

Bundle-RequiredExecutionEnvironment: J2SE-1.5

The plugin with such a definition will not be activated during start up if the workbench jre does not fit the requirement. So far so good. I start a new workspace. Changed the workspace path to a new folder during startup. Eclipse looks ok on the first look but after a while you miss some functions. Lucky if some error dialogs pops up. You see normally only in the error log exception stacktraces containing lines with

java.lang.UnsupportedClassVersionError: …

The good thing is that eclipse can be configured to work in such conditions. Eclipse has command line arguments to define the workbench jre. So you have under windows to create a new link on the desktop to the eclipse.exe. Under options of this link is the full path to the eclipse.exe. Just add after a white space e.g. the following corresponding to your environment and paths:

-vm C:\java\jdk1.6\bin\javaw.exe -vmargs -XX:MaxPermSize=128m -XX:PermSize=128m -Xmx768M -Xms768M -Dsun.lang.ClassLoader.allowArraySyntax=true

This defines the workbench vm (-vm), vm arguments (-vmargs) like memory settings (-XX…) and a system property (-Dkey=value) e.g. to work properly with the bea plugin.

With these changed settings eclipse can start all plugins as expected. Functions, perspectives,.. formerly hidden appear in the workbench. No Magic but a little bit more messages in the ide would be nice for users… In the preferences of eclipse can under Java -> Installed JRE the default JRE be defined. Each project can define under project properties -> Java compiler -> compiler compliance level e.g. 1.4. So run ant inside a dos box to compile with java 1.4 (as defined for windows), start eclipse with java 6 and compile the project inside eclipse with 1.4. Very flexible but you need to know where to put the screw driver on.

tech stuff, development news…