C Popp

You can try wrapping the Image object in an ImageIcon object which does implement Serializable.

http://download.oracle.com/javase/1.5.0/docs/api/javax/swing/ImageIcon.html

Google Wave isn't the only Google application utilizing the GWT framework. In one of the past Google I/O conferences (I can't recall the specific one), the speaker gave examples of how GWT is used for the Google ad words management tool. This shows GWT is embedded into a core business application so will likely continue to see some level of support.

Yikes.. it's difficult to say what the exact source of your problem is without more information.

I would suggest a couple of places to look.

Are all your objects being returned by the RPC call serialized?

In my project we found the best type of objects to pass to the client are just plain old java objects (POJO's). If you can boil all your class fields down to primitives like String or int it works even better.

Remember the only thing you should be passing down to the client is the absolute minimum information needed for the Presenter (if you are using the MVP pattern) to render the display. Any business logic or complex calculations should be done on the server before generating your response.

On the server side, you can continue to use whatever approach you currently use. Once you start connecting your GWT client with your server code you will need to use a RPC service.

This link to the google docs should help get you started.
Making Remote Procedure Calls

Intro to GWT on NetBeans

The GWT compiler generates 6 different permutations of the javascript file. You can modify the default setting to create fewer permutations.

Compiling a GWT Application

It does take into account the differences in various browsers but you might still have to do some tweaking on the CSS styling especially if you are using some of the newer CSS styles formats.

GWT does not require an advance level of java knowledge, but you do need to have some knowledge of java GUI development (unless you use the ui xml pattern which requires some knowledge of html). You should also have some idea of how web technologies and ajax works.

All RPC requests are handled as separate transactions. There isn't any built in caching of serialized objects. If you know a List is not going to frequently change or is static, you can persist that object on the client side after it's retrieved from the server. You can store it in the client session and if the List is called again, just add a method to check if the object is in the session first before making a RPC call back to the server.

I like to run my JUnit tests as a small application. I have result processing classes and jsp's to display results. You can use pojo's and beans for tracking different aspects of the tests being run. If you need a dependency, you can try using a hashtable or map with the key = testname and value = pass/fail. Then it's pretty trivial to do a key lookup as part of a test setup with logic to handle if the key value is pass or fail.

If this was a job interview question, there probably wasn't one right or wrong answer. The interviewer was probably looking at your thought process. He/She would be more interested in how you approach the problem, what past experiences you might have had, and how quickly you can think on your feet. When I've conducted job interviews for test positions, you would be surprised at how many people just freeze and have no clue what to do if you start talking about testing without a script.
[ November 19, 2007: Message edited by: C Popp ]

Hi,

I believe this would be a starting point:

From the error log:
The /WEB-INF/web.xml was not found.

I would verify your struts and web app environment are configured and setup correctly. It looks like the web.xml file is having an issue.

Originally posted by Robertovich Wartins:
thank for response,

i have created hibernate classes for some database table and i want test those classes (test operations like insert, select, update and delete). Should i put all those tests in one test case class (in separated methods) or for any of those tests create new test case.

I like to make my tests as small and modular as possible. I end up with more code, but I believe this makes it easier to find debug errors and maintain tracability from tests back to requirements.

I've never used Coverlipse, but Agitar is just amazing. I agree with the previous poster, it's fun to use.

I would probably start with a combination of Selenium and AutoIt. I would use Selenium to drive the GUI and AutoIt for screen recognition and testing. AutoIt has some easy to use functions for defining a region of a screen, watching for changes in the region, and comparing to user variables. AutoIt makes it easy to find the checksum of a region and using that checksum value for other script controls.