Give it a read and let me know if you have any feedback.

• What are you using (FlexMonkey, FlexUnit, Mercury QTP, Selenium, RIATest, FunFX, etc)?
• Have you run into major pitfalls with your Flex testing?
• Are the tools you are using allowing your team to achieve your testing goals?
• How would you rate the state of testing in Flex development?

Thanks! I look forward to hearing from you.

FlexMonkey is an open source Adobe AIR application used for testing Flex and AIR applications.  It provides the functionality to record, playback, and verify Flex UI interactions.  It also generates ActionScript / fluint tests that you can easily include within a continuous integration environment.

Who is FlexMonkey for?

Like many in the Flex community, we came to Flex development after many years of developing traditional web applications, where there are multiple tools that allow a developer to simulate a UI client for testing (like HTMLUnit).  For testing our Flex user interfaces, we had a strong desire to continue using this paradigm for our developer level testing.

As software consultants, ensuring quality is an important part of what we do.  So, we created FlexMonkey to fill the gap between a unit test and what’s really needed to ensure quality (and protect against regression) when developing user interfaces.  While we initially created FlexMonkey for developers, many in the FlexMonkey community have found it just as valuable for use in creating QA / Functional tests.

So, we believe that FlexMonkey is useful to both software developers and quality control engineers on Flex development projects.

So, what’s so game changing about it?

If you ask Mike Labriola (one of the Fluint/FlexUnit 4 guys), ‘How do I unit test my Flex user interfaces?’  He will quickly respond, ‘You don’t.’  He won’t tell you this because he doesn’t believe in unit testing (as I can assure you that he strongly believes in Unit Testing), but because there are parts of a user interface that cannot be sufficiently tested with unit tests.

So, as we jump into the reasons why FlexMonkey matters for you, I’ll note that we do not believe that FlexMonkey replaces the need for traditional unit tests.  There are still parts of a Flex application that are best tested this way (e.g. service clients, business logic, etc.).  If you are building a framework, it’s likely that the majority of your testing will be done through unit tests, but if you are focusing on features and functionality you will likely be able to achieve upwards of 80% of your testing goals with FlexMonkey.  Nevertheless, here are the high level places where FlexMonkey changes the game:

1. FlexMonkey fills the gap left between unit testing and a truly successful test suite,  as it provides a tool for doing developer level functional tests that replace traditional unit testing activities that developers perform in other tiers of application development.
2. FlexMonkey allows developers to write their user interface code in the most natural way. Typically, when developing a user interface application without a tool to simulate a client of the running application, the code has to be written differently to support testing.  This often leads to many unnatural acts, which can be avoided when using FlexMonkey.

   For the Java folks, I like to bring up the SpringFramework here, as it provides us with a powerful example of a framework that largely succeeded because it allows developers to write their code in a natural way.  Over the long run, the frameworks and tools that make it possible for developers to work in the most natural way will always win out.

   Tools like FlexMonkey are able to facilitate this simply through the paradigm they provide for testing user interfaces, as they allow developers to test applications without any knowledge of the underlying code.

3. Cost. FlexMonkey is free.  So, even though it’s not a perfect tool, it does enable one to build and deploy a successful test suite, a task that is extremely difficult and costly in Flex development without FlexMonkey.

You can find the FlexMonkey project at: http://flexmonkey.gorillalogic.com/

FlexMonkey is built and open-sourced by Gorilla Logic, Inc.

Now, they are attempting to expand from just a framework to some sort of Flex MVC framework think-tank. Yakov Fain’s reaction to this is priceless:

First, let me ask Java developers a question. Imagine that one day you wake up and read the following announcement, “As of today, Spring framework is the foundation for delivering of successful J2EE projects. In contrast to earlier versions, many parts apply across frameworks. So, if you are using Struts, JSF, and especially Tapestry, just forget about all these complex to pronounce names. From now on, no matter what framework you use, you are actually using Spring’.

Some of you would think, “Yakov is either out of his mind or is writing this blog sitting in one of the coffeshops in Amsterdam”. Wrong! I’m just reading an announcement about the upcoming release of the popular Adobe framework Cairngorm 3: http://opensource.adobe.com/wiki/display/cairngorm/Cairngorm+3

When I first saw Cairngorm (post), it was easy to understand what they were trying to do, as the framework is so heavily influenced by the original version of Struts and the Core J2EE Patterns. Unfortunately, Struts was barely a fit for the original problems it was attempting to solve, and are completely inappropriate for building client-side applications within a real runtime. Struts used an adaptation of the MVC pattern called MVC2/Model 2, which was created to deal with building user interface within request/response model. This model is completely at odds with building a rich Internet application.

So, obviously, I find the move to hold Cairngorm up as a template for architecting Flex applications a bit concerning. Beyond that, one of the other things in the Java community that never proved to bring fruit was the desire to standardize anything and everything, which this idea seems born out of.  I am actually foggy at best with what they are trying to accomplish, and find it completely bizarre that they would try and morph a framework into this “thing.”  Do we really need some sort of Flex specific think tank on a long established pattern?  Also, it seems very bold to attempt to group the work of so many others under their umbrella.

Is this just an attempt to keep the brand name relevant of a framework whose time has come and gone?

Here is my write up: http://www.infoq.com/news/2009/11/flash-10.1-and-air-2.0

Honestly, this episode makes me wonder if our recipe for Drunk on Software success is more of Chet and less of James and I.

Original Post

Original Post: http://www.drunkonsoftware.com/2009/07/26/episode-15-universal-mind-spatialkey

Additional Resources:

• Component for Role Based Permissions in Flex
• Flex Security Component Updates
• Project Home