Realtime hit counterweb stats

Archive for the 'Adobe Flex' Category

Global Error handling in Flash Player 10.1

Sunday, November 8th, 2009

A few years back I had the need to write a global exception management framework in order to provide a means from which unhandled errors could be handled gracefully throughout an application. Initially, I assumed there must be a listener available which could be added to SystemManager or Application, and from there all unhandled errors could be caught; however, after spending many hours trying to find a solution, it seemed this wasn’t possible to achieve in the Flash Player API at the time.

Just recently I found myself once again in need of such an API, and thought there may be something new in Flash Player 10 that I wasn’t aware of. However, as it turns out the only thing I was able to find actually happened to be in the Adobe Bug database; specifically, issue FP-444. It was then that I referenced the feature set for the next revision of Flash Player – version 10.1, and learned that in addition to many other significant new enhancements Flash Player 10.1 will finally introduce the ability to manage exceptions globally!

This is an invaluable feature for all Flash based applications as it will allow for the ability to add a global listener from which you can catch any and all uncaught RTEs. The ability to provide a global exception management mechanism will undoubtedly improve Flash applications significantly as uncaught RTEs can then be handled gracefully by responding to the user accordingly.

This may be one of those new features that isn’t necessarily the main focal point in marketing campaigns. However, it is APIs like this that make life for developers (and QA) much easier and, equally as important, ultimateky have huge payoffs for users.

Coincidentally, this also just so happens to coincide with an open source Exception Framework I am planning to release. The framework will provide a foundation from which more robust exceptions can be defined (in comparison to the generic Error class), as well as common exception types which implement the framework. So if you are interested, check back soon as I plan to publish the framework this month.

Maven dependency resolution with javacio.us

Wednesday, October 21st, 2009

Ever need to determine what versions of a specific Maven dependency are available in a public repository? Or more importantly, determine if a dependency even exists in a public repository? If so, then javacio.us, a free, subscription based service available to anyone with a Google account can be used to help quickly and easily resolve maven dependencies.

Usage is simple:

  1. Search Google for the name of the dependency (e.g. concurrent pom) and you will get a list of all groups in which the dependency is under a public repository.
  2. Select the specific group you require
  3. Select the version you require

And that’s it! Now you can simply copy the coordinates and the dependency has been resolved.

While there are certainly other services available which do more or less the same thing (e.g. Maven Repository Browser, etc.), the simplicity of being able to just search from Google is rather convenient.

Thoughts on Cairngorm 3

Sunday, October 18th, 2009

A week or so prior to MAX, the Cairngorm committee had a rather interesting discussion, during which Alex outlined what the team at Adobe Technical Services had been considering for Cairngorm 3. The meeting was focused on providing everyone with an overview of the collective ideas which Adobe had been gathering internally for some time now, and to also inquire feedback prior to the public announcement of Cairngorm 3.

Prior to the meeting I had anticipated the discussion would be based around a few new patterns and best practices which are currently being advocated, and possibly some additional libraries which help to address recent challenges in RIA development. However, what we discussed was actually quite different – in a good way.

As you are probably aware by now, Cairngorm 3 is focused around tried and tested best practices and guidelines which aid Flex developers in providing solutions to their day to day RIA challenges. These guidelines are primarily based upon that which has been realized by Adobe Technical Services, and also from the Flex community at large. Teams can leverage these guidelines where applicable to help deliver successful RIAs using frameworks of their choosing. While there may be specific frameworks and libraries recommended in Cairngorm 3, these are just that – recommendations. There is generally a framework agnostic approach which I must emphasize is highly preferable to that of suggesting one framework over another. This is precisely what I think is needed in the Flex community, for there is rarely a one size fits all approach to software architecture, especially in terms of specific framework implementations. This is a pretty easy concept to comprehend, as, what works for one team, in one context, may not always be appropriate for another team, or in another context.

Cairngorm 3 is a step forward towards what (IMHO) should be a general consensus in the Flex community at large; there are many existing frameworks out there which help address specific problems, with each providing unique qualities and solutions in their own right. This is the kind of thought leadership which helps a community progress and grow; it should be encouraged, as allowing for the shared knowledge of fundamental design principles and guidelines is something which provides value to all Flex developers, regardless of which framework they happen to prefer.

If there is one suggestion I would propose, it would be to have an entirely new name for these collections of best practices, guidelines and general Flex specific solutions. Personally, I would like to see the name Cairngorm (which, after all these years, I still pronounce as Care-in-gorm) refer to the original MVC framework, i.e. the framework implementation itself, as keeping the name the same while undergoing a different direction is bound to cause confusion to some extent. Whatever the new name would be is insignificant as long as the original name of Cairngorm applied to that of the actual framework implementation. This would perhaps be more intuitive as it would allow for the name Cairngorm to be used to describe a concrete framework as a potential solution, just as one could describe other frameworks; e.g. Spring ActionScript, Mate, Swiz, Parsley, Penne, Model-Glue, PureMVC, Flicc etc.

Most importantly, however, is the prospect of choice, as choice is always a good thing. Moreover, an initiative being lead by Adobe in this area sends a very good message to the Flex community as a whole. I happen to leverage a number of different frameworks and patterns which address different problems. As new problems arise, I employ new solutions where existing solutions may not suffice, or develop custom solutions where none are currently available; never blindly choosing one solution over another. However, in every case, there are typically some basic, fundamental guidelines which hold true and can be followed to help drive a design in the right direction. Regular readers of this blog have probably noticed that the basis of the majority of my posts are heavily rooted within these fundamental design principles, as it is from these fundamental design principles and guidelines that developers can utilize the frameworks which work best for them in order to and meet their specific challenges.

Essentially, Software Architecture is all about managing complexity, and there are many fundamental patterns and guidelines which can help developers mange this complexity. The specific framework implementations are of less concern, for it is the understanding of these patterns and principles – and more importantly, when to apply them, which will ultimately drive the decisions to leverage a one framework over another. In my experience, I have found that the only constant in software architecture is that a pragmatic approach should be taken whenever possible, whereby context is always key, and simplicity is favored as much as possible. Cairngorm 3, I feel, is a nice illustration of this principle.

Vector Iterator for Flex

Friday, October 2nd, 2009

One of the many welcome additions to the Flex 3.4 SDK is the inclusion of the Vector class. Vectors in particular are especially welcome as they provide compile time type safety over what would otherwise not be available when implementing custom solutions, such as a typed collection.

Essentially, Vectors are just typed Arrays. And while not as robust or powerful, Vectors are similar to Generics in C# and Java. When it is known at design time that a collection will only ever need to work with a single type, Vectors can be utilized to provide type safety and also to allow for significant performance gains over using other collection types in Flex.

I recently wanted to convert quite a few typed Array implementations to Vectors, however, the Arrays were being traversed with an Iterator. In order to reduce the amount of client code which needed to be refactored I simply implemented a Vector specific Iterator implementation.

If you are familiar with Iterator Pattern in general, and the Iterator interface in particular, then usage will prove to be very straight forward. You can use the Vector Iterator to perform standard iterations over a Vector. Below is an example of a typical client implementation:

var abc:Vector.<String> = new Vector.<String>(3, true);
abc[0] = "a";
abc[1] = "b";
abc[2] = "c";

var it:Iterator = new VectorIterator( alpha );

while ( it.hasNext() )
{
    trace( it.next() );
    // a, b, c
}

Using an Iterator with a Vector ensures only a linear search can be performed, which proves useful with Vectors as they are dense Arrays. However, one consideration that must be made when using an Iterator with a Vector is that you loose type safety when accessing items in the Vector via iterator.next(). It is because of this I would suggest only using Iterator’s with Vectors to support backwards compatibility when refactoring existing Arrays which are being used with existing Iterators.

The VectorIterator and it’s associated test are available below:
VectorIterator
VectorIteratorTest