By mid Jan I had nailed a decent chunk of core system functionality, including:

• Listing public Games
• Entering a Game
• Pulling Game Submission Pairs
• Rating Pairs
• Submission Scoring
• Viewing Submission Details
• Viewing Fullsize Submissions
• Logging In
• Uploading Submissions
• Customizing/Editing Submissions
• Writing Comments on Submissions

And Then…

About two weeks ago I came to the painful realization that the client-side architectural framework I was using was adding unnecessary complexity to the project and generally slowing me down. I started looking at other options but each had drawbacks that I was not willing to put up with.

In the end though, it was not really the Document-Based nature of the project that was giving me hassles. I wanted a system that could automatically attach Mediators to view components as they appeared, without all the hassles of manual Mediator instantiation, naming, registration and removal.

So, against all “better” judgement, I borrowed ideas from PureMVC, Mate and Swiz, and rolled-my-own. And it was fun!

Conceptually it’s very much like PureMVC, except:

• It uses Dependency Injection instead of the Service/Model Locator Pattern
• It uses Flash Events instead of a custom Notification scheme
• Mediators can be automatically created for view components

There is actually very little to the framework: A Mediator Factory (for automatically creating Mediators when view components arrive on stage) and a Command Factory (for wiring Commands to Events) – everything else is taken care of by Dependency Injection.

I’m using the SmartyPants IOC library to handle injection. I sent a mail to the author of said library to explain my framework idea and the problem it is aiming to address. In his response he mentioned that my problem might be one commonly referred to as “robot legs“. Rad! While not entirely accurate, I dig that phrase, and hereby name my framework “RobotLegs”!

I’m in the process of migrating Pijin over to RobotLegs, and so far everything is going very smoothly. I’ll post a more detailed description of RobotLegs once I’ve spent a decent amount of time “eating” it.

UPDATE: RobotLegs 0.1, with demo Flex app, has been released. More info here:

http://shaun.boyblack.co.za/blog/2009/04/16/robotlegs-an-as3-mvcs-framework-for-flash-and-flex-applications-inspired-by-puremvc/

July 2008:
Started working on Pijin on a more full time basis – reduced Hello Computer work to 2 days a week.
Serious evaluation of technologies, tools and approaches.

August 2008:
Decided on Tools and Technologies:

• Eclipse with plugins: PDT, Aptana, Subclipse, and Flex Builder (likely to switch to FDT later)
• PHP 5.2+ with Zend Framework (AMF integration with CodeIgniter was too messy),
• PureMVC MultiCore for AS3 framework,
• Flex 3 for front-end prototype,
• AS3 for FP10 for deploy,
• SabreSMF for Remoting (likely to switch to Zend_Amf later)

Played with the Zend Framework
Decided on packaging and naming conventions
Basic server setup
Tested WebOrb, but decided on SabreAMF instead (WebOrb was too bloated, though I like the admin tool)
Started Pijin Core AS3 Library
Started Pijin Flex3 PureMVC MultiCore Project
Started Pijin PHP Library (Zend Framework naming conventions – allows use of Zend_Loader)
Started Pijin PHP Remoting Services
Wrote the Pijin AMFGateway bootstrap and CallbackServer for SabreAMF
Wrote the Pijin AS3 RemoteService
Started implementing Remoting security features for Services

September 2008:
Started simple Flex prototype interface
Integrated my CMS tool for quickly and easily setting up the Pijin V3 db Schema
Wrote/adapted Pair Retrieval algorithm from Pijin.net V2
Wrote/adapted Pair Rating algorithm from V2 – now with “weighting”
Defined Item Schema
Figured out how Styles will work
Upgraded to SabreAMF 1.2

Here’s why:

March 2006:
Pijin.net V2 went live.
It took me over 6 months to build it.. On the Flash timeline.. Nested code everywhere.. Globals. Nasty. Scary.

April 2006:
Started planning Pijin V3. Lots of feedback. Lots of ideas. Too many ideas.

July 2006:
Realized that it was completely impossible for me to build Pijin V3 at my skill level, or without the funding to put together a development team.
Continued planning and doing conceptual work.
Explored the possibilities of getting funding to hire experienced developers.

March 2007:
Joined Hello Computer to improve my skills and work in a team.
Started learning about Object Oriented Programming.
Wrote my first Class – an important step towards building V3!

April 2007:
Developed my first OOP project in AS2 using FlashDevelop and MTASC – another step closer.

August 2007:
Developed my first AS3 project – another step closer.

November 2007:
Developed my first PureMVC AS3 project -  another step closer.

Jan 2008:
Developed my first fully deep-linkable AS3 project with PureMVC and SWFAddress – an interesting exercise.

March 2008:
Developed my first CodeIgniter PHP project – another step towards V3.

April 2008:
Developed my first ExtJS JavaScript project – JavaScript: a step backwards!

May 2008:
Developed a CMS with ExtJS and CodeIgniter – useful for Pijin V3.

June 2008:
Decided I was finally ready to start building V3 again.

July 2008:
Started working on Pijin on a more full time basis, reduced HC work to 2 days a week.

So, I’m back on track. There is still an insanely huge amount of work to do, but at least it’s work that I am finally capable of doing. I’m hoping to have the Flex prototype ready by December, and a public Beta by March 2009.