Where there's a will, there's a Makefile

2009-07-07T00:41:00.000-07:00

We have finally switched over to Google Code! It took the NRG a while to complete the switch, and there are still a few things to patch here and there, but it looks like we're well on our merry way!

That was the good news. The bad news, for some, is that iCognition can now be built using the GNU make utility. There, there, Windows users, I'm pretty sure it's going to work with Cygwin or MingW/MSYS in the nearest future.

Well, If you're like me, and you use Linux (or OS X) and 'vim', you should be ecstatic about this news! You don't need that fancy Adobe Flash CS3 IDE to write your Actionscript anymore! And, what's even better, you can now SSH in from anywhere and work on your project in the good old terminal. "But wait! The pretty 'Publish' button is gone!", one might say. Yes, absolutely, but isn't the Open Source world a great place to be in? (at times, when tools compile and there aren't any segmentation faults...). I've taken the liberty to compile iCognition using the Adobe Open Source Flex SDK. Since then I've hardly been using the FlashCS3 IDE, because 'vim' is better in so many ways.

My workflow has changed too, since I know have to run commands to build iCognition, as opposed to clicking a button. Logically, it didn't take long for a Makefile to be born, and along with it - a mighty configure script! The 'configure' script has been written from scratch and performs a bunch of tests on your system to ensure that the build process will go smoothly. The Makefile is pretty straightforward at this point, but it is slowly growing into a 'real' Makefile - you know, one of those long, twisted and complicated Makefiles nobody enjoys to read? Like that, only more elegant, where possible.

Obviously, you'll need the Flex SDK if you want to use the Make-based build system. I'm going to support publishing using Adobe Flash IDE, mostly for debugging purposes (I'm not a fan of command line debugging, unless there is no other way. I'd rather use trace() over 'fdb'), but at this point it seems that Flex is the way to go. And with that comes the question of whether I should support Adobe Flex Builder, to which I can only answer: "Why not? But later". FlashDevelop with the debug addon seemed like a good candidate too, but all in good time.

Speaking of Flex, I might switch iCognition to Flex completely in the future, since SWC files look pretty attractive for encapsulating Flash tasks at this point.

That being said, here are a few links related to iCognition's development:
iCognition@GoogleCode
iCognition Discussion Group
iCognition SVN Repository

Welcome to the iCognition Blog

2009-06-17T10:43:00.001-07:00

Hello, world!

Interestingly enough, we're switching from Sourceforge to Google Code just when development of iCog 2 has been restored. For a few weeks, we’ve been working on integrating the data exported by iCog1 with an in-house application we’re using here at the NRG@Harvard and.. wow, that took us a while! But let’s not jump too deep into the matter just yet. I believe I should explain what’s going on briefly and perhaps even introduce myself.

My name is Victor Petrov and I am the lead developer behind the iCognition Project. I am part of the Neuroinformatics Research Group at Harvard University - a group of developers heavily focused on building reliable software to “facilitate the integration, mining, and sharing of neuroimaging and associated data” (source). And we’ve decided the time has come to build iCog.

iCognition (aka iCog) is a software package that aims to become the world’s leading test administration tool. Our focus is on cognitive and behavioral tests, but it is by no means limited by that - we also plan on including support for creating and administering surveys, as part of the 2.1 release. However, the building blocks behind iCognition allow it to be extensible and morph into any kind of testing software the developer wants - be it cognitive, behavioral, or generic assessment tests. In fact, I bet I’ll be able to morph it into a bug tracker or into a massive multiplayer online poker game. Yes, that’s the kind of flexibility we’re injecting into iCognition!

Have I mentioned user-friendliness and web services? We’re very excited about Web 2.0, Javascript toolkits such as ExtJS and Dojo, PHP frameworks like Zend and CodeIgniter, formats such as JSON and XML, REST services and everything else that makes it possible for users to feel great when interacting with dynamic UIs, for developers to harness the power of our elegant REST API, all while still accomplishing tasks with minimal effort.

With that being said, the goal of this blog is to periodically shoot into the dark corners of the Internet some updates on the current status of iCognition, a few ideas that pop into our heads and key issues we encounter while working on iCog. Hopefully, someone will throw a message back at us, expressing interest in the project, asking questions or, who knows, maybe even criticizing my decision to use Google Code instead of Sourceforge. Either way, please contact me directly at: victor [underscore] petrov [at] harvard [dot] edu . Eventually, iCognition will get it’s own website, but for now we’re going to make full use of the features Google Code has to offer.

Speaking of iCog’s status, there are a few things you should know - and that includes some history, too:

What we call “the first version of iCognition” or iCog1 already exists - we’ve developed it and are actively hosting it on the Web (the URL is somewhat private). iCog1 includes a set of cognitive tests written in Flash, an administration panel (go jQuery!) and data exporters for CSV, XML and JSON formats. However, it has its limits and is heavily customized to serve the needs of the Cognitive Neuroscience Laboratory at Harvard. It is, of course, open source and doesn’t come with a manual or a user guide (it is mainly a proof of concept, something to test things out and show that the data we’ve captured is valid and can be used reliably to capture subject data online). It does come with commented code, so anyone is free to download the source and start reading it. I would recommend against doing so, though, because we want to release iCog2 to the world (with an Open Source license, of course). I believe the license will be GPL, mostly because ExtJS is GPL’d, but we’ll make a definitive decision closer to the date when we start implementing it. The source of iCog1 is available only in the iCog SVN@Sourceforge, in a ‘v1′ folder - I’ve decided to use the SVN@GoogleCode for iCog2 development only. However, we’ve halted all development of iCog1 - we’re only bug-fixing it and that will soon be stopped too, so iCog1 is slowly becoming more and more obsolete, as iCog2 gets built.

iCog2, or just iCog, is the second version of iCognition and is what this blog is all about. Since iCog1 is not really intended for other people except the CNL@Harvard, we’re refering to iCog2 as iCog or iCognition, with no version number, as this is going to be the first public or “release-quality” version. iCognition is going to be a complete rewrite of iCog1, with reusability, user- and developer-friendliness in mind. We’re going to allow people to create cognitive and behavioral tests with drag-and-drop ease. At least that is the goal. So far, we’re taking care of the architecture and we’re making sure it is rock solid. We’re following a few guidelines on software construction and we’re definitely documenting the process as much as we can. Sourceforge already contains a few documents on iCog’s architecture: Vision and Scope, Software Features and Requirements, Use Case Scenarios and Software Requirements Specification. These are still considered a work in progress, so things might change in the future.

That is pretty much all that has been going on here at the NRG@Harvard. We’re planning on introducing a few innovative features into iCog, so please check back often to read more updates on iCognition!