/Word stream

Adobe has finally updated their Flash Player Version Penetration statistics page with details for December 2008, and things are looking good. Just how good? Well, basically, Flash 10 has enjoyed the fastest adoption rate of Flash since it has been introduced, having now a 55.9% penetration on mature markets roughly two months after it has been released.

I’m actually quite surprised that my own prediction of 90% penetration in 6 months for Flash 10 – previously it would take one year for Flash to reach that mark – seems more realistic now. That, coupled with other recently announced numbers like 100 million Adobe AIR installations and a reported 80% market share on online videos, make the future look pretty bright for the Flash platform.

And let’s not forget the fact that, even with current penetration numbers aside, the adoption rate of a certain competing technology is vastly inferior to Flash’s (click on “Line” to see how each different version of each platform has been performing, then compare Flash 10 and SilverLight 2); while SilverLight is obviously gaining some penetration, it seems sponsoring certain exclusive online video offerings isn’t doing as much to the platform as one would think it would.

Finally, strangely enough, Flash Player 9.0.115 (the first to include h.264 video support and some other features like hardware scaling of fullscreen content) is not featured on the official penetration statistics anymore, and even the old numbers have been removed from the tables. I guess that’ll become irrelevant pretty fast – it already was at 90% on september – but historically, it’d make sense to still list them there.

Notes: the above graphic was created with data gathered from past updates of the Flash penetration statistics pages. You can access the full data set here (or click the image) – it also includes a chronological graphic. Some version lines start with a penetration rate higher than 0% at month 0 because it has been released the same month the statistical data was gathered. Due to this, consider some error margin translation of around 15 days (left or right) for each line.

FlashTerm, Peter Nitsch’s ActionScript Telnet client, has just gone live. It works with a sample BBS, and it’s a great example of what’s possible with AS3ANSI, his AS3-based ANSI parsing library. Read Peter’s blog for more information.

For what’s it worth: I’ll be giving an (online) talk about ActionScript development in Brazil December 18 (next thursday), at 22hs local time, courtesy of the Adobe User Group for Rio Grande do Sul (AUGRS). This talk will be in Brazilian Portuguese only, and it’s aimed at people who want to know more about the Actionscript development field; I’ll be talking about my experience in the area, as well as how I see it working for other people. It’s not overly technical, but I’ll be talking a bit about the tools developers have available; it’s not overly personal, but I’ll be showing some of my work to illustrate my points and talking a bit about my experience to put things into context. Find more information about the event (including link to the presentation room) here.

The presentation will also be recorded and available later, as well as all material involved.

Update: the starting time has changed, and it’s now set as 22hs instead of 21hs. Sorry for that.

Update (2): the talk was recorded and is now available online. It lasts for around 1 hour for the main presentation, and another 45 minutes for questions. Thanks to everybody who attended! The turnout and the result was better than I expected.

This is news to me: Google Analytics is updating their tracking features and adding better support for Flash tracking – it was possible before, but it’s just getting more native, as Google is even providing an AS3-based framework for that. Below’s the email that I’ve received through my Google Analytics admin account.

We are happy to let you know that the Event Tracking feature is now available in all profiles for the following Google Analytics Account ID: XXXXXX. Please note that you are receiving this email update because you are an ‘Admin’ or designated contact for the account listed above.

When you log in to these profiles, you will see a new set of reports called “Event Tracking” under the Content section. As posted on our blog, this is a limited release currently available only to select profiles.

Event Tracking allows you to track interactions with Web 2.0 style content such as Flash, AJAX, Silverlight, social networking apps, etc. We recently made tracking Adobe Flash even easier with the release of a new Flash Tracking client library. It allows for much simpler tracking of Flash content with drag and drop functionality and an open source framework.

To use Event Tracking, you will need to upgrade your site to use the new ga.js javascript. Detailed instructions on how to set up Event Tracking on your site are available on our CodeSite.

To find your ga.js code snippet, edit the settings for your profile and click the “Check Status” link on the upper right corner of the page. You can now track interactions beyond just pageviews.

Check the links above – it looks pretty awesome.

As everybody knows by now, Adobe has made Alchemy – their C/C++ compiler for AVM2 (AS3-ish) SWF bytecode – available for download. We’ve seen highly visual demos, like Doom, and the old Quake 1 video where we got the first glimpse of what would become Alchemy. They’re very cool to show, and there’s some additional example libraries, but I guess Alchemy is ready for actual production uses of the technology; there’s not many such libraries shared yet.

Unfortunately I don’t have a long beard and I’ve never learned to be a proper C++ programmer, so the task of porting anything must fall on better hands. So, I’ll allow myself to post suggestions of a few libraries that I believe would make nice Alchemy ports in the hopes that someone more capable takes pity in our high-level, memory-managed struggle.

• OpenCV is the standard free library used in Computer Vision projects – this means face, gesture, and motion recognition, as well as a lot more crazy skynet-like features. It’s normally used in platforms like Processing and vvvv. Having that available for Flash projects would be a real step forward.
• Box2D is an open source 2D physics engine, largely responsible for the proliferation of physics-driven gameplay in amateur games. While we have Box2DFlashAS3 and other physics engines in Flash, having a straight port with better performance would revolutionize physics in Flash games.
• ARToolKit is a library used for augmented reality – overlaying digital images into real world images. There’s also FLARToolKit now, but as with Box2D, a straight (if adapted) port would probably be nicer in terms of features and speed.

Honestly I have no idea of the portability for these projects; all I know is that they’re based on C/C++. However, they’re straight data processing libraries (maybe except for the rendering part of ARToolKit), so I think they wouldn’t be too limited by the player’s external capabilities and would make nice libraries to have available.

I’m not very much of a radical, I believe, but I tend to judge development communities by the amount of negative or positive energy they seem to spread around. This is hard to put in objective terms, but as a positive example, I’d point to the PHP website – you have a fairly straightforward website and visiting the reference for each command will likely present you with hundreds of examples added by the community on how to use that command for all ranges of tasks. I’m not a professional PHP developer, but implementing features in PHP has always been super easy simply because of the community – the way they’re unattached to knowledge and unafraid to share it.

For me, the other side of the spectrum would be developing with Microsoft technologies. There are amazing exceptions like the XNA framework (which may be due to the nonexistence of a commercial exploration model until very recently), but overall finding answers or additional (free) libraries for certain problems using MS technologies has always been pretty difficult – for some reason, I always end up on forums and websites that require payment for solutions. There’s MSDN which is thankfully public, but I can’t forget how you had to actually buy the references in CD until some years ago, and I think that sort of approach influences how the community at large behaves.

So here I am, rushing to add Sound support to Fnk – the final version of the software for my thesis is due tomorrow. I manage to get FileReference (using the new Flash 10 capabilities which are now old hat in the whole project) to successfully load the data from a Sound file. Now I just want to play that. Simple, eh?

Not so fast. You can’t just feed Sound data to a Sound object – actually you can’t do that with any kind of “native” file that the player loads (to read a Bitmap, for example, I have to actually load the data as if it’s a Loader and .draw() it to a BitmapData).

But back to the point. I want to play sound from loaded sound data. A quick search on Google turns out this page, in which a guy talks about the very same issue and makes AS3 libraries for MP3 parsing freely available for download. Wonderful, I think. I mean, I’d love to implement it myself, or initiate an Adobe Make Some Byte Parsing campaign for a more native way to do it, but I’m in a hurry. So I quickly adapt it to my needs and 5 minutes later, Fnk is happily loading and playing MP3 data.

Then I compile an AIR 1.5 version of the app and it doesn’t work. Apparently it thinks I’m trying to load executable data and prevents the MP3 parsing classes from loading at all, throwing a security exception when the library tries to parse it.

It turns out a quick search on Google gives me this page where someone else tell what’s it all about and how to solve it. Bam, I patch my version of the libraries and I have it working 100% on both the Plugin and the AIR version of my software.

The ActionScript community may not have a central point of community knowledge like the PHP community does. But it’s amazing the amount of helpful folks and blog posts we can find when doing a Google search. This issue I had isn’t even something old and established – it relates to a very recent issue of a very recent feature. That there were people on it so fast and posting the solutions online amazes me to no end. Flash development may be something frowned upon by some of the most fervent free software advocates, but for some reason the community has managed to retain the ethos of a group where knowledge must be free and shared among everybody. I don’t know why that is, but it’s awesome.

Keith Peters said something similar (via Mike Chambers) a while ago and it got me thinking. It’s something I hadn’t realized at first, maybe because I already take that for granted. But the above experience I had just now was a wonderful realization of the impact of that all. I was thinking I’d need to implement my own MP3 parser, something that would probably be fun but take a while, or wait for Adobe to implement the data loading capabilities to this and maybe a dozen other classes… either way, I wouldn’t have that ready for my thesis. I read all kinds of ActionScript feeds daily, and I had seen no mention of that issue until then, so I really thought I was out of luck and ready to shove the standalone version aside and worry about it later; solving something so specific in less than half an hour was the kind of glad surprise I don’t always get.

So, thanks.

ANSI art visualization in the browser has matured: meet Peter Nitsch‘s AS3-based ANSI Viewer.

He talks about it a bit more here. That work led him to create as3ansi, an ANSI parsing and visualization library for ActionScript 3. There’s a lot more about that here, with some sample code to illustrate; his event-driven ANSI parsing code is particularly awesome, considering the peculiarities the standard poses.

This may be hard to understand for people uninvolved with the underground BBS art scene around the 90′s, but for people who breathed that kind of 16-colored oxygen, seeing that come to life in a browser is wonderful – I spent a lot of time doing that and running a local ANSI art group, and heck, look at my business card. I’ve had my share of similar experiments about ANSI viewing in ActionScript in the past, but Peter’s work more than surpasses that. I can’t wait to see what the guys over at SixteenColors will make of it.

Holy deployables, Adobe has silently made AIR 1.5 available for installation today. Among other things, it now supports Flash 10; you can find more information here (source: Jen deHaan).

As I’ve just been waiting for it, I’ve now made my own crazy application/language/editor/experiment Fnk available as an AIR 1.5 application. You can install it from here (or a direct link to the .air file of the current version here). No fancy-schmancy AIR features yet; it has no auto updates or unrestricted file reading/writing or anything, so it runs pretty much like the browser version. But it’s certainly nice to see it running as a standalone application. As much as I enjoy having the ability to run something from the web (and it’s a core feature of Fnk), being able to double-click .fnk files and have them open on an editor is certainly convenient. Sort of ironic I guess, but since AIR allows us to deploy applications on both fronts almost seamlessly, I don’t have to choose what to support exclusively.

Straight and to the point, hopefully as a reference, because I had trouble finding the answer to problems on specific parts of the process. This considers you already have a normal Flex SDK project setup in whichever tool you’re using (I’m using FDT).

1. Be sure to download the latest version of Flex SDK command-line tools (I recommend the “Adobe Flex SDK” files, whichever version – stable if you wanna play it safe, nightly if you’re feeling lucky). You do not need to edit the config.xml files – it’s set to Flash 10 as default.
2. Change your regular AS3 compilation procedure to use amxmlc.bat instead of mxmlc.exe for compilation. The same compilation syntax applies so this should be pretty straightforward.
3. Create an AIR project. This page has a nice tutorial on how to do it. You don’t need to create the .mxml file though – as long as you use the same syntax you use to compile normal .as-based projects in mxmlc, it should work. Very important: the AIR XML file header should be set to use version 1.5. So the application tag start should read <application xmlns="http://ns.adobe.com/air/application/1.5"> instead.
4. Run your application with adl.exe instead of flashplayer.exe.

Optionally, the complete AS3 reference (including AIR) is available online. You can also download it in zip form from this link.

And remember AIR applications built for Flash 10 will require AIR 1.5 to install (the current version is 1.1). This means you can compile and run your application locally without a lot of trouble, but if you try installing your packaged Flash 10-based application on AIR 1.1, you’ll run into a misleading message that says the package requires an “older” version of AIR and that you need an “upated” version of the packaged application. For the installation to work, we’ll have to wait for AIR 1.5, which will come “this fall”. You can find more information about AIR 1.5 here. AIR 1.5 is now out so applications built with the method mentioned above will work.

I don’t know if anyone has noticed, but Adobe’s Flash Player Version Penetration stats page is still stuck showing statistics for June 2008. Considering the previous versions of the survey, the next update would have a column for September, and according to my not-so-empirical analysis of archived versions of the page, Adobe’s website would usually update with new information the same month of the column, or the following month tops; we’re already on November, however, and there’s no update for September so far.

Adobe may have just delayed the survey for one month – they did so in 2006, for example, when there was no March 2006 survey, but an April 2006 one (shortly before dumping NPD in favor of Millward Brown, by the way). Or maybe there’s something else, like a change in methodology or a delay so that some early Flash 10 penetration can be included. All my own speculation, though, as I can’t think of any real reason for such a delay.

Update: The stats page has been update with numbers for September. Thanks Adobe.