It’s a problem with any centrally distributed environment. HTML5 will have this problem and it will be made worse by the politicized standards process. Plugins will have these problems as long as the execution speed of the API and built in features is much faster than than the speed of a developer’s code.

For example if the built in vector rendering isn’t quite right, there is no point of writing your own in AS3 through BitmapData, it won’t match the performance.

I think technologies like NativeClient or Microsoft Xax which are promised to be secure alternatives to ActiveX are what we should be getting excited about. We should have nothing less than all the speed and features of desktop applications in the browser.

It would end the endless cycle of feature requests. If a framework or API doesn’t have a feature, you could implement your own without worrying about performance. There still would be a lot left to do, but at least the possibilities will be on-par with native applications.

As far as having gateway languages around like Basic, JS, AS2, AS3, Processing etc.. this can be achieved with frameworks or compilers as it always was. No need to handicap professional programmers for that. A framework or language designed to be a gateway platform would work much better than AS3.

Myself I would never compile Tamarin or made some changes in AS3 compiler. It is out of my league. What pissing me off is Adobe’s inability to polish AS3 and API. They have too much bugs, too much inconsistencies, and documentation is not sufficient. Adobe just moved forward, and that is behavior inherited from Macromedia. I have feeling that they don’t care. Just because 90% of developers doesn’t care. I care.

I care cause 70% of my development time goes in hacking Flash. Trying to force Flash to do what is supposed to do. 90% testing time goes in performance testings ’cause there’s no destructor. Whole language has beta feel for me.

For Adobe (formerly Macromedia) it isn’t a big issue. They are happy as long designers can make “stunning” RIA or any fool can make “stunning” RIA. So, they have a Flex. If I wanted do make RIA using tags, why would I use Flex? HTML+JS+XMLHttp does very good job for me.

I’m not a top-notch developer as you are. Maybe I’ll never be. I know that your effort (which is out of my scope) will make as3 development more comfortable for me.

The one truly outrageous thing about AS3 is its inability to recognize AS2, which has probably crippled many a project. Imagine if photoshop version Y wouldn’t recognize photoshop version X files? Madness

Why isnt there a ‘destroy’ or ‘dispose’ function implemented (eg in case of ‘Movieclip’ or ‘Sprite’), where all childs and all EventListeners are removed?
i cant wrap my head around these simple things;
they should be the most basic features of a modern oop framework language, which is supposed to produce big ria applications.

Also, may I suggest that for reporting bugs or enhancement-requests of this sort, reporting it at https://bugzilla.mozilla.org/enter_bug.cgi?product=Tamarin is a good idea (in addition to bugs.adobe.com).

Dont get me wrong I really love AS3 and the platform. As an “interaction designer” AS3 is perfect for me to quickly prototype ideas and manage large libraries to later reference (packages are great)… without ever touching a FLA. But really the language additions (as nice as they would be… please give us Abstracts) performance should really be their main concern.

I have been lucky enough to have the past year to focus on “Surface” based interaction in Flash 10… (think multi-touch/fidicuals) and performance is even more crucial as each time you add a new cursor (finger) the available vars and thus dynamic nature of software can change times (n). I really think Adobe has fallen behind in the user input department… for example all of Flex is rigidly coupled around mouse and keyboard… with no abstraction available much of this has to be overrided or rewritten to work with my own “TouchEvent” class… and with Windows 7 coming out (Sept) focused on touch computing and we have heard null from Adobe/Flash Team on their roadmaps for dealing with touch input… when MS already has tutorials on IE8/Silverlight implementations in the browser. Also, touch is just the beginning of where emerging sensing solutions will bring… with each being more performance demanding e.g. augmented reality, voice or computer vision…

Either way love Adobe and I hope they have some secrets up their sleeves ;) And Joa you rock too for bringing up these topics… someone’s gotta do it :)

[T]hese days, if I catch a programmer writing a linked list, that person had better have a very good reason for doing so instead of using an implementation provided by a system library.

–Michi Henning, “API Design Matters,” Communications of the ACM 05/2009, p. 55