Exactly - Event-Model - ENTER_FRAME is just an event as we all know - its just attached to a timer that's
running on the stage.

On Sep 23, 2008, at 6:25 AM, Josh McDonald wrote:


Quoth Alex:

"Player is pretty simple, at least in Windows. I haven't seen many threads if any. Input is based on Windows messages, I/O uses non- blocking, etc."

Player most likely won't do anything while your code is busy chewing up CPU time. Once you go back to waiting for an event however, it goes about its business.

-Josh

On Tue, Sep 23, 2008 at 2:14 PM, Samuel Colak <[EMAIL PROTECTED] >wrote:
Ralf,

Thanks for the link - unfortunately its making a number of guesses and to be honest its not entirely accurate.

Flex / Flash obviously is an event driven paradigm. The ENTER_FRAME is simply an avenue to do time-based event processing however this is not to say that the renderer itself is suspended at any time.

It is just a coordinated timely event which unfortunately doesn't help in the UI rendering mechanism.

Im pretty sure that certain graphic operations have been coordinated such that the UI Renderer effects things such as bitmap draw and fill operations in a locked fashion, but i would hazzard to state that the whole graphics pipeline does not get suspended for large number of modifications but rather per-operation followed by a "flush"
instruction.

This is why i'm kinda looking into a way of seeing if this exists on some low-level point.

On a side-note, there exists functionality to suspend all background processing (basically suspending the event- notification recorder) - something i noticed a bit of time ago which gave me hope - since i wondered if in the
bowel's of flex's core, there was such a UI function?

I have attached a sample movie file - testTdCode.swf - about 300k which shows off the TDRenderer i wrote.

Please excuse (Adobe that is) the use of the Flex Logo - obviously this is their property.

It does a number of 3D manipulations (again my code for the renderer is open if people are interested) and then restructures a bitmap in realtime to the new dimension. You'll note that the hit on the CPU is very low until the mouse-move / mouse-down
operations are detected within the "overlay".

Regards,
Samuel



Im-At-Home BV
http://www.im-at-home.com

Overtoom 238-II / 1054HZ Amsterdam / The Netherlands
Tel: +31 20 750 8304 (Amsterdam, NL) / +1 646 385 7345 (Manhattan, US) / Mobile: +31 64 328 5922

Skype: samcolak / MSN: [EMAIL PROTECTED] / AIM: [EMAIL PROTECTED]

On Sep 22, 2008, at 7:38 PM, Ralf Bokelberg wrote:

You might want to have a look at this excerpt from Colin Mook's AS3 book:
http://www.moock.org/blog/archives/000235.html
It should give you a good understanding of how rendering works in Flash/Flex.
Cheers
Ralf.

On Mon, Sep 22, 2008 at 7:15 PM, Ralf Bokelberg
<[EMAIL PROTECTED]> wrote:
> Afaik the Flashplayer does this for you. Nothing is rendered as long
> as you are in a script. You can try to draw a line and then do a
> simple while loop for 5 seconds. You will not see any updates of the
> screen.
>
> Cheers
> Ralf.
>
> On Mon, Sep 22, 2008 at 5:05 PM, Samuel Colak <[EMAIL PROTECTED] > wrote:
>> Alex, Adobe Guys, Community, Romans.....
>>
>> Is there a way to halt the UI Graphics Renderer so that (in a way) you
>> can achieve the ability to post UI changes (in effect double
>> buffering) before
>> the renderer performs any UI update ? An obvious although troublesome >> way of doing this is with bitmap however the routines for drawing do not
>> appear to be common between BitmapData and DisplayObject.
>>
>> Such like under DisplayObject
>>
>> graphics.unlock; // disassociate graphics device from
>> renderer...
>> graphics.[do stuff here].... // misc graphics routines...
>> graphics.lock; // re-associate renderer to graphics device and
>> flush activity to renderer
>>
>> (lock/unlock might be switched depending upon your perspective to the
>> renderer)
>>
>> I might be jumping the gun with you guys producing some hardware
>> acceleration so if i am apologies in advance.
>>
>> I am also aware that there is a possibility that if the re- associate >> does not take place, then all other render activity might be lost....
>> so obviously
>> this is not something that you do on a whim. Since everything is more >> of less sequential then im pretty sure that most programmers would not
>> forget to do this.
>






--
"Therefore, send not to know For whom the bell tolls. It tolls for thee."

http://flex.joshmcdonald.info/

:: Josh 'G-Funk' McDonald
:: 0437 221 380 :: [EMAIL PROTECTED]



Reply via email to