Flash memory management and Actionscript - actionscript-3

There was many discussions about this problem, but I want to pay attention on the situations that IMHO seems not so clear:
Yes the general rules are:
Remove chachedAsBitmap
Stop movieClip if playing
Remove events
Delete references
etc.
But let's look:
First Example:
I have nested sprite (ex: mainSprite), it contains other sprites with dynamic textFields in it (and are chached as bitmaps), just textFileds and MovieClips with event listeners on it (with weak reference).
When I need to remove this sprite I need first to remove all it's nested content via loops or just
removeChild(mainSprite);
mainSprite=null;
is just enough?
Second Example:
I have some sprite in which I'm loading bitmap and manipulating with bitmapData, later I'm just replacing content of this sprite with another bitmap, is allocated memory for older bitmap automatically erases and is overwritten or it still exists?
Third example:
I have some "graphics template" MovieClip (in library with Export for Actionscript property set on it) which I'm adding on the stage and filling with dynamic data (and adding event listeners), let's say that it's one scene of the app, on another scene I need same MovieClip with other dynamic data, but inbetween need to clear my stage (need something like transition animation which is also library MovieClip), what's the best way: to set this MovieClip visible property to false (while transition animation is plays) and then reuse it, or just remove it with removeChild and then add when add with addChild once more?
All I wrote is more about Air Mobile, cause in most cases for the desktop these situations aren't so problematic, but in case of mobile development they are.

You can visually monitor memory usage along with fps etc using this lib: http://code.google.com/p/flash-console/
hope that helps.
P.S. gc in flash is always a weird thing :)

First example: removing mainSprite from display list is enough if there are only weak listeners on its children.
Second example: I'd advice reusing the same object with visible = false. Recreating the same object is more resource expensive plus you get another instance of the same thing being in memory before it gets gc'ed.

Related

as3 how can i prevent that a new instance is created by entering a frame?

i am working with several nested movieclip objects in a project. but i get into trouble with the buttons i created and implemented in the nested movieclips:
to describe it in a simple way:
I have a main movieclip with five frames, including two buttons with listeners to browse between the frames. Then inside of one Frame I have another movieclip with its own buttons. i instanciated it by hand not through code and gave it a specific name like "nestedMc".
Now I dont want to build the Listeners for those buttons inside the class of the nested movieclip class but in its parent class, which works fine until i then goto another frame in the main movieclips timeline and come back.
obviously every time flash enters a frame its contents get created anew (and therefore get new instance names). I could now try solve this through filling the frames via code.
But maybe there is another way to make sure the frame contains the same instance everytime i enter?
Timeline scripting is a dirty business, and really, a carry-over compatibility layer for Actionscript 2 projects. Whenever possible, I highly recommend not doing it, and simply keeping all of your code in your document class. As you're experiencing, timeline code causes headaches.
Consider instead just creating both states of your Stage (it sounds like that's what your two buttons are jumping between) and simply hiding them offstage or setting their alpha to zero and their mouseEnabled state to false. Furthermore, if the purpose of your frames is to play animation (a tween), consider instead switching to a much more powerful suite such as TweenLite. Moving an object over a hundred pixels (smoothly) can be as easy as:
TweenLite.to(redBall, 3, {x:100});
Now, if you're manually adding these items to the stage, as long as the object is a dynamic one, you can assign an instance name to it which will be saved between frame loads. Be aware the object name is not the same as the instanced name. For example:
var redBall:Ball = new Ball();
redBall.name = "bubbles";
The object's name is Ball, but it's represented as a variable called redBall. Its actual DisplayList name will likely be ambiguous (such as "Instance71"), and I can manually define it as "bubbles". 3 different names for the same object, all very different and necessary.
Even if you give the object a displayList name, you may not be able to reference it through code unless you enable Automatically declare stage instances, which basically creates on each object a pointer to the displayList object.
That said, you can always fetch the object by other means. Obviously, your buttons are always appearing, but you're trying to find a very specific object on the stage. At this point, we can use getChildByName() or getChildAt().
Hope that helps.
-Cheers

AS3 How to communicate between frames

I have been writing a game in timeline code. I want the different frames (rooms) in the game to be able to share information between each other. Of course, timeline code is limited to the frame it is written in.
After doing quite a bit of reading ("Foundation Game Design with Flash" and a number of articles, tutorials, forums etc) I decided to employ a document class. This did not work either. It seems it only works for frame one but not the rest of the frames (I have four).
How can I have frame four respond to something that happpened in frame one? For example, if the player achieves something in frame one, I want a movie clip in frame four to be visible.
If You are writing your code on the timeline, My suggestion would be to create two layers in the timeline, one for 'frame-actions' - in this layer you insert the code specific to a single frame (will work when the movieclip is stopped on that particular frame).. And also create one more layer called global-actions (for the entire timeline). Only the first frame will be a key frame and there should be empty frames till the end of the timeline.
In this layer actions write the code that you want to access from any keyframe in the same timeline.
If you define a variable in the actions which are written for the whole timeline (global-actions) then that will be available on all the frames.
Now if you want to go to a different frame based on some action, just write some functions in the layer which contains global actions and call that particular function through the frame actions. To go to a different frame use the 'gotoAndStop(frameNumber)' function of flash.
I want to tell you that while it will work, I would not recommend using it in this way.
HTH.
You can use static variables - these are variables which are linked to a class, rather than an instance of it.
Suppose your document class was called Document.as, and you wanted a variable, playerLives, to be visible from any part of the program.
Declare it inside Document.as:
public static var playerLives:int = 3;
You can then reference this directly from anywhere else in your code with:
Document.playerLives
(note that the variable is a member of the class itself, not an instance of it).
You could use a dedicated Statics class to hold these variables if you want to keep your document neat, or attach them to the relevant classes (eg Player.lives)
I've not used timeline/frames for some years but I believe this is how I used to do it!
NB Statics will be fine for your purposes but they are, in some ways, an equivalent to the _global variable in AS2 (at least, they can be used in the same manner) - many would not approve of their use, or over-use, as they are freely accessible from anywhere in your program (thus anathema to the OO concept of encapsulation), but personally I try not to worry about it in small cases - the most important thing to know about the rules of any design pattern is when they can be broken!
They are also slightly slower to access than instance members, but you won't notice this unless you are constantly accessing/changing them (making things like player velocity, which will need to be referenced/changed every frame, static, is not a good idea).
Hope this helps.
You may find the simplest way to link everything with the document class is to move your four frames into a movieclip together and have that on the first frame, then interact with that movieclip.
E.g. in the document class, where the movieclip instance on the timeline is called 'game'.
game.gotoAndStop(4);
game.objectToDisplay.visible = true;
If you encounter reference errors in the IDE then you can avoid these by using [] notation to refer to the properties of game, e.g. game["objectToDisplay"].visible = true;
Note that it's not really best practice to do this, but it will at least help you to finish that first game which is really more important at this stage in your learning. Afterwards, if you want to learn more then I'd recommend "The Essential Guide to Flash Games" by Jeff Fulton from 8bitrocket.com - it will teach you how to use the document class effectively.

Minimize memory usage with a multi-frame MovieClip

I have a MovieClip in Flash with 100 frames. Each frame contains a certain icon I need to use in a project. I create instances of this icon MovieClip wherever I need an icon to appear, and gotoAndStop to a certain frame to display that icon.
Will storing a 100 icons in a single movieclip cause every single icon to be created in memory whenever I create an instance of the MovieClip? If I stored each icon in the library and attached only the icon that is needed, would that consume less memory than creating this MovieClip that has all the icons in it?
To answer your question: When you go to a frame with an icon on it, Flash will create a new instance of that icon. When you leave that frame, Flash will make that icon eligible for garbage collection, unless you force it to hold the instance in memory somehow, such as using addEvenListener where the method that is the listener is the icon or somewhere inside it.
I think the memory usage is likely to be higher for the goToAndStop vs. the new instance. If you are not experiencing problems with goToAndStop(), you are unlikely to experience additional issues by instantiating a new icon each time you switch from one icon to another. The other people who have answered are quite right that you will use fewer CPU cycles by instantiating all the icons only once (by whatever method), and then simply using the same one every time you use that icon. However, your overall memory footprint will be higher, because you will have all of the icons (even ones you are not currently using) in memory all the time.
If you want to go the route of only instantiating each on once, I'd suggest you go with Lazy Loading, where you only instantiate each icon when it is first used. One way to do this is to use what you already have and visit the frame the first time you want to use a specific icon, then store the BitmapData or a reference to the icon itself after that and reuse it. Another way is to build a swc and use a similar pattern.
None of this requires a static variable BTW, since it sounds like you're not using tons of different copies of your icon MC. Even if you are, it's probably better to handle referencing the icons you have through dependency injection.
I think you are probably asking about file size, however, vs. actual memory usage. The answer to that questions is that all assets that are used by your fla get compiled into the swf, regardless of whether they are in a MovieClip, a SWC, or in the library with Export for Actionscript in Frame N checked.
I get it, I've got boobs, I'm guessing
Try this, to verify my "guess."
Create a swf that has a keyframe on frame 1.
Draw a circle there.
Put another keyframe on frame 10. Draw a square.
In Publish Settings, check "Generate size report."
Now, you know (or you should know) that this swf can display the circle on frame 1, even if the assets compiled on frame 10 have not yet been downloaded. So, is there any possible way that the square could be loaded into memory before Frame 10 has been downloaded? Hint: the answer is no.
Now ask yourself this: Do you think Macromedia wrote a special version of MC that is incapable of lazy loading that the MC that is the main Document Class that Flash generated for the movie you made above so obviously handles so well?
The Macromedia engineers did a lot of things that in hindsight look pretty stupid, but they're not that incompetent.
If each icon that goes into the 100 frames MovieClip is an image you can easily enough export that image for actionscript and access only the image you need.
Another idea is to create a static class that stores each icon frame as a symbol in a static array(if you want to access the icon by index) or Dictionary(if you want to retrieve it by name).
e.g.
package{
import flash.display.Bitmap;
public class Icons{
public static const assets:Vector.<Bitmap> = Vector.<Bitmap>([new Icon01ASExportName(),//instance bitmaps once
new Icon02ASExportName(),//reuse them later multiple times
...
new Icon99ASExportName()]);
}
}
Yet another is to generate a SpriteSheet, either once and save it as a file, either at runtime, then copyPixels() to paste the icon from the right rectangle of the main spritesheet. You've got multiple tools available for generating the spritesheet:
Zoƫ
SWFSheet
Flash CS6
TexturePacker (if you have movieclips, you can export them as an image sequence to drop into the software)
There are also ways to generate a spritesheet at runtime.
For some nicely explained video tutorials on spritesheets and BitmapData by Lee Brimelow: Sprite Sheets and Blitting - Part 1,2 and 3.
If your icons are Bitmaps then each movieclip will share the same data - the only memory increase will be due to more movieclips, rather than their contents (which shouldnt be a problem).
If your icons are not Bitmaps (other movieclips, shapes, buttons etc), then everything gets duplicated, so memory usage will increase a lot faster as more movieclips are added.
Another consideration is that lots of movieclips on screen will have more of an effect on FPS than simpler objects so you may want to consider adding the icon itself rather than the movieclip even if you are using Bitmaps.

Instantiate MovieClip with its layers?

I have searched a lot lately, I found that I can load an external swf file to my haxe project at compile time, and use the movieClips via their IDs as Classes types ..
That's cool& nice, but how it would work when I instantiate a MovieClip that has layers?
What I have is MovieClips with layers, layer of image, and a layer of text over the image layer.
So, is this achievable? will I be seeing instances of movieClips (images& texts over them) ?
Let's first dissect the flash terms in terms of code,
Layers :
Consider the layers as grouped z indexes. A single layer with multiple objects will also assign sub z indexes to each clip (see the send to back option in context menu)
The flash IDE provides you a nice interface to group & lock a few instances of objects, while working on the other. Each such group is a layer.
At runtime, every layer (with it's objects) will be concatenated into one single list (the display list) and the objects placed onto the stage in order of their position in the list.
Frames :
Frames are slices of time controlled by the fps property of the flash player. So if you have 12 fps, it means that whatever code you put in that frame of time will receive attention for 1/12th of a second.
Of course there is code being written for anything you do inside a frame or layer. The code, for example that you put in manually in IDE is added in by calling the addframescript internally.
MovieClip
The MovieClip class itself is actually an IDE related thing. The split being Sprite Class plus a timeline (collection of frames) plus associated properties & methods.
The Movieclip class thus provides you with properties like currentFrame, totalFrames, etc & methods like gotoAndPlay(), stop(), etc to interact with the flash controlled class.
Considering all of the above together, you should expect to see different images/text/objects at different intervals of time as defined in the frames when you access the movieclip.
I haven't spent a lot of time in flash, but the answer to your question comes from having a better understanding of how Flash/AS3 works, and not so much to do with Haxe.
Quoting this answer:
Layers only exist in the Flash IDE. They are not part of Flash Player's display list system. So you can't specify what layer a child goes into. Use addChild() or addChildAt() to add children to containers.
So that means Haxe will not have any concept which object is in which layer, nor would AS3 for that matter. The correct approach seems to be to use containers with IDs, which are recognised by AS3 or Haxe, and then add children to the containers.
TLDR: use containers movie clips, not layers.

symbols placed on the timeline become undefined if stepping backwards

I am using the frames in the timeline of a .swf as pages in a flash app. The user can advance to the next page by clicking a button that takes her to the next frame. Similarly, it is possible to navigate to the previous frame/page as well.
Most of the content is placed on the stage (i.e. created by dragging an instance of a library symbol to the stage) but properties of those instances, such as .visible might be changed via actionscript. Also, some objects are loaded from external flash files and displayed programmatically with addChild / addChildAt.
The problem is, if I am on Frame N+1 and there is an object displayed on the stage programmatically (i.e. with addChild, not by having it placed on the stage) and navigate to Frame N where there is an object that is placed on the stage (i.e. dragged from the library),
then the instance of that object is undefined/null and throws an error if I try to set its properties (like .visible).
The error does not occur if I am moving to the NEXT frame, only if I am moving to the PREVIOUS one. Therefore I assume that some kind of initialization is not getting called while going one frame back.
I was also thinking that the objects would just not "live" to the next timeframe, that is, their value would be lost and re-initialized because of scope, but if there is no dynamically created object on the stage, I can navigate back and forth just fine.
Is there a way to ensure that the objects created on the stage do not disappear while navigating back to the previous frame?
The first, and more useful, part of the answer is this: timeline keyframes and scripts can give conflicting information about display objects - whether they should exist, where they should be, and so on. For example, when you add an item by playing into its frame, and then delete it with script, and then play into its frame again. When this happens, there's no unambiguously correct thing for Flash to do, so it tends to be unpredictable. I believe what generally happens is that once you fiddle with a given object via script, it's considered to no longer pay attention to the timeline - but your mileage will vary.
Having said that, the reason things are different when you play backwards is the second and more arcane part of the answer. Internally Flash functions differently when seeking forward and backwards on the timeline. Flash internally treats keyframes as changes to be applied in the forward direction, so as you play forward, it applies those changes in sequence. When you move backwards, however, from frame N+X to frame N, it doesn't scan through the intervening X frames reversing those changes - it jumps back to frame 1 and fast-forwards along to frame N. Normally, it amounts to the same thing and you don't need to worry about it, but when you get into the twitchy area where scripts and the timeline have a different idea of what should be on the stage, you're liable to see things behave differently depending on which way you jump (as you are now).
The super-short version is, for things to work predictably, try to ensure that any given object gets added, updated, and removed the same way - either all via script, or all via the timeline. When that seems impossible, fiddle with your content structure - usually, the best solution is to change your object into two nested ones, so that the things you want to do with script occur one level higher or lower than the things you want to do with the timeline.
I'm not sure I got your question right, but as3 does not instantiate elements on the timeline as soon as you gotoAndSomething, but later that frame.
That is, you can't
this.gotoAndPlay(10)
this.elementOnTimelineFrame10.DoSomething()
without errors.
I remember using this chunk of code in the past to work around this problem. It uses the Stage.Invalidate() function to wait for an Event.RENDER before trying to access and children, more info (although vague as hell) is here
private function init():void
{
stage.addEventListener(Event.RENDER, stage_renderHandler);
}
private function stage_renderHandler(evt:Event):void
{
// Run your code here
updateChildren();
}
private function enterFrameHandler(evt:Event):void
{
// triggers the RENDER event
stage.invalidate();
}
This also might me very costly (performance wise). I would strongly advise against dynamically adding/removing objects to an existing timeline, is there any way in which you can place an empty Sprite above the timeline animation and use that for all your dynamic content?
Hope this helps