Hi everyone! Has anyone encountered the same bug/behavior?
Steps to reproduce problem:
- First of all my every scene is replaced with transition (CCTransitionFade);
- I have RootScene, SceneA and SceneB classes.
- I call pseudocode: [ccdirector replaceScene:transitionScene1WithSceneA]; // let this transition be named "transition1"
- Then I want to wait until transition is finished.
- When transition is finished I want to call another [ccdirector replaceScene:transitionScene2WithSceneB] // let this transition be named "transition2"
So I override SceneA#onEnterTransitionDidFinish method:

// SceneA class
-(void) onEnterTransitionDidFinish {
    [super onEnterTransitionDidFinish];
    [CCDirector sharedDirector] replaceScene:[CCTransitionFade transitionWithDuration:transitionDuration_
                                                                                                                           scene:[SceneB node]
                                                                                                                     withColor:ccc3(0, 0, 0)];
}

setNextScene with transitions works in several stages.. Between them replaceSceen from transition is called.
But replaceScene with transition in onEnterTransitionDidFinish causes the problem in CCDirector#setNextScene:

-(void) setNextScene  //! Is called when transition1 is finished (runningScene_ = transition1
                                    //! and nextScene_ is an instance of SceneA class
{
	Class transClass = [CCTransitionScene class];
	BOOL runningIsTransition = [runningScene_ isKindOfClass:transClass]; //! YES
	BOOL newIsTransition = [nextScene_ isKindOfClass:transClass];            //! NO

	// If it is not a transition, call onExit/cleanup
	if( ! newIsTransition ) {
		[runningScene_ onExit];      //! <---- here is -onEnterTransitionDidFinish message to my instance of SceneA class is send
                                                          //! so you remember I have called -replaceScene: in onEnterTransitionDidFinish and now
                                                          //! nextScene_ is transition2 scene.

		// issue #709. the root node (scene) should receive the cleanup message too
		// otherwise it might be leaked.
		if( sendCleanupToScene_)
			[runningScene_ cleanup];
	}

	[runningScene_ release];

	runningScene_ = [nextScene_ retain];
	nextScene_ = nil;

	if( ! runningIsTransition ) {
		[runningScene_ onEnter];
		[runningScene_ onEnterTransitionDidFinish];
	}
}

When transition2 is finished i have RootScene present at the screen, SceneA present at the screen doing animations (onExit was not called) and SceneB at the screen too.
If I call replaceScene with delay everything is ok.
The question is it an expected behavior of replaceScene in onEnterTransitionDidFinish? How can I know when transition is completely finished and I can call replaceScene: without worry of crashing and wasting time for waiting?
Will appreciate any help :)

Thanks alot! Apperently you cant replace a scene in the initmethod and when i moved it to be called after init was done everything worked out. Its strange though cause i do recall i had problems before i put it in the init.

@MrAzulay,

If you put the replaceScene in onEnter or onEnterTransitionDidFinish, it will probably work like you are wanting it to:

-(id) init
{
        if( (self=[super init])) {
    }

    return self;
}

-(void) onEnter//TransitionDidFinish
{
   [super onEnter/*TransitionDidFinish*/];
   [[CCDirector sharedDirector]replaceScene:[GameLayer scene]];
}

-Mark

@zaaroth : It used to be in a IBAction, i just stripped everything down too make it easier for people to see the code.
It seems to work without that error now but the sprites wont draw out when the scene is called with replacelayer but from the appdelegate it works :S

You are calling replace scene on a init method. It may not be wrong from a tech perspective, but sounds odd nevertheless. I dont see the point of calling replace scene on the init that will be ran in order to create another scene.

I get SIGABRT on this row:
[scenesStack_ replaceObjectAtIndex:index-1 withObject:scene];

I tried to figure this out for hours but dont understand it
Loading the scene with appdelegate works without any problem but with replaceScene i get this problem.

http://pastebin.com/wCyEPwfw
This is all the code that exists in the layer that tries to replace the scene.

Pleease help!

Hi,
just wanted to say that the order of execution changes if you are using a transition between scenes or not..

i.e. if you are going to SceneTwo from SceneOne:

**Without a transition
--SceneTwo init called
-SceneOne onExit called
-SceneOne dealloc called
--SceneTwo onEnter called
--SceneTwo onEnterTransitionDidFinish called

**With a transition
--SceneTwo init called
--SceneTwo onEnter called
-SceneOne onExit called
--SceneTwo onEnterTransitionDidFinish called
-SceneOne dealloc called

So, as far as I know, if you are using a transition between scenes, you should be using onEnterTransitionDidFinish of the new scene for stuff you are waiting to be completed from onExit of the previous scene.

-Mark

I would make sure riq gets to know about this and your fix. register a bug?

Ok, I found the solution. I hope it's right.
Change your onExit method as below in CCLayer.m.
Check if the Delegate of UIAccelerometer is self first. If it isn't, leave it alone. Otherwise, set the delegate to nil as original code.

-(void) onExit
{
	if( isTouchEnabled )
		[[CCTouchDispatcher sharedDispatcher] removeDelegate:self];

	if( isAccelerometerEnabled )
		//[[UIAccelerometer sharedAccelerometer] setDelegate:nil];
		if ([[UIAccelerometer sharedAccelerometer] delegate] == self) {
			[[UIAccelerometer sharedAccelerometer] setDelegate:nil];
		}

	[super onExit];
}

I know the reason.
There are 2 screens. One is old and going to disappear, another is new and going to be shown.
The new one enter, the method is called first.

-(void) onEnter
{
	// register 'parent' nodes first
	// since events are propagated in reverse order
	if (isTouchEnabled)
		[self registerWithTouchDispatcher];

	// then iterate over all the children
	[super onEnter];

	if( isAccelerometerEnabled )
		[[UIAccelerometer sharedAccelerometer] setDelegate:self];
}

Then the old screen is called OnExit second.

-(void) onExit
{
	if( isTouchEnabled )
		[[CCTouchDispatcher sharedDispatcher] removeDelegate:self];

	if( isAccelerometerEnabled )
		[[UIAccelerometer sharedAccelerometer] setDelegate:nil];

	[super onExit];
}

So the onExit method clears the UIAccelerometer.
But I don't know how to figure it out.
I have the same problem.

So I found a temporary solution for now, which is still scary that it is needed.

- (void)ContinueButtonTapped:(id)sender {
	self.isAccelerometerEnabled = NO;
	[[CCDirector sharedDirector] replaceScene:[CCCrossFadeTransition transitionWithDuration:0.5 scene:[Space3Scene scene]]];
}

as you can see I added self.isAccelerometerEnabled = NO; to my code anytime I was about to move from one scene to layer that had it enabled to another. when i move to the main menu or game over scene i don't need that because by default it is off so is getting the self.isAccelerometerEnabled = NO; when that loads. This is very strange that this happened out of no where but at least I can continue on for now.

If you figure it out please let me know, if not at least I have it working again. :)

anyone have any thoughts?

this is where i put the self.isAcc... you know :)

-(id) init
{
	if( (self=[super initWithColor:ccc4(0,0,0,0)] )) {

		[FlurryAPI logEvent:@"Space Demo Started"];

		CGSize winSize = [[CCDirector sharedDirector] winSize];
		self.isTouchEnabled = YES;

		ballsLeft = 10;
		MenuExist = FALSE;
		blockerBall = 1;
		blockerExist = FALSE;
		/*
		CCSprite * bg = [CCSprite spriteWithFile:@"Blackblue.png"];
		bg.position = ccp(160, 240);
		[self addChild:bg z:-1];
		*/
		// enable accelerometer
		self.isAccelerometerEnabled = YES;

like that, and then it continues

- (void)accelerometer:(UIAccelerometer *)accelerometer didAccelerate:(UIAcceleration *)acceleration {
  //do stuff in here
}

and then that is how i use the accelerometer. do i have to call an update on it like this kind of thing?

[self schedule:@selector(tick:)];

notice you have to set that in your layers not in your scenes, are you doing that?
Also are you implementing the method for receiven the accelerometter events and setting the update interval in each one?

i do. and i do it for each scene. each new file, a .h and .mm, is a scene, and i change scenes between each level like so:

[[CCDirector sharedDirector] replaceScene:[CCCrossFadeTransition transitionWithDuration:0.5 scene:[Space3Scene scene]]];

to move to the next level.

I have the self.isAccelerometerEnabled = YES; on each one in the init but it does not seem to work past the first level that uses it

Sorry i didn't really understand your problem.
You should set self.isAccelerometerEnabled = YES; for each layer you want to receive accelerometter events.

I think that by default it is off.

I am having a strange issue where as i move from one level to the next (scene) once one has been accelerometer enables, all the ones following can't be. I need in all, and if i disable in level 1, level two is and thats it, and so on.

any ideas?

self.isAccelerometerEnabled = YES;

thats what i use

Thanks for all your help guys! I finally solved thanks to @Lam's tip. Playing around with the posVar got it working!

@degborta
I think positional variance would help.
You should try Mark's approach but instead of changing speed, change the emitters posVar.

Cache the old posVar during onEnter, set the new posVar to a large number, maybe the screen size.

Then onTransitionDidFinish reset the posVar.

It also might help to speed the emission rate. Since pos var let's it emit particles in a larger region, younow need to emit faster.

So change position var and emission rate , then reset during a finished transition.

AkumaStreak: Can you explain further what you mean? I have tried playing around with the step- only, but it seems nothing happens at all. Couldn't find the update method on the emitter.

Maybe I am doing something completely wrong.

I also vote that you should run update with a white lie delta time (one way or another) in order to effectively initialize them the way you want.

@badawe yes!

Guys i have one question about particles, can i have more than one particle emitter positions?

Like i have 4 finger imputs and in all of then i want some particles, so i need 4 particles emitters? thats right?

Thanks for your input, I have tried both of your ideas now, but neither seems to solve my problem.

I set the speed to 100 and schedule a timeout for 3seconds on init, in the timeout I reset the speed to 20, as it is supposed to be.

But the problem here is that all particles that were created when I had the speed set to 100 keep falling at that speed, so they all quickly fall down below screen and disappear.

I absolutely have no idea as to what I can do to solve this. Any ideas are appreciated.

I haven't tried it, but maybe you could start the particle system in 'onEnter' with a really fast speed, then in 'onEnterTransitionDidFinish' slow the particle system down to normal speed.

dunno ..

whoops, I meant step, not update. So yes I've tried looping as well, big and small values for both ways, for some reason it just does not take; I debugged a little and everything seems ok... but xcode is being flaky (debug values are sometimes not there or zero) since I updated the sdk.

What if you called -step:4.0 instead? Or called -step:(4.0/kFPS) in a loop for (4.0*kFPS) times where kFPS=30.0 or whatever your frames per second is.

It makes sense to me that if you created any ParticleSystem and manually called update on it with a value of say 4 seconds that it would then appear to have "run" for 4 seconds, however I did a quick test and could not get it to work right (aka it had no effect). I did something similar to this:


ParticleSystem *ps = [ParticleSnow node];
[ps update:4];
[self addChild:ps];

Something like this should work, but it apparently needs something more that I'm missing... maybe I'll get back to you.

In your subclass of your ParticleSystem you could add the ability to save and load its state, ie serialize it, to a file. Make a version where you let it run a bit, save its state, and in the game you load that state as your starting point for the system. Tricky but doable.

I am playing around with the particle examples and using it on one of my layers in my testproject.

Everything is working great but I have a question.

In my project I have a Layer that fades into another Layer (the one that should have the snow on it). I am adding the snow particles on init in the second layer.

When the transition between the scenes is finished, the snow particles haven't reached that far. They are on the top third part of the screen. It takes a second or two before they reach bottom.

This makes it pretty obvious that I just started the snow animation. I would like it to cover the whole screen with particles, so when the transition finishes it appears it has been snowing for a long time. But I am clueless to what I should do or look into.

Any tips or ideas are greatly appreciated.