IOS Touch latency problem.

mehmetuysal

Hi,

I am developing my new game with Gideros after 3 games in Corona. I gave up Corona because of android performance issues.

I have tested game on android tablets and some android phones, performance is really fine. But, when I started to test on iPhone, there is observable touch latency problem. Android can detect touch perfectly, but ios detects in half second. I have used EVENT.MOUSE_DOWN, EVENT.MOUSE_UP events. Is there any difference between "touch" and "mouse" events ? Please help me, I have finished everything with gideros. I like that very much. So I do not want to move to other development platforms and rewrite code.

ar2rsawseen

@mehmetuysal do you use latest Gideros version?
I don't remember since which version, but there was an optimization for mouse/touch events.

Earlier mouse events were for desktop and touch events for mobiles, now you can select in project properties to dispatch events on both. So there should not be any huge difference.

One thing to try you can put event:stopPropagation() in all you touch/mouse events to stop events and prevent them form dispatching on other elements.

mehmetuysal

I am using v2012.09.9. I think it is latest version.

I thought, converting mouse events to touch event can cause performance issue so I changed mouse events to touch. But both has same problem.Also I'm using event:stopProgation().

Is there any other way to optimize touch events performace ?

ar2rsawseen

@mehmetuysal not that I know of.

well the problem is, that I don't know any other who would experience performance issues with touch events on IOS platform.

Is it through the Gideros player or exported app? Maybe you have older Gideros player?
Maybe it's some specifics you do inside touches, what exactly you do in them?

mehmetuysal

I am building game like diamond dash. 8x7 touchable boxes on screen.
I have tested in gideros player, android tablet and an old android phone with player and built apk. All has perfect performance.

With gideros player(on device) and exported built has poor touch performance on IOS device. I have tested on Xcode native ios simulator now. That works well because of computers CPU.

If issue causes because of specifics inside touches, same problems should occur on android devices too ?

function Engine:onMouseDown(event)   if(game_paused==false) then if(self:hitTestPoint(event.x,event.y) and idle == false) then self.isFocus = true ... ... .. ... event:stopPropagation() end else event:stopPropagation() end end

mehmetuysal

In gideros

function Engine:onMouseDown(event) print('test') end

writes 'test' more than 50-60 times. So I guess that checks all objects on the screen, until find the correct one

in corona

function Engine:onMouseDown(event) print('test') end

writes 'test' 2-3 times.

I think this can cause the problem. Does gideros needs some core code optimization ? or am I doing something wrong. Is there anything to check in my code ?

bowerandy

@mehmetuysal, mouse events are sent to all objects on the stage that register for them. They are not specific to whether the event actually occurs inside the object bounds or not.

I have not experienced any latency issue with Gideros and iOS. Are you using a particularly old (i.e. slow) device? There should be no problem searching 56 objects for a touch location - this should occur almost instantly.

Are you sure there are only 56 objects on the stage? Previously people who have experienced performance issues have inadvertently been adding far more objects than they think they have.

Also, your code looks slightly odd to me. Is Engine a class of which there is a single global instance or are there 56 of them. Perhaps you are registering all 56 objects with your single engine class? If this is the case then your onMouseDown() function is hit testing against the engine and not the target of the touch.

best regards

ar2rsawseen

@mehmetuysal no this is how the things are supposed to be in Gideros.

Basically touch/mouse event is called on all objects that registered to it. Then inside it you can check using hitTestPoint was this touch meant for specific element.
This gives you a flexability of, for example, increasing touch area of small images/buttons, or checking that specific image on top or bottom received the event.

but once you are inside if self:hitTestPoint ...
then it means you have found what you needed and you can call event:stopPropagation() so it will not be dispatched further.

I hope @atilim could shed more light on the issue, in the meantime, it might be useful to prepare a small example of a slowdown that you could share with others to test.

mehmetuysal

@bowerandy
I have box and engine classes. I am creating 56 boxes from box class and add to engine sprite class. I am adding touch/mouse eventlistener for all boxes. I didnt create touch events in box class, because box needs interaction with other boxes. I am counting max 30-35 touch calls until hittest returns true.

May be my code needs optimization. But as I said before, I just want that should work like on my andorid device( 2 years old HTC Desire). I am testing on IPhone4. I will test on blank project only for touch performance.

Thank you very much for your answers

bowerandy

@mehmetuysal, okay. In that case maybe you should be hit testing against event.target rather than self in your above code? It seems to me that in the above, self is always the engine. The event.target field should have the box object that received the event. Does this make any difference?

best regards

atilim

I also think @bowerandy is correct. hitTestPoint function is called on whole Engine object 56 times.

mehmetuysal

Yes,I think that checks whole stage child sprites. So what am I supposed to do ?

@bowerandy, I change to event.target. Nothing changed.

Android is ok, but ios still has noticiable latency. I will test with blank project tomorrow. I hope I will solve the problem.

atilim

Also playing an .mp3 file with each touch can cause such latency. If you're doing so, convert your .mp3 file to .wav.

bowerandy

@mehmetuysal, try the following code in a main.lua and, once you've validated that it works okay, adapt it for your app:

local BOX_COUNT=100   MyBox = Core.class(Sprite)   local engine   function MyBox:init(size, id) self.id=id   local shape=Shape.new() shape:beginPath() shape:setFillStyle(Shape.SOLID, 0xff0000) shape:moveTo(0, 0) shape:lineTo(size, 0) shape:lineTo(size, size) shape:lineTo(0, size) shape:closePath() shape:endPath() self:addChild(shape)   self:addEventListener(Event.TOUCHES_BEGIN, engine.onTouchBegin, engine) self:addEventListener(Event.TOUCHES_END, engine.onTouchEnd, engine) end   Engine=Core.class()   function Engine:init() engine=self for i=1, BOX_COUNT do local x=math.random(application:getContentWidth()) local y=math.random(application:getContentHeight()) local box=MyBox.new(30, i) stage:addChild(box) box:setPosition(x, y) end end   function Engine:onTouchBegin(event) local target=event:getTarget() if target:hitTestPoint(event.touch.x, event.touch.y) and not(target.touchId) then -- Remember the touch id as a focus indicator. -- This allows you to track different fingers if you need to. target.touchId=event.touch.id print("pressed box", target.id) target:setScale(0.5) event:stopPropagation() end end   function Engine:onTouchEnd(event) local target=event:getTarget() if target.touchId==event.touch.id then print("released box", target.id) target:setScale(1) target.touchId=nil event:stopPropagation() end end   Engine.new()

I tried increasing the BOX_COUNT to 10,000 and, although the creation process takes 2 seconds, the touch latency is imperceptible (fine work @atilim). I am using an iPad 3 though, so slower devices may not perform quite so well.

best regards

mehmetuysal

@bowerandy this code works exactly like mine. This has no latency. So I rechecked everything at my code. I have 2 sounds playing after a box touched. After I've removed sounds, latency has gone. Than I have increase BOX_COUNT to 300 and added my 2 sounds to @bowerandy test code, then problem come back.

As a result, sounds caused my problem. I am using mp3. What am I supposed to do for better sound performance? I attached test code with sounds.

thank you all, for your helps

GregBUG

I think is better to use wave format for sound effects; if I remember correctly mp3 is better for background music (because mp3 is more CPU intensive than wav).

Hope it helps.

bowerandy

@mehmetuysal, see @atilim 's (and @GregBUG 's) comment above. MP3 sounds will cause such latency. Convert them to WAV and you should be okay.

best regards

mehmetuysal

mp3 was the problem. I convert mp3s to wav then latency has gone. You saved my life. Thank you all.