The function removeTween in Tweener has:

if (ft._prev != null)
{
    ft._prev._next = ft._next;
}
else
{
    if (ft._next != null)  //  Here is the issue?
        _tween = cast(ft._next, Tween);
}

That null check means that _tween will not be null when there is not a next tween.

If you look at addTween, it does:

ft._next = _tween;

So, the infinite loop occurs if you have one tween added, removed, then added a second time.

var testTween:Tween = new Tween(blah);

//  ft._next == null since _tween == null  AND _tween is later set to reference the test tween.
addTween(testTween);

// ft._prev != null evaluates to false
// else
// ft._next != null evaluates to false
// _tween is never set to null so the tween continues to update after being removed
removeTween(testTween);

// ft._next == _tween == testTween.  Infinite Loop during updateTweens.
addTween(testTween);

Hopefully that made sense. I would submit a fix but I have a development version of haxe right now...

Possible solutions:

1. Simply remove the null check.
2. If cast throws an error:

_tween = ft._next == null ? null : cast(ft._next, Tween);

Hi Matt,

NME 3.3.2 handles the mapping of classes a little bit differently. Since Flash and HTML5 do not support joysticks right now, the JoystickEvent class is not available.

Do you think that we should be handling this with a silent failure, so you can compile applications but receive no joystick input, or do you think it should be more strictly enforced, where it throws a compile error if you try to use a joystick while targeting a platform that does not support it?

If you think this should be a silent failure, we can probably update NME to handle it that way (as it was before). Otherwise, would you mind updated your code so it includes joystick support "#if (nme && (cpp || neko))" instead of "#if nme" (or something like that)?

Thanks!

_field is not always a child of _parent.

The issue appears when the destroy function is called:
https://github.com/HaxePunk/HaxePunk/blob/master/com/haxepunk/graphics/Text.hx#L155

I have 'hacked' a fix by doing:

if (_parent != null)
{
_parent.removeChild(_field);
_parent = null;
}

I haven't had time to look into the issue beyond a quick fix so i am not sure if this might cause an issue elsewhere. I'll submit a pull request when I can.

Hi,

Found a bug when you change the scale of the screen, say HXP.screen.scale = 2;, then, if you resize the flash window, a error is thrown :

ArgumentError: Error #2015: Invalid BitmapData.
    at flash.display::BitmapData/ctor()
    at flash.display::BitmapData()
    at com.haxepunk::HXP$/createBitmap()[C:\Users\bender\dev\nme\HaxePunk102\src\com\haxepunk\HXP.hx:810]
    at com.haxepunk::Screen/resize()[C:\Users\bender\dev\nme\HaxePunk102\src\com\haxepunk\Screen.hx:65]
    at com.haxepunk::HXP$/resize()[C:\Users\bender\dev\nme\HaxePunk102\src\com\haxepunk\HXP.hx:210]
    at com.haxepunk::Engine/resize()[C:\Users\bender\dev\nme\HaxePunk102\src\com\haxepunk\Engine.hx:220]
    at MethodInfo-209()[C:\Users\bender\dev\nme\HaxePunk102\src\com\haxepunk\Engine.hx:186]

Found out it tries to create a BitmapData with negative size. I think it's because of this 2 lines in HXP.resize() :

width = Std.int(width / HXP.screen.fullScaleX);
height = Std.int(height / HXP.screen.fullScaleY);

When a PreRotation graphic is rotated it is not rotated around the center of the image. I believe the best way to fix this is to translate the image to the correct center point when rotating it.

For some reason the null check if (options) doesn't work. Once you replace it with if (options != null) everything works. Everything worked fine targeting flash, it wasn't until I tested the windows target that I got this issue.

HXP.hx line:854 is

if (options)

The line should be

if (options != null)

Specifically I was seeing the on complete function not get registered and with this change it's properly registered.

When attempting to use the angle or scale-properties that leads to matrix transformations, this puts an Entity with graphic = Image to 0,0 and does not do any transforming at all.

It's a straight port, not sure why it's missing. Here it is:

/**
 * Immediately stops the Tween and removes it from its Tweener without calling the complete callback.
 */
public function cancel()
{
    active = false;
    if (_parent != null)
    {
        _parent.removeTween(this);
    }
}

Subj. Looks like refactoring isnt' finished.

When using Image class constructor to crop an image asset, the resulting instance has the size of the source asset. This seems to cause centerOrigin() to not work properly.

While working thru the tutorial (103) and having just added the ship, I added -debug while running under Neko. I clicked on the green box in the upper left & I got a crash with the following stack trace. Doing the same thing under Flash causes no problems.

Called from sys.io.FileOutput::$statics line 1
Called from ApplicationMain::main line 39
Called from nme.Lib::create line 77
Called from native.Lib::create line 73
Called from native.display.Stage::nmeProcessStageEvent line 382
Called from native.display.Stage::nmeDoProcessStageEvent line 292
Called from native.display.Stage::nmePollTimers line 718
Called from native.display.Stage::nmeCheckRender line 247
Called from native.display.Stage::nmeRender line 731
Called from native.display.DisplayObjectContainer::nmeBroadcast line 141
Called from native.display.DisplayObjectContainer::nmeBroadcast line 141
Called from native.display.DisplayObjectContainer::nmeBroadcast line 156
Called from native.display.DisplayObject::nmeBroadcast line 165
Called from native.display.DisplayObject::nmeDispatchEvent line 179
Called from native.events.EventDispatcher::dispatchEvent line 80
Called from native.events.Listener::dispatchEvent line 201
Called from com.haxepunk.Engine::onEnterFrame line 250
Called from com.haxepunk.debug.Console::update line 448
Called from com.haxepunk.debug.Console::updateDebugRead line 927
Uncaught exception - Invalid field access : toString

When running under Neko, HXP.random should return values from 0 to 1, but sometimes returns negative values:

    HXP.randomSeed = 0x20000000;        
    trace(HXP.random); // returns -0.249998043873346

This seems to occur when using Std.int() to convert the float value back to int. I don't think that this is necessarily a great solution, but adding 1 if it goes negative brings the value back into range.

    /**
     * A pseudo-random Float produced using FP's random seed, where 0 <= Float < 1.
     */
    public static var random(getRandom, null):Float;
    private static inline function getRandom():Float
    {
            _seed = Std.int((_seed * 16807.0) % 2147483647.0);
            var result = _seed / 2147483647.0;
            return (result < 0 ? 1 + result : result);
    }

    /**
     * Returns a pseudo-random Int.
     * @param       amount          The returned Int will always be 0 <= Int < amount.
     * @return      The Int.
     */
    public static inline function rand(amount:Int):Int
    {
            return Std.int(getRandom() * amount);
    }

Here is the commit:

https://github.com/MaskedPixel/HaxePunk/commit/4c22663853066dfece93662f0389991e7fa1b9f7

See the description for commit details.

I would submit a pull request except I have several commits from the main repo from when I merged. I see every other pull request is 'clean' and don't know how to do that because git.

EDIT: Nor can I figure out what to google for.

EDIT: Checking the source here I see that I don't have the same version and there shouldn't be a problem with the collide. Must have made a mistake when I tried to update.

Recycling works with this change to the create function,

    /**
     * Returns a new Entity, or a stored recycled Entity if one exists.
     * @param   classType       The Class of the Entity you want to add.
     * @param   addToWorld      Add it to the World immediately.
     * @return  The new Entity object.
     */
    public function create(classType:Dynamic, addToWorld:Bool = true):Entity
    {
        var className:String = Type.getClassName(classType);
        var fe:FriendEntity = _recycled.get(className);
        if (fe != null)
        {
            _recycled.set(className, fe._recycleNext);
            fe._recycleNext = null;
        }
        else fe = Type.createInstance(classType, []);
        var e:Entity = cast(fe, Entity);
        if (addToWorld) return add(e);
        return e;
    }

I seem to be having an issue with spritemap not displaying any data to the screen, the entity is there, however the image is not. :(

Here is the code I am using, pulled from the tutorial on the website. This code is in my entities added() function.

    _sprite = new Spritemap("gfx/ship.png", 18, 18);
    _sprite.add("idle", [0]);
    _sprite.play("idle");
    this.graphic = _sprite;

Just for good measure, the image data appears appropriately when using this code instead.

    this.graphic = new Image("gfx/ship.png");

I am using haxe 2.10, nme 3.4.0, and HaxePunk 1.6.6 from haxelib.

When viewing a scene in fullscreen the masks do not rendering over the objects like they should. It needs to take screen.fullScaleX/Y into account.

The update to hardware acceleration messed up the offset values for images and sprites. When scaling the x/y values it is also scaling the originX/Y values as well.

In Atlas.PrepareTile:

[code]if (_layer.index > drawCallThreshold)
{
_layer.prepare();
getSpriteByLayer(layer).graphics.drawTiles(_tilesheet, _layer.data, smooth, _renderFlags);
}[/code]

Shouldn't that > be changed to a <? I was unable to get my atlas-based Images to render until I made that edit.

When playing too much sounds at once the game crashes on Android, not on other platforms.

You can recreate this error by too sounds at once or rapidly playing them.

When running in debug mode on Android, the console on the pc gives me this error :

E/AndroidRuntime(10114): FATAL EXCEPTION: GLThread 861
E/AndroidRuntime(10114): java.lang.NoSuchMethodError: no static method with name
='getSoundPosition' signature='(II)I' in class Lorg/haxe/nme/Sound;
E/AndroidRuntime(10114): at org.haxe.nme.NME.onRender(Native Method)
E/AndroidRuntime(10114): at org.haxe.nme.MainView$Renderer.onDrawFrame(Ma
inView.java:358)
E/AndroidRuntime(10114): at android.opengl.GLSurfaceView$GLThread.guarded
Run(GLSurfaceView.java:1516)
E/AndroidRuntime(10114): at android.opengl.GLSurfaceView$GLThread.run(GLS
urfaceView.java:1240)
E/DASH - ak897x( 403): ak897x_activate: unable to set accelerometer delay!

This happens because the if that checks if the source is a TextureAtlas should be an else if.

Fix available here: https://github.com/MaskedPixel/HaxePunk/commit/14bf5c46fc5244938c8a8f5781364cc57b0f2d5c How the hell do I attach this? I still hate git.

While testing the profiler in Flashdevelop I noticed that memory would steadily increase even while not doing anything. Looking through the objects it seems that, as far as I can tell, lots of functions are built up over time causing memory to go up (from 50MB up to about 100MB in my case).

Image with the profiler ( just noticed memory use is cropped out by accident) and what I belive is the function that is created all the time:

http://imgur.com/lNjG4

Bug tested on 2.1.1 and reported here http://forum.haxepunk.com/index.php?topic=312.0

The problem come from this line :
if (color < 0xFF000000) color = 0xFF000000 | color;

I'm not sure what you wanted to do here (and this method is quiet different from the FlashPunk one) but here is a solution that work. 0xFF000000 is equal to -16777216 in base 10. So it will almost never be true. Change the previous by this one :
color = 0xFF000000 | (0xFFFFFF & color);

The color will be displayed correctly, even if the user gives a value with an alpha channel as argument (it will be ignored).

If you don't like this behavior, and want to let the user set an alpha channel in the color, even though ' alpha ' (method's argument) is set to 1, then use this line instead to check if the color has an alpha channel
if ((color >> 24) & 0xFF == 0) color = 0xFF000000 | color;

On a side note, the second ' if ' of the method is useless :
if (color >= 0xFF000000) color = 0xFFFFFF & color;
Once again, 0xFF000000 == -16777216, so it will be true almost every time.

EDIT: completely changed my message after reading about colors and bitwise stuff.

Sorry to keep hurling these at you :)

Currently, I am running into either a bug with mouseX, or a simple misunderstanding of what mouseX is supposed to give me. Here is a build with the bug, and here is the source code. It's present on the Handle element below the Radio buttons on the left side. Also, the source code is currently using mouseFlashX, but the issue is still the same regardless.

When I select that bar to move, rather than use the pointer side of the my OS's mouse graphic, mouseX uses the opposite side. Besides the obvious issue of not going where the user wants it to, there is another oddity that deals with collidePoint and mouseX. When using collidePoint, the correct side of the mouse graphic is used, so this causes the range of the Handle slider to be shifted to the right equal to the width of the mouse graphic, meaning I can't reach the far left side, and the slider moves off the edge of the far right side.

_region.draw(_point.x, _point.y, layer,
HXP.screen.fullScaleX * scaleX * (_flipped ? -1 : 1),
HXP.screen.fullScaleY * scaleY,
angle);

I think each scaleX and scaleY is missing being multiplied by scale

Using a Grid mask, I'm getting worse performance than using individual entities. I'm using a 640x480 grid with 16x16 tiles. Unfortunately, I'm programming in spanish (it's first time this noobish mistake is a problem), so my code could be pretty hard to read if I copy-pasted it here.
Actually, this same thing happens with tilemaps. Here's some code explaining this better:

//Creating the world like this:
...
switch(level_data[i][a])
{
    case WALL:
    myTilemap.setTile(i,a,0);
    myGrid.setTile(i,a,true);
}
...
add(new Entity(0,0,myTilemap,myGrid)).type=Main.TYPE_WALL;

//...makes the update() loop slower than creating the world like this:

switch(level_data[i][a])
{
    case WALL:
    add(new Wall(i*16,a*16));// i*16/a*16 being the x/y position
}

The Grid and Tilemap classes need to be optimized or wiped out the lib, as their purpose is to make the game faster than using individual entities.
http://dl.dropbox.com/u/45951683/github%20grid%20issue.zip

Here are two .swf's of my WIP game. An swf uses Grid and Tilemap and the other one uses individual entities.

I'll be glad to help if you can't reproduce this problem. I might even take a look at the source files and try to optimize both classes later.

Currently, as far as I know, the only way to wrap a string is via manual placement of \n. FlashPunk seems to have a word wrapping feature; and it doesn't look too difficult to implement.

It seems also that you skipped over the options section. Haxe has anonymous objects, so there shouldn't be anything fancy required to get that up and running.

I might try fixing up Text myself, but I'd rather wait until my other pull requests get attention instead of adding a new random feature to it.

It appeared broken as below, when the text of a Text is modified.

C++ (Mac):

Flash:

Here is the sample code:

import com.haxepunk.*;
import com.haxepunk.graphics.*;

class TestWorld extends World {
    override public function begin():Void {
        super.begin();

        var text = new Text("Testing", { resizable: true });

        text.text = "TestingTestingTestingTesting";

        addGraphic(Image.createRect(text.textWidth, text.textHeight, 0xFF0000));
        addGraphic(text);
    }
}

class Main extends Engine {
    public function new():Void {
        super();
        HXP.world = new TestWorld();
    }

    static public function main():Void {
        nme.Lib.current.addChild(new Main());
    }
}

In FlashPunk, the moveBy function checks moveCollideX/Y for a boolean value before moving: here and here.
By default the 'moveCollide' functions return true, meaning the Entity should stop on collision.

First, I have confirmed that this issue was introduced in 2.2.0.

I have this code to create a scene and switch to it:
Main.ascene = new AScene();
HXP.scene = Main.ascene;

Later I switch to a new world:
HXP.scene = new BScene();

And switch back like this:
HXP.scene = Main.ascene;

Prior to 2.2.0, this worked perfectly and wold switch scenes exactly as expected. Now it seems to only half switch. AScene's update function is being called, and BScene is no longer the active scene, but any entities that were added to the scene before the switch don't appear to exist (not there visually and don't run update functions). New entities can be added, however.

The feature to scale a Tilemap would be useful in some cases.

Btw: Some labels for the github issues would be nice to at least differ between bugs and feature requests.

In tween.hx l.53 is the following code :
_time += HXP.fixed ? 1 : HXP.elapsed;

I think fixedFramerate is when you have a timer driven update rate, so time should be mesured in elapsed time, and non-fixed framerate is drive by "onFrame" events and time should be mesured as +1 frame. I think the condition is swapped and the expecting code is :
_time += HXP.fixed ? HXP.elapsed : 1;

Tutorial defaults works properly, but setting Engine's width/height to anything (even in constructor) throws following:

An ActionScript error has occurred:
ArgumentError: Error #2015: Invalid BitmapData.
    at flash.display::BitmapData/ctor()
    at flash.display::BitmapData()
    at com.haxepunk::HXP$/createBitmap()[/home/dsirijus/Libraries/haxepunk/com/haxepunk/HXP.hx:838]
    at com.haxepunk::Screen/resize()[/home/dsirijus/Libraries/haxepunk/com/haxepunk/Screen.hx:65]
    at com.haxepunk::HXP$/resize()[/home/dsirijus/Libraries/haxepunk/com/haxepunk/HXP.hx:228]
    at com.haxepunk::Screen/set_scaleX()[/home/dsirijus/Libraries/haxepunk/com/haxepunk/Screen.hx:200]
    at com.haxepunk::Engine/resize()[/home/dsirijus/Libraries/haxepunk/com/haxepunk/Engine.hx:223]
    at com.haxepunk::Engine/setStageProperties()[/home/dsirijus/Libraries/haxepunk/com/haxepunk/Engine.hx:186]
    at com.haxepunk::Engine/onStage()[/home/dsirijus/Libraries/haxepunk/com/haxepunk/Engine.hx:241]
    at com.haxepunk::Engine()[/home/dsirijus/Libraries/haxepunk/com/haxepunk/Engine.hx:96]
    at Main()[src/Main.hx:29]
    at Main$/main()[src/Main.hx:20]
    at Function/http://adobe.com/AS3/2006/builtin::apply()
    at Function/<anonymous>()
    at Reflect$/callMethod()[/usr/lib/haxe/std/flash/_std/Reflect.hx:56]
    at ApplicationMain$/begin()[bin/flash/haxe/ApplicationMain.hx:69]
    at ApplicationMain$/preloader_onComplete()[bin/flash/haxe/ApplicationMain.hx:96]
    at flash.events::EventDispatcher/dispatchEventFunction()
    at flash.events::EventDispatcher/dispatchEvent()
    at NMEPreloader/onLoaded()[bin/flash/haxe/NMEPreloader.hx:98]
    at ApplicationMain$/onEnter()[bin/flash/haxe/ApplicationMain.hx:88]

Doing graphic = new Text("hi"); currently breaks the game. The native Windows build stops responding immediately and the flash build just doesn't show anything.

When trying to output the haxepunk template to html5, it fails first with this error:
Called from ? line 1
Called from be/haxer/hxswfml/Main.hx line 242
Called from be/haxer/hxswfml/FontWriter.hx line 164
Called from format/ttf/Reader.hx line 27
Called from format/ttf/Reader.hx line 89
Called from C:\Motion-Twin\haxe/std/haxe/io/Input.hx line 237
Called from C:\Motion-Twin\haxe/std/haxe/io/BytesInput.hx line 61
Uncaught exception - Eof

and when you remove font assets from the nmml, it errors with this:
haxelib/HaxePunk/1,6,5/com/haxepunk/debug/Console.hx:361: characters 2-63 : Invalid arguments
haxelib/HaxePunk/1,6,5/com/haxepunk/debug/Console.hx:361: characters 2-63 : Function 'draw' requires arguments : source, ?matrix, ?colorTransform, ?blendMode, ?clipRect, ?smoothing

It's really important to me to have a working html5 exporter.

You can't really define a sub-region of an AtlasRegion because it is a limitation with the TileSheet class. It simply can't be done as of now.

The most obvious solution is to simply create a new AtlasRegion that is a subregion of the original. To do this you need:

• access is needed to the AtlasRegions parent.
• access to the AtlasRegions name.
• access to the AtlasRegions x and y values.

Maybe there is clipping somehow possible with drawtiles, but my googling has turned up nothing.

EDIT: It may even be advisable to have access to an AtlasRegion's sub-regions with the AtlasRegion itself. This could allow TileMaps to be passed an AtlasRegion instead of a bitmap. The benefit here is several parallax tilemaps could be drawn within a single draw call with a whole lot of other Graphics.

When using TexturePacker, if you create an atlas with rotated sprites, there are numerous issues with hardware rendering. The images created with cannot be rotated, scaled, or their origins changed reliably. Also, their y position is always off when rendered without the aforementioned properties modified.

I tried fix the issue by reworking a large portion of the rendering code to no avail. It could be done by someone with better maths than me so maybe someone can pull it off.

This is what needs to happen on the math side during rendering for rotation in the atlas to work:

rotate 90
translate height (this resets the origin to the top left)
translate originx/y
scalex/y
rotate angle

The problem shows up during translation. You cannot translate with the 2x2 matrix available through drawTiles. You need a 3x3 matrix for translations. The math has to be done by hand which would add a lot of unnecessary congestion to the rendering code. Although, performance-wise, on native it probably won't make THAT much of a difference.

Then again, I might be wrong. Tell me I am wrong and it can be done. Please...

The F(unktion) and num keys are not working on cpp windows build,
on flash everything is fine.

I get 80 as keycode for F1 on windows, which is equivalent to Key.P

if (Input.pressed(Key.F1))
    trace("F1");

if (Input.pressed(80)) // equals Key.P
    trace("F1 - Win");

if (Input.pressed(Key.NUMPAD_0))
    trace("NUM 0");

Creating a Text object on flash does always produce a ghost text at position 0, 0 with the same text.

var txt:Text = new Text("Test", 100, 100);
addGraphic(txt);

Without the addGraphic line the ghost is shown only.

I'm trying to set the origin of a spritemap in HaxePunk 2.1.1, but it only seems to work on flash.
Doesn't work on Windows, Windows Neko and Android.

For example, when I set the originX and originY of a Spritemap, the sprite stays at its original origin (0,0) instead of the origin I set.
It looks funny because it turns around the center thereby dragging the entire sprite around its axis. For more details on how it looks I included some screenshots :

If this is already fixed in the next version , disregard this issue.

They gives me an error: "On static platforms, null can't be used as basic type Float"
to fix it you need to replace the line:
if (a == null)
with:
if (Reflect.getProperty(_object, property) == null)
in both files in tween() methods (line 45 in VarTween and line 57 in MultivarTween)

it should be "#if (flash || html5)" at line number 30 in com.haxepunk.utils.Key class

sorry for my english.....

When trying to test my Clash demo on a Linux/Neko target, I get this error:

Couldn't set OpenGL mode: Couldn't find matching GLX visual
Called from sys.io.FileOutput::$statics line 1
Called from ApplicationMain::main line 39
Called from neash.Lib::create line 64
Called from neash.display.Stage::nmeProcessStageEvent line 358
Called from neash.display.Stage::nmeDoProcessStageEvent line 270
Called from neash.display.Stage::nmePollTimers line 651
Called from neash.display.Stage::nmeCheckRender line 228
Called from neash.display.Stage::nmeRender line 663
Called from neash.display.DisplayObjectContainer::nmeBroadcast line 129
Called from neash.display.DisplayObjectContainer::nmeBroadcast line 129
Called from neash.display.DisplayObjectContainer::nmeBroadcast line 142
Called from neash.display.DisplayObject::nmeBroadcast line 248
Called from neash.display.DisplayObject::nmeDispatchEvent line 259
Called from neash.events.EventDispatcher::dispatchEvent line 80
Called from neash.events.Listener::dispatchEvent line 177
Called from com.haxepunk.Engine::onEnterFrame line 232
Called from com.haxepunk.Engine::update line 115
Called from com.haxepunk.Engine::checkWorld line 318
Called from Test::begin line 29
Called from clash.Button::$init line 18
Called from clash.Button::new line 43
Uncaught exception - Invalid field access : __construct__

It compiles to both Linux/CPP and Flash, so there seems to be something wrong with the Neko target. The source code for the issue is in my Clash repository, with the issue seeming to come from the Button class.

I wish I had more to tell you, but I don't really know what the issue is.

I have code that looks like this:

highscoresText.x = (HXP.screen.width * .5) - (highscoresText.textWidth * .5);
highscoresText.y = (HXP.screen.height * .65);

This works fine in all versions of HaxePunk on the flash target, placing the text centered nicely. However, the text is in the wrong place on android using HaxePunk 2.1.0 and up (I haven't tested other devices, such as iOS or Blackberry). The text appears much lower, and to the right of where it should, nearly off the screen in fact.

When calling lineStyle in Draw.linePlus it causes a crash on Windows.

com.haxepunk.Draw:175
Graphic.lineStyle

Discovered that I hadn't tested the flipping properly. When two or more images tries to be flipped it will only over use the first image that gets flipped. This is because a hash is used to store the images and since all the bitmapdata objects have the name "BitmapData" the same iamge will be used. I have tried to come up with something to fix it but I can't come up with an idea that works neatly due to bitmapData not having a unique identifier.

The two fixes I have come up with is either:

1. When creating a new image (or a class that extends Image) you also pass an (optional?) identifier that is used as the _class variable.

2. Pass in Bitmaps instead of only the bitmapData and rely on HXP.getBitmap() ( I believe it would need some changes or a new function would be made).

(This is while using Samhaxe, may be other ways otherwise).

I am pretty new to Haxe and HaxePunk but this function worked back in version 1.7.1 but in this new version when I try running my game with HXP.console.enable() it outputs this http://pastebin.com/JbyK6FeQ.

Thanks in advance!

You'll have to inverse the "if" and "else if" conditions cause BitmapData is Dynamic too ;)

com.haxepunk.graphics.Backdrop:22

So I noticed a memory leak in my Clash UI project, and I don't really see where there could be a problem in my code to cause it. If I toggle between different skins, my memory goes up each time. When I talked about (something similar) to this on the forums, you said that all Graphics classes pull from a cache of bitmap data, so I wouldn't think there would be any change in memory after the first call to a new set of bitmap data.

Is the new memory just the overhead of the Image class, not necessarily the bitmap data? Here is the current Flash build of Clash, with the debug information visible to show the leak (if you are using FDB).

it seems the circle polygon collision is buggy. it expects the circle to have a centered offset.

http://forum.haxepunk.com/index.php?topic=239.msg428#new

Hello,

Great work you've done here! Look forward to building my next game with HaxePunk. :-)

I ran into a problem tonight with your Canvas class. It was throwing an "Error 2015: Invalid BitmapData." Here's the stack trace:

ArgumentError: Error #2015: Invalid BitmapData.
at flash.display::BitmapData/ctor()
at flash.display::BitmapData()
at com.haxepunk.graphics::Canvas()[src/com/haxepunk/graphics/Canvas.hx:57]
at com.haxepunk.graphics::Tilemap()[src/com/haxepunk/graphics/Tilemap.hx:44]
at com.haxepunk.tmx::TmxEntity()[src/com/haxepunk/tmx/TmxEntity.hx:57]

It seems when using Flash 10, the limitation on BitmapData is 4096x4096, and you have an arbitrary limit of 4000 set by default on line #32 of Canvas.hx, which is great:

_maxWidth = _maxHeight = 4000;

But on Flash 9, this breaks when you attempt to load an image greater than 2880x2880 because Flash 9 imposes a different size limitation. I was able to fix this for Flash 9 simply by changing this line to:

if flash9

    _maxWidth = _maxHeight = 2880;

else

    _maxWidth = _maxHeight = 4000;

end

Hope this helps.

Thanks,
Marc