weimingtom / nekonme Goto Github PK

There's no email address or contact form, on these pages or on your
homepage/blog ... I would have posted a comment on your blog, but there
doesn't even seem to be any way to set up an account for that? So this was
the only way I could sort out how to write something that would reach you.

I'm very interested in your SDL bindings for Neko. I'm considering using
HaXe, with your SDL bindings, for a project.

I'd love to tell you more, but as this information is publically visible,
perhaps this is not the place... (in fact, you should probably delete this
after reading it)

Please find my email address on http://www.mindplay.dk and get in touch?

Hope to hear from you!

Regards,
  Rasmus Schultz

What steps will reproduce the problem?
1. Here is the test code, if you have an iPod / iPhone / iPad, start debugging 
and pay attention to Debug Console to see the output (on the Simulator it 
almost bever occurs...):

import flash.Lib;
import flash.events.Event;

class IPhoneMain
{
   private static var _frameCount:Int = 0;

   function new()
   {
      flash.Lib.current.addEventListener(Event.ENTER_FRAME, OnEnterFrame);
   }

   public function OnEnterFrame(evt:Event):Void {
      _frameCount++;    
      trace(" --> OnEnterFrame :: " + _frameCount);
   }

   public static function main ()
   {
      var w = 768;
      var h = 1024;
      var fps = 60;
      var bgColor = 0x000000;
      var flags = 
           nme.Lib.FULLSCREEN |
           nme.Lib.BORDERLESS |
           nme.Lib.RESIZABLE  |
           nme.Lib.HARDWARE   |
           nme.Lib.VSYNC      |
           0;
      var name = "iPhone Application";
      nme.Lib.create(function(){new IPhoneMain();},w,h,fps,bgColor,flags, name);
   }
}

2. Look that sometimes the trace output stops writing on the Debug Console, 
i've done this little test without rendering because it does not matter, it 
stalls with or without rendering

3. Can it be the GC?

What is the expected output? What do you see instead?
Sometimes the application stalls and cause a little hang...

What version of the product are you using? On what operating system?
r539 on iPod 4.0 / iPad 3.2

Please provide any additional information below.
By the way, i currently coded Multiply and Screen blend modes on .hx (not cpp) 
layer cause i needed it to a project, if you think its useful i can share the 
code ;)

It would be great to have methods that can convert (or better cast) between 
haxe.BytesData, ByteArray, BitmapData.

Currently the input of BitmapData.loadFromBytes/BitmapData.loadFromHaxeBytes is 
the bytes of a encoded file, but I would like to load a raw pixel array into a 
BitmapData.

If the raw pixel array is referenced but not copied, it would be the best case.

before r535,my app can run 45 fps on iphone,but after i update to r535,it only 
8 fps.
i guess is there some thing going wrong ?

The flash.Error class has moved.

Compile the following code with:

-main Main
-cpp bin
-lib nme
--remap flash:nme
-resource foo.png@foo (attached)

class Main
{
  static function main():Void
  {
    nme.Lib.create(function()
    {
      var data = nme.display.BitmapData.loadFromHaxeBytes(haxe.Resource.getBytes('foo'));

      nme.Lib.current.addChild(new nme.display.Bitmap(data));

    }, 320, 240, 30, 0xffffff, (1*nme.Lib.HARDWARE) | nme.Lib.RESIZABLE);

  }
}

The problem seems to be image dimensions, which are 320x32. If I change them 
for example to 320x100 the program doesn't crash and the image is shown 
correctly.

111

Add right mouse button stuff.

Hi Hugh.

On Wed, 10 Feb 2010 22:27:20 +0800
"Hugh Sanderson" <[email protected]> wrote:
> I don't know too much at all about Linux sound.  I'm using SDL
> to do the "heavy lifting". Maybe I need to add sound options
> to the compile, or upgrade the mixer library.  Or maybe I simply
> have not initialized to correctly.  I'm working on the next version
> of NME at the moment, so I will have a look at the sound soon-ish.

I suspect that the missing sound support has to do with pulse.
I tried some options I found in some forums, but couldn't make it work.
Probably configure --enable-pulse might change something. There is also
a CheckPulseAudio() function in configure.in. Did you use some special
configure options for creating your sdl-static makefiles?
Also, would it be possible to dynamically link against an installed
libsdl?

> There is no "dirty rectangle" detection, so the render load is
> similar if nothing is moving.  If you set the frame rate low enough,
> then I think there should be reduced CPU usage, as it does a
> system wait between frames.

I see. I'll try some lower framerates.

--
Mockey

Attached is a patch to compile nme on linux for 32/64 bit x86 architectures.

What steps will reproduce the problem?
1.when i update to r557 the retina scale is not 1:1 pixels.
2.when i delete      evt.x = backingWidth;
                     evt.y = backingHeight;
it is ok.

Added a AVAudioPlayer implementation for Sound and SoundChannel, parallel to 
the OpenAL.

AVAudioPlayer is now used by default, unless inForceSound is set.

What steps will reproduce the problem?
1. Compile and run a 01-Sample the samples/ directory with
-cp ../..
-main Sample
-lib nme
-lib hxcpp
--remap flash:nme
--remap neko:cpp
-cpp cpp

2. Run the example with
#> cpp/Sample 

What is the expected output? What do you see instead?
Error : Could not load module nme@nme_create_display_object__0

What version of the product are you using? On what operating system?
nme/2,0,0 on Ubuntu 10.04 64bit desktop

Hi Hugh,

Thank you very much for your help!

I was just able to get this to work. It looks A LOT better.

The first steps to use sub-pixel rendering are to modify FreeType's headers to 
enable this feature (prior to May 2010 it violated patents), then to use 
FT_RENDER_MODE_LCD or FT_RENDER_MODE_LCD_V instead of FT_RENDER_MODE_NORMAL. 
This resulted in improved font rendering, but it still felt too bold and too 
aliased.

When you use one of these rendering modes, the values for each glyph are 
returned as three separate R, G, B values instead of a combined value. In order 
to get it to work initially, I needed to write code which would read all three 
values and merge them together. For some reason, I had been using the alpha 
from the G value to apply to the whole pixel. Averaging the alpha from all 
three components removed some of the blockiness.

It still looked like the font was not being smoothed as much as it should have 
been, so I played more with LCD filter modes and other settings to try and 
obtain a better result. Finally I tried FT_LOAD_FORCE_AUTOHINT, which ignores 
the custom hinter in each font file and uses FreeType's auto-hinting instead. 
This looks MUCH smoother.

It doesn't work because TTF has not (anymore?) a text() method


(for my needs I use 
class DTTF extends TTF {
    public function setText(text:String):Void {
        this.text = text;
    }
}
)

Hiding the mouse cursor is needed for most non-interactive presentations, but 
nme.ui.Mouse.hide() is not implemented.

Disregard the previous one.

Davide

What steps will reproduce the problem?
Load an image from haxe.io.Bytes into BitmapData.loadFromHaxeBytes

What is the expected output? What do you see instead?
It should load the image (works on mac) but it fails with the following error 
(iphone sim):
closerlib.iphone(8705,0xa04cd4e0) malloc: *** error for object 0x7075a08: 
pointer being freed was not allocated

p.s. Thank you for adding that and fixing the loadfrombytes for iphone

In the attached test, 3rd click on the text steadily crashes vm (64-bit 
Ubuntu). Sometimes kernel says:
*** glibc detected *** neko: malloc(): smallbin double linked list corrupted: 
0x0000000000ea7a10 ***
======= Backtrace: =========
...

3rd version of the patch. Fixes a couple of problems of the second version.

D

Could you include in next NME submit on haxelib the haxelib.xml XML output so 
that we can get online documentation on http://lib.haxe.org/p/nme ? Check 
http://haxe.org/haxelib for more infos.

Oh, and maybe the project infos need some update as well 
http://lib.haxe.org/p/nme

Check the attached example. Clicking the textfield won't scroll the field. 
Comment out the 
tt.selectable = false;
Now the scrolling works.

Currently (r586), nme.display.Loader can only load files from the local 
filesystem. This patch extends this functionality to loading images from a 
network location. I've tested this with JPEGs on the CPP target and it works as 
expected. This patch also provides a couple of minor fixes, such as correcting 
the typo reported in issue 45 and removing an unused import statement. For the 
CPP target, loading of network images occurs synchronously, which means the 
application freezes until the image is fully loaded. There are likely 
platform-specific ways around this. For example, for the iPhone, there is 
NSURLConnection.

with the name RGBA wouldnt it accept an RGB and A?

maybe convert to BGR->

inline static public function RGBA(inRGB:Int,inA:Int=0xff) : Int
{
    var BGR = (inRGB & 0x000000ff) << 16 | (inRGB & 0x0000FF00) | (inRGB & 0x00FF0000) >> 16;

    #if neko
    return BGR | ((inA & 0xfc)<<22);
    #else
    return BGR | (inA <<24);
    #end
}

What steps will reproduce the problem?
  1. Right- or middle-click and handle the fired MouseEvent.

What is the expected output? What do you see instead?
  When right-clicking, the rightIsDown property of the fired event should
be true, but leftIsDown is true instead. The same goes for middle clicking
and middleIsDown.

What version of the product are you using? On what operating system?
  NME 0.3.0, Windows XP service pack 2.

What steps will reproduce the problem?
1.
2.
3.

What is the expected output? What do you see instead?


What version of the product are you using? On what operating system?


Please provide any additional information below.

[yelbota@yelbota haxecar]$ ./cpp_out/Game -opengl
Couldn't set OpenGL mode: Couldn't find matching GLX visual

Looks like classic GLX visual problem with color depth.
http://code.google.com/p/nekonme/source/browse/trunk/project/nme.cpp#828

Recently I've been on a mission to find a GUI library that works on both
flash and cpp targets of haXe. In my struggles to get haxegui to work in
that fashion, I have accumulated a few changes to NME which I'm posting here
before it becomes an full-blown thermonuclear patch bomb.

Some of the patch is just code from neash that may have been overlooked.
Other stuff was just missing, so I attempted to add it based on the haXe
flash api. In a mad dash to make something work, I added quite a few empty
stubs and ad-hoc implementations (for instance, drawRoundRectComplex just
calls drawRoundRect) so it's far from perfect but might highlight some of
the more important missing parts. Hopefully someone finds part of it useful.

NME is a great library already, but if more people dedicate time to filling
in the blanks it could make c++/flash target compatibility really smooth.

Big thanks to all the NME and haXe contributors!
-Adam

Recently I am looking into using some Flash 3D engines on Neko/C++. I found
that all of them(at least fp10 branches) depends on the Vector3D and
Matrix3D class which are not in neash(or nme) yet. So, I've created the two
classes, and hope it can saves you some time 
Vector3D is complete but Matrix3D is not, but almost. I have tried a
few algorithms found on the web but none of them have the original result of
the method "pointAt"... And for "decompose" and "recompose", they only
support the default orientationStyle (eulerAngles).
The zip file also contains the unit test I created for comparing the result
between the classes in flash and neash. I do not have time to add comments
in the source yet... so if you have any problem, tell me...

It would be great if it was possible to generate sound in Neko and cpp targets. 
See the following example from as3 reference 
(http://help.adobe.com/en_US/FlashPlatform/reference/actionscript/3/flash/media/
Sound.html#event:sampleData)

var mySound:Sound = new Sound();
function sineWaveGenerator(event:SampleDataEvent):void {
    for ( var c:int=0; c<8192; c++ ) {
        event.data.writeFloat(Math.sin((Number(c+event.position)/Math.PI/2))*0.25);
        event.data.writeFloat(Math.sin((Number(c+event.position)/Math.PI/2))*0.25);
    }
}

mySound.addEventListener(SampleDataEvent.SAMPLE_DATA,sineWaveGenerator);
mySound.play();

thoughts?

What steps will reproduce the problem?
1. No flash.geom.PerspectionProjection yet :)

What is the expected output? What do you see instead?
awesomeness use with your drawTriangles implementation, instead i see compile 
error

NME 2

Merge webOS changes

Using sample 03-Bitmaps I have changed the constructor to simply:


   super();
   Lib.current.addChild(this);

   var shape = new Shape();
   addChild(shape);
   var gfx = shape.graphics;
   var mtx = new Matrix();
   gfx.beginBitmapFill(image2,mtx,true,true);
   gfx.drawRect(0,0,300,300);  


Upon running the image appears to be repeated correctly in the x-axis, however 
has lots of gaps in the y-axis. Furthermore, if the image is resized from the 
default (130x101) to 100x100 for example the x-axis repetition is broken as 
well.

In short, repetition in both axis appears to be buggy when used.

Running hxcpp 2.07, nme 2.0.1, haxe 2.07. Compiling with the command:

haxe cpp.hxml

There is a typo on line 94 of revision 586 of trunk/nme/net/URLLoader.hx:

bytesLoaded = bytesTotal= this.data.lenght;

The word "lenght" should probably be "length".

I was just perusing the code, so I don't know if this typo actually has any 
negative effect.

Cheers,
Josh

I am developing additional extensions to nme, like some robots communication 
interfaces, remote sensors & etc, is there a way to only use the nme headers 
(c++) and build something based on it in another ndll ?

What I do now is always fork nme and add the extensions directly to it, i am 
sorry, i am not fluent with c/c++ but it would so much better if there would be 
a nme.lib (nme.a) file together with the headers that would allow people to use 
the nme ByteArray C++ classes.

Also, can you also provide what versions of dependencies are you using for the 
svn version (i mean, what zlib version, libjpeg, libpng, sdl, sdl_mixer, etc.)

Thank you for this amazing project

The attached test file contains all of the test cases.

1) It seems that "name" property disappears when mouse click event is called on 
that object.

Immediately after object creation:
Test.hx:33: DisplayObjectContainer 4 : test!

On clicking:
Test.hx:39: DisplayObjectContainer 4 : null

2) Bitmaps are not clickable. Not sure if it's supposed to be this way or not.

3) setting backgroundColor property of textfields doesn't do anything. 

4) opaqueBackground changes textfield background but makes textfield 
unclickable.

changed name of inForceSound variable to inForceMusic, so by default OpenAL is 
used.

using openAL sound (i.e. sound that is not music) causes an app to be 
terminated when in the background. 

This happens because polling of the sound completions causes a render to 
happen, which causes a SIGABRT.

This workaround makes it so a check is made of whether the app is animating. If 
it isn't it means that the app is in the background so OnPoll shouldn't be 
called.

Hey guys,

I've been quietly working on my cross-platform game engine on Haxe.
Still no luck with ZIP, and TAR is only working on my cpp target, but
my code works on both CPP and SWF if I download all of my resources
from URLs individually. I've now got a Mac set up and successfully
used Huge's XCode template to get the sample running in the simulator.
I've also been able to get my code to compile and run.

However, I'm getting a run-time error when trying to open my resource
file. Even adding a "test.txt" to the original sample project and
trying to open that is failing. I've double checked that my file is
indeed added to the build target in the "Copy Bundle Resources" tab,
and even opened up the .app package and verified that the file is
placed at the root folder there. Simply calling
File.getContent("test.txt"); fails with the generic error message:
Error [file_contents, test.txt]

I'm old-hat at iPhone, so I'm wondering if I need to prepend the path
with something like: NSString *filePath = [[NSBundle mainBundle]
pathForResource:@"MyFile" ofType:@"txt"];  and if so-- how would I
expose this Obj-C snippet to Haxe (or hopefully that someones' written
an iPhone SDK utility library to expose things like this already).

Any feedback would be appreciated.

-Paul Zirkle

What steps will reproduce the problem?
1. using a sample of nme2 on iphone simulator

What is the expected output? What do you see instead?
It should run on simulator (it's running on device), but instead, it's creating 
a lot of link error.

What version of the product are you using? On what operating system?
NME 2.0, haxe 2.0.6, hxcpp 2.0.6 on macosx 10.6.4 with xcode 3.2.3.

// Tested on the current svn version

tt = new nme.text.TextField();
tt.multiline = true;
tt.wordWrap = true;

// this makes Line1 and Line2 on the same line then Line3 on the next line
tt.htmlText = "Line1<br>Line2<br>Line3";

// inserting \n makes the first br work as expected and somehow the second too
tt.htmlText = "Line1<br>\nLine2<br>Line3";

// \n is treated as br (probably a bug?)
tt.htmlText = "L1\n\nLine2\nLine3";

// line1 and text on the same line, then ok
tt.htmlText = "Line1<br><font color=#55dd55>Text</font>Line2<br>Line3";

Like previous one, only the debug is off

Ubuntu 10.04, 64-bit

Also how can I find out which mouse button was pressed from the event?

What steps will reproduce the problem?
1. Create a BitmapData with transparency
2. Do data.clear(0);
3. Draw it somewhere

What is the expected output? What do you see instead?
Expect to see nothing, since it should have alpha 0 (which is how it works in 
ActionScript/Flash). Instead I see a black bitmap.


What version of the product are you using? On what operating system?
Latest trunk as of right now, on windows.


Please provide any additional information below.
nme_bitmap_data_clear does "| 0xff000000" on the color, which kinda explains it.

Attempting to build on Ubuntu x64, I have to following error:

In file included from ByteArray.cpp:1:
ByteArray.h: In member function ‘void ByteArray::MakeSpace(int)’:
ByteArray.h:35: error: ‘memcpy’ was not declared in this scope
ByteArray.h:38: error: ‘memset’ was not declared in this scope
ByteArray.cpp: At global scope:
ByteArray.cpp:3: warning: non-local variable ‘<anonymous struct>*
k_byte_array’ uses anonymous type
ByteArray.cpp: In function ‘_value* nme_read_file(_value*)’:
ByteArray.cpp:68: warning: ignoring return value of ‘size_t fread(void*,
size_t, size_t, FILE*)’, declared with attribute warn_unused_result
make: *** [ndll_objs/ByteArray.o] Error 1

var o = new DisplayObject();
o.name = "test test";
trace(o.name.substr(0, 3));

passes the compiling fine but crashes miserably. I know why it happens but it's 
still a bit annoying.

I think the default pan value for a SoundTransform is 0, i.e. centered sound.

In the file project/sdl/SDLSound.cpp the function getLength() references a 
member of Mix_Chunk that doesn't exist:

    double getLength()
    {
       if (mChunk==0) return 0;
       return mChunk->length_ticks;
    }

mChunk->length_ticks doesn't exist.

 hi all!
A sound would stop in 3-4 second on iphone. if I turn off
GC,"cpp.vm.Gc(false)",every thing runs well.It seems similar to the bug with 
ByteArray.readFile in version 2.0.6.0.


It seem cause by some thing havn't been implement,i add this to sound.mm  ,it
works. 
  bool isComplete() {
      ALint state; 
      alGetSourcei(mSourceID, AL_SOURCE_STATE, &state); 
      return (state == AL_STOPPED);
   }

   void stop()
   {
      alSourceStop(mSourceID);
   }

Fixes tiny memory leaks when loading an image.