philmoe / cerberus Goto Github PK

From the Monkey-X forum by user Pakz:

When using the drawtext command in the onrender on imagecanvasses it does not work in html5 but it does work in glfw.

Import mojo2

Global canvas:Canvas
Global image:Image
Global imagecanvas:Canvas
Global image2:Image
Global imagecanvas2:Canvas
Global image3:Image
Global imagecanvas3:Canvas

Class MyApp Extends App
	Field refresh:Int
	Method OnCreate()
		canvas=New Canvas
		image=New Image(128,64,0,0)
		image2=New Image(128,64,.5,.5)
		image3=New Image(238,64,1,1)
		imagecanvas=New Canvas(image)		
		imagecanvas2=New Canvas(image2)		
		imagecanvas3=New Canvas(image3)						
		createimages()
		createimages()		
		SetUpdateRate(60)
	End Method
	Method OnUpdate()
	End Method
	Method OnRender()
		canvas.Clear 0,0,0
		canvas.DrawImage image,DeviceWidth()/2,50
		canvas.DrawImage image2,DeviceWidth()/2,50+96
		canvas.DrawImage image3,DeviceWidth()/2,50+(96*2)
		canvas.Flush
	End
End Class

Function createimages()
	imagecanvas.Clear
	imagecanvas.SetColor(0.1,0.1,0.1,1)
	imagecanvas.DrawRect(0,0,image.Width(),image.Height())
	imagecanvas.SetColor(1,1,1)
	imagecanvas.DrawText("Image...1",0,0)	
	imagecanvas.Flush
	imagecanvas2.Clear
	imagecanvas2.SetColor(1,1,1,1)
	imagecanvas2.DrawRect(0,0,image.Width(),image.Height())
	imagecanvas2.SetColor(0,0,0,1)
	imagecanvas2.DrawText("Image...2",0,0)	
	imagecanvas2.Flush
	imagecanvas3.Clear
	imagecanvas3.SetColor(1,1,1,1)
	imagecanvas3.DrawRect(0,0,image.Width(),image.Height())
	imagecanvas3.SetColor(0,0,0,1)
	imagecanvas3.DrawText("Image...3",0,0)	
	imagecanvas3.Flush
End Function

Function Main()
	New MyApp
End

Here the original link in the beginners forum : http://www.monkey-x.com/Community/posts.php?topic=10553

From the Monkey-X forum by user Maverick69:

When I'm using AppODeal 3rd Party SDK to show video ads, monkey crashes in the following lines because the MonkeyAppDelegate isn't found:

int gxtkGraphics::BeginRender(){
	MonkeyAppDelegate *appDelegate=(MonkeyAppDelegate*)[[UIApplication sharedApplication] delegate];
	width=appDelegate->view->backingWidth;

MonkeyAppDelegate *appDelegate=(MonkeyAppDelegate*)[[UIApplication sharedApplication] delegate];
	[appDelegate->view presentRenderbuffer];

I've replaced the code through the following lines which fixed the crash:

int gxtkGraphics::BeginRender(){

    BBMonkeyAppDelegate *appDelegate = BBIosGame::IosGame()->GetUIAppDelegate();
	width=appDelegate->view->backingWidth;

void gxtkGraphics::EndRender(){
	if( !CFG_OPENGL_GLES20_ENABLED ) Flush();
    BBMonkeyAppDelegate *appDelegate=BBIosGame::IosGame()->GetUIAppDelegate();
	[appDelegate->view presentRenderbuffer];
}

For the Monkey-X forum by user Maverick:

The Android Browser in my Samsung Galaxy S3 (Android 4.4.2) does have a
problem that I haven't encountered in other Browsers:

Creating a buffer view with an buffer that has a size that isn't a multiple of the base type creates an Exception:

ArrayBuffer length minus the byteOffset is not a multiple of the element size

This happened to me in a game where I'm loading a Binary File. Of course it is possible that the size of this Binary File isn't a multiple of 4.

Now when Monkey tries to load in the File into a DataBuffer object the above-mentioned Exception is thrown.

I've created a workaround in this pull request (blitz-research/monkey#93) because I weren't able to come up with a cleaner solution:

If the length of the buffer isn't a multiple of 4, I copy/resize the buffer to be a multiple of 4 in length.

Note: I keep the original length property so that the seeking logic remains the same.

Unless the developer goes in and modifies Cerberus-X's source code for Android it always includes 4 permissions. In my opinion they are important ones too. If the developer needs to add permissions to the game they are making, they can use ${ANDROID_MANIFEST_MAIN}

https://github.com/cerberusxdev/cerberus/blob/develop/targets/android/template/gradletemplate/app/src/main/AndroidManifest.xml#L7

android.permission.INTERNET
android.permission.ACCESS_NETWORK_STATE
android.permission.READ_EXTERNAL_STORAGE
android.permission.WRITE_EXTERNAL_STORAGE

Read / write external storage should never be used by games ever. If they have a save file backup system (and they always really should...), it'd be better to use the Storage Access Framework (Android 4.4+) or the system clipboard. Apps that require the internet permission but don't use it really chaps my hide too.

What do you think? May I do a PR for this?

I have a question/documentation issue and I am not seeing it anywhere so I'd like to know. When you run a binary of Cerberus X on any Linux host, the available target options are usually Android, HTML5 and the Linux target, both debug and release. However, according to this topic here https://www.cerberus-x.com/community/index.php?threads/crosscompiling.1493/

It seems like crosscompiling from one platform to another is possible. As the title of my post implies, I am seeing if compiling from Linux to Windows through MinGW could be possible. As per the official site at (https://mingw.osdn.io/) it is possible to build and run MinGW on a Linux host and generate binaries from there. As Cerberus X already has MinGW support, could this be extended to apply to Linux as well? So that a workflow of: HTML5, mobile, Linux and Windows could be there in one go? Of course as long as the user has provided their own MinGW compiler.

Hey! I would like to ask something. See, we are trying to get Cerberus-X on the FSF directory-listing, and we are looking at adding this project to the directory. After some discussion, we found this module here that looked concerning
https://github.com/PhilMoe/cerberus/blob/develop/modules/opengl/gl.h

The license seems to be explicitly forbidding the sharing and usage of this header and we found this quite strange as gl2.h allows redistribution of it so if possible, we would like some clarification on what's happening with this header and if any info could be provided.

Cheers!
MetalMaxMX

For the Monkey-X forum by user Xaron:

Hey Mark, in your mojo2/graphics.monkey module I found this:

Method Init:Void()
		_dirty=-1
		For Local i:=0 Until 4
			_lights[i]=New LightData
		Next
	End

Should the for loop not initialize till MAX_LIGHTS and not just to 4, so something like:

For Local i:=0 Until MAX_LIGHTS

like everywhere else?

Because a few lines below I see a potential access violation then when you have more than 4 lights:

Local light:=_lights[i]

Hi Phil

Are there any build instructions fro Linux around? I see the website/forum is no longer active and the instructions here on github are a bit too vague for my level of knowledge!

Cheers

Paul

What the title says.

Ted doesn't compile with XCode 11.1 anymore. QT 5 Creator complains about a missing string.h header file.

Gamepads are not detected if they are connected after the game has already started. This happened on Windows, Linux, and Mac.

This issue on GLFW looks like it may need to be updated to version 3.3 to fix the problem: glfw/glfw#1218

On Git the hub development branch. An alteration to the WinNT builder script's default parameters to menu option to show on start up breaks the option to do automated and deploy.

Why are the native libraries included in the Android target used instead of Java? They seem to only contain two small methods for parsing numbers from strings. It would be nice to see it be pure Java on Android, so it'll be compatible with any possible processor architecture that comes out in the future.

https://github.com/MikeHart66/cerberus/blob/develop/targets/android/template/langutil/jni/langutil.c

The documentation needs to be enhanced regarding the building process and settings for Android Apps. For an example:

• How to change the icons
• How to add permissions or libraries

When specifiying cellWidth and cellHeight and padding=true, the offset in the y is not taken into account

	Function LoadFrames:Image[]( path:String,cellWidth:Int,cellHeight:Int,padded:Bool=False,xhandle:Float=.5,yhandle:Float=.5,flags:Int=Image.Filter|Image.Mipmap,shader:Shader=Null )

fix: (added y and height just like x and width)

		Local x:=0,width:=cellWidth
		Local y:=0,height:=cellHeight
		If padded x+=1;width-=2;y+=1;height-=2
		
		Local tx:Int = material.Width/cellWidth
		Local ty:Int = material.Height/cellHeight
		Local frames := New Image[tx*ty]
		Local i:Int = 0
		For Local yi:Int = 0 To ty-1
			For Local xi:Int = 0 To tx-1
				frames[i]=New Image( material,xi*cellWidth+x,yi*cellHeight+y,width,height,xhandle,yhandle )

SetClipboard("à") copies "<?>" into the clipboard.

For the Monkey-X forum by user Peeling:

After upgrading (?) our test devices to iOS10, all our apps have started exhibiting the same problem:

In a ~1cm strip down the left hand edge (portrait), static touch input is only registered after pressing the screen continuously for ~1 second. Swipes are registered instantly.

I've tried logging out the result from TouchDown directly. If I don't touch the screen, it's zero. If I touch the screen outside that 1cm strip, it instantly changes to 1. If I swipe the screen inside that strip, it instantly changes to 1. But if I statically touch the screen inside that strip, it takes a continuous press of ~1 second before the output changes to 1.

This is a pretty big deal as we have a few buttons that live in that left hand strip and now give the impression of being unresponsive.

By July 2021 new submissions to the Google Play Store will allow only Android App Bundle, no APKs anymore.

https://developer.android.com/guide/app-bundle

Hi,

I managed to compile the current Version on Opensuse Tumbleweed. There where a few errors regarding the c++ version (7.3.1) that I"ve fixed locally.

Attached patch.

Cheers
Franz

linux_patch.txt

Just comment out the one line that imports the app compat library, and the base size of a game will go from 6.25MB to 668KB (based on the bouncy aliens example when installed on my phone).

There should probably be a way to include it again using the config.cxs file.

https://github.com/cerberusxdev/cerberus/blob/develop/targets/android/template/gradletemplate/app/build.gradle#L45

Dead code really gets on my nerves. ha ha

What do you think of this? I'd like to do a PR for this.

When setting #HTML5_CONSOLE_SHOW to False it leaves a blank spot where the console would have been. Also the console is shown for a split second when the game first starts.

GrantOjanen@67bd842#diff-0ca17d8b1de08089e88e6fed11442b345e8dedce835e535e7020906f0c438350

Hi folks,

just want to leave a note here that any mojo2 example doesn't work anymore after transcc got updated lately. No matter which platform I tried.

Cheers
Michael

The Desktop target for Windows talks about using VS Express 2013 with Update 4... at time of writing Update 5 is the latest, but if you manage to find and download VS Express 2013, builds will fail.

The project\project.buildv2020-05-09\glfw3\msvc\CerberusGame.vcxproj file generated will refer to the 141 build dependencies of MSBuild which is... VS2017? As these don't get installed with VS2013, there's a bit of headscratching involved.

Not sure which is the preferred route here, whether to document how to fix the project file so VS2013 works (easy enough, in the vcxproj file, just find <PlatformToolset> in both the debug and release builds and replace with v120 instead of v141) or, probably, the better route to update the docs to suggest whichever version of VS Express/Community/whatever is the build the CX devs themselves are using.

If you import mojo2, #COLOR_USE_NORMALISED_COMPONENTS is inititalised which allows floats in Color. If you import mojo2.graphics, #COLOR_USE_NORMALISED_COMPONENTS isnt initialised and floats arent allowed. It might not be a bug and import mojo2 might be the documented way, rather than import mojo2.graphics.
For my project I didnt want to include mojo.audio, but I can see most game projects do want to include all mojo components

Import mojo2.graphics
col = New Color(Rnd(),Rnd(),Rnd())
' always 0
Print col.r

When writing (alpha shl 24 + $ffffff) to a databuffer and then writing that to an img, when drawn it doesnt display the alpha and just draws solid.
This happens in GLFW and HTML targets

The attached demo,will set the image to either RED GREEN BLUE or WHITE and draw it to the screen. (press R G B or W for the relevant image). This image is drawn on the left

In data there is also an image it loads that is set to WHITE with variable alpha. This draws correctly. See the right of the screen

This seems to only happen when using $FFFFFF white
test_writepixel.zip

Being able to get the display name and button/axis count of a controller would be really nice to have. It should be pretty easy to put in.

GLFW: glfwGetJoystickName(), glfwGetJoystickButtons(), glfwGetJoystickAxes()
HTML5: gamepad.id, gamepad.buttons.length, gamepad.axes.length
Android: InputDevice.getName(), It looks like Cerberus uses hard coded button/axis counts on Android.

I'd like to add a link for Linux on "The Desktop target.cerberusdoc" to the installation and build requirements guide. https://www.cerberus-x.com/community/index.php?threads/install-and-build-requirements-for-linux-includes-raspbery-pi-updated.741/

Also move up the "Notes" section, so that it is with the rest of the Windows information.

As a beginner these changes would have saved me a bit of time.

File:
https://github.com/cerberusxdev/cerberus/blob/develop/docs/cerberusdoc/Target%20SDKs/The%20Desktop%20target.cerberusdoc

Proposed Changes:
GrantOjanen@776cf43#diff-aa190177179f157f4f3d36e69263ff3db9031c668c37f4f74c9075cbd0a12b48

I’d like to contribute to cerberusX.
I manage to compile 64bit liblangutil NDK library for 64bit android.
It should be in our android target ..targets\android\template\gradletemplate\app\src\main\jnilibs

From the Monkey-X forum by user malublu:

Reference: http://www.monkey-x.com/Community/posts.php?topic=10564

Some names are not useable: u_color, ucolor, U_color

Currently font.load loads in font image sheets with default filter. This means that scaling a font is always filtered. It be nice to pass in the image filter
Example - Turn off filtering while loading font
fnt = Font.Load(file,0)

in mojo2.graphics.font
line 1380 Function Load:Font(url:String,flags:Int=Image.Filter)
line1453 _pages[pageNum] = Image.Load(filename,0,0,flags)

For the Monkey-X forum by user Maverick:

On FireFox keyevents with umlauts are fireing with keycode '0'. Those events shouldn't get eaten by Monkey or we won't receive those events within monkey (KeyDown works - GetChar doesn't work).

Here is my pull request:

blitz-research/monkey#91

When compiling this showed an error:
g++: error: transcc/transcc.build/cpptool/main.cpp: No such file or directory
I compared this version of Cerberus with an anterior version, this directory transcc.build exists in an anterior version but not exists in the new version.

For the Monkey-X forum by user JaviCervera:

When compiling an application for the GLFW3 target on macOS, I get the following error:

"fatal error: bracket nesting level exceeded maximum of 256"

This seems to happen because I have a VERY long Select statement that is expanded into a set of indented if else blocks in the generated C++ code.

I tried adding this to the top of my program:

#CC_OPTS+="-fbracket-depth=1024"

But it didn't work because it seems that CC_OPTS is only taken into account in the C++ Tool target (which is a shame, it would be great to be able to specify additional compile options on any C++ targets).

After compiling:
Ide is version: V2018-03-02
but version of compiler transcc is still V2017-10-24.

I think there was change because of this line:
In the file VERSIONS.TXT, in the line 15:
NEW: [TRANS] Added character literals to the language. (Author: Pixelpaladin)

For the Monkey-X forum by user Pit:

Dear,

some sound function don't work anymore on IOS.
here are the message I recieved on Xcode:

'AudioSessionInitialize' is deprecated: first deprecated in iOS 7.0
'AudioSessionSetProperty' is deprecated: first deprecated in iOS 7.0
'AudioSessionSetActive' is deprecated: first deprecated in iOS 7.0

For this code:

gxtkAudio::gxtkAudio():musicState(0),musicPlayer(0),musicVolume(1){

// BMH - set the session category to ambient (works with other music apps)
AudioSessionInitialize( NULL,NULL,NULL,NULL );
UInt32 sessionCategory=kAudioSessionCategory_AmbientSound;
AudioSessionSetProperty( kAudioSessionProperty_AudioCategory,sizeof(sessionCategory),&sessionCategory );
UInt32 allowMixWithOthers=true;
AudioSessionSetProperty( kAudioSessionProperty_OverrideCategoryMixWithOthers,sizeof(allowMixWithOthers),&allowMixWithOthers );
AudioSessionSetActive( true );

What can I do in order to play music ?

Pit

one precision: now it works (I think I used an old version of my code :-/ ), but I still have the messages as Alert.
So I suppose this must be taken in account for future release of Monkey...

Pit

From the Monkey-X forum by user Beaker:

The native js print function clashes with the built in window.print() function. I can't see an easy user fix for this. I had to do a manual search and replace: "[tab]print(" to "[tab]bbprint(".

forked repos can't be searched on github which sucks

From the Monkey-X forum by user SOAP:

BUILD FAILED
_mainfile.buildv87a\android\build.xml:83: Cannot find \Android\sdk\tools\ant\build.xml imported from _mainfile.buildv87a\android\build.xml

Total time: 0 seconds
TRANS FAILED: Android build failed.
Abnormal program termination.
Exit code: -1

It looks like in latest they completely removed the ant stuff, expects you to only build gradle projects.

We really need an official gradle project based Android target.

https://developer.android.com/studio/releases/sdk-tools.html

Last good tools for Monkey Android builds using ant

https://dl.google.com/android/repository/tools_r25.2.5-windows.zip

https://dl.google.com/android/repository/tools_r25.2.5-macosx.zip

https://dl.google.com/android/repository/tools_r25.2.5-linux.zip

With OSX Catalina, applications have to be 64bit. So the launcher has to be rebuild as a 64bit app. This can be done via BMax-NG or maybe it can be converted to a CX app.

https://www.cerberus-x.com/community/threads/app-crashes-when-clicking-on-window-minimize-32-bit.484/

User Phil7 wrote:

Hi,
my app crashes with a memory access error (error shown only with build and run)
when I click on the minimize Icon of the window or when I try to resize the window.

You can reproduce it with examples\mojo\difference\clock (didn't try any other examples)

I built once and changed #GLFW_GCC_MSIZE_WINNT="32" in config. cxs
Click on the edge of the window to resize (if enabled) or on the minimize icon.

Btw it works fine in debug mode
I am on Windows 10
Using cerberus v2018-04-21 and v2018-03-02

For the Monkey-X forum by user Soap:

Unloading fraps in mojo 2 apps causes memory access violations. Mojo 1 doesn't have this problem. Tested with mojo 2 bouncing aliens demo.

Tested with fraps 3.599 (latest)

The free version can be downloaded for testing http://www.fraps.com/download.php

For the Monkey-X forum by user Gerry Quinn:

#CPP_DOUBLE_PRECISION_FLOATS=True
Print

"Cos " + Cos( 90.0 )
Print "Cosr " + Cosr( 3.141592653589793 / 2.0 )

Output:
Cos 1.2167964475066282e-08
Cosr 6.123233995736766e-17

I'm guessing there must be an explicit float conversion done internally in translation that evades the C++ compiler.

EDIT: The constants in monkey.math could become an issue with double precision too, if people don't know they are SP

When using double quoted strings, which span over at least 2 lines, inside multi line comments, TRANS reports the wrong line number in error messages during the parsing.

See: https://www.cerberus-x.com/community/threads/cerberusx-reporting-incorrectly-number-of-line.960/#post-7610

For the Monkey-X forum by user Paul - Taiphoz:

Using the example that comes with it and building for GLFW3 it will not get anything.

It works fine in html5 tho, but I need it on Windows.

For the Monkey-X forum by user Maverick:

On Samsung Devices (on my Galaxy S4Mini) playing a sound effect sometimes results in a crash, throwing the following Exception:

InvalidStateError: DOM Exception 11

I've found out that it could happen that chan.waSource.stop( 0 ) throws this exception. However, as far as I understand the source, this happens if stop is called twice or when waSource Object isn't playing. So this if-branch shouldn't be reached, but it does.

I've fixed it by catching the exception and ignoring it, which is fine in this case.

I'm sure there must be a more cleaner solution, but I wasn't able to find one. This fix worked fine for me.

My bugfix pull request is here: blitz-research/monkey#92

From the Monkey-X forum by user Nobuyuki:

Hello,

I don't know when this happened, but on my Windows 10 machine, the glfw target doesn't seem to want to cooperate and let me do arbitrary stencil operations. Example code can be downloaded here:

https://github.com/nobuyukinyuu/nDrawExts2

Check the screenshot on the Github page for what the arbitrary stencil SHOULD look like. When compiling to Android (API_10), the code works just like in the screenshot, and in previous versions of Monkey, this also worked (although needed a bit of a workaround to get it to work on MacOS; re-initializing the window with 8 bit depth allocated to the stencil buffer, I'm assuming because it defaulted to a nonsupported value). In the new version of Monkey, I had to use SetDeviceWindow, but I think the code was working even then, although I could be wrong, because it was updated "in a hurry" just to get it to compile again.

Here's what it looks like now; the behavior's clearly broken, and the stencil mask is clearly passing everything to be drawn.

I tried it without SetDeviceWindow(), and the result was the same. Am I missing some voodoo here, or did the change over to this as a documented feature disable stencil bit processing entirely?

Figured out the issue. My current workaround involves editing the GLFW3 target. Please consider a patch!
In \targets\glfw3\modules\native\glfwgame.cpp , line 579, change the following line:

glfwWindowHint( GLFW_STENCIL_BITS,0 );

to this:

glfwWindowHint( GLFW_STENCIL_BITS,8 );

A new flag for SetDeviceWindow could also do, if you don't want the stencil channel to be enabled by default....

I've created a pull request for this issue, with a conservative approach just in case there's a reason to have the default stencilbits hint set at 0. Please consider...
blitz-research/monkey#89

As OpenGL ES2 is the minimum requirement for TEDs help doc viewer, it sometimes crashes if the hardware doesn't support it.

https://www.cerberus-x.com/community/threads/cerberus-x-crashes-because-of-issue-with-opengl.838/