Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Yes it works you just need to change it in ALL the Android.mk because it's defined in each script file. That's why I said that I need to make another sumission adding optional optimization WHICH will overwrite every .mk.inwhen it's added to the main library. 2011/9/13 Marc Sticksen > Hi, > > only commenting the NEON-line in the Android.mk doesn´t work. Well, not > here in my system, maybe it works for you? Have you tried it? > > Cheers, > Marc > > -- > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=42699#42699 > > > > > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Hi, only commenting the NEON-line in the Android.mk doesn´t work. Well, not here in my system, maybe it works for you? Have you tried it? Cheers, Marc -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=42699#42699 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Yeah... bright mind, changin templates than can change form version to version it's better than change the scripts of your build. 2011/9/12 Marc Sticksen > > Jorge Izquierdo Ciges wrote: > > Neon extensions maybe don't make so much acceleration with devices but v7 > makes some improvements depending the device. If you change the Android.mk > in the example and in the OSG build and quit the Neon part of the script > (just comment the LOCAL_ARM_NEON:=true) and it should work for you. > > > For all that are searching a solution on how to disable NEON in the OSG > build: just comment the above section in file > OpenSceneGraph_root/PlatformSpecifics/Android/Android.mk.modules.in > > Cheers, > Marc > > -- > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=42604#42604 > > > > > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Jorge Izquierdo Ciges wrote: > Neon extensions maybe don't make so much acceleration with devices but v7 > makes some improvements depending the device. If you change the Android.mk in > the example and in the OSG build and quit the Neon part of the script (just > comment the LOCAL_ARM_NEON:=true) and it should work for you. For all that are searching a solution on how to disable NEON in the OSG build: just comment the above section in file OpenSceneGraph_root/PlatformSpecifics/Android/Android.mk.modules.in Cheers, Marc -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=42604#42604 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Neon extensions maybe don't make so much acceleration with devices but v7 makes some improvements depending the device. If you change the Android.mk in the example and in the OSG build and quit the Neon part of the script (just comment the LOCAL_ARM_NEON:=true) and it should work for you. 2011/8/23 Jorge Izquierdo Ciges > OH GOD > NOW I GET IT! > > How could I be so blind! damn! > > It's a Tegra Device! Tegra2 chipsets can compile armeabi-v7 BUT can't use > NEON extensions. If you see my examples and the OSG build files use NEON > extensions with armeabi-v7 by default for Google's advice. But Tegra2 has > the extrange weirdness to be an armeabi-v7 compatible processor BUT it > doesn't have any Neon extension module! > > Now all makes sense =_= > > Sigh... well I'll send a patch when I can adding options to the Cmake > building to add choice in versions and Neon. > > > 2011/8/23 Luca Vezzadini > >> Big news!!! >> I've been able to make the OSG example run on my Honeycomb device! >> Now, something is definitely screwed somewhere... here is what I did in >> order to make it run. >> - build for API platform 9 both OSG and the examples >> - build the example ONLY for armeabi, not armeabi-v7 >> - target Android 2.3.3 in the Eclipse build. >> >> By doing that the code runs fine. >> To me, this means that previously my Honeycomb device (Acer IconiaTab) was >> trying to load the armeabi-v7 version, as it should support it, but for some >> reason that build is not working. Forcing it to use the armeabi (by removing >> completely the build for v7) works fine. >> >> This is where I have to stop though... if you guys are in touch with the >> Android Dev community and know well enough the depths of OSG build for >> Android, you might want to see if and how this can be fixed... I am >> available to run more tests of course, so let me know if there is anything >> else I can do ok? >> In the meantime I'll finally go on with some real coding :) >> >>Luca >> >> -- >> Read this topic online here: >> http://forum.openscenegraph.org/viewtopic.php?p=42182#42182 >> >> >> >> >> >> ___ >> osg-users mailing list >> osg-users@lists.openscenegraph.org >> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org >> > > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
OH GOD NOW I GET IT! How could I be so blind! damn! It's a Tegra Device! Tegra2 chipsets can compile armeabi-v7 BUT can't use NEON extensions. If you see my examples and the OSG build files use NEON extensions with armeabi-v7 by default for Google's advice. But Tegra2 has the extrange weirdness to be an armeabi-v7 compatible processor BUT it doesn't have any Neon extension module! Now all makes sense =_= Sigh... well I'll send a patch when I can adding options to the Cmake building to add choice in versions and Neon. 2011/8/23 Luca Vezzadini > Big news!!! > I've been able to make the OSG example run on my Honeycomb device! > Now, something is definitely screwed somewhere... here is what I did in > order to make it run. > - build for API platform 9 both OSG and the examples > - build the example ONLY for armeabi, not armeabi-v7 > - target Android 2.3.3 in the Eclipse build. > > By doing that the code runs fine. > To me, this means that previously my Honeycomb device (Acer IconiaTab) was > trying to load the armeabi-v7 version, as it should support it, but for some > reason that build is not working. Forcing it to use the armeabi (by removing > completely the build for v7) works fine. > > This is where I have to stop though... if you guys are in touch with the > Android Dev community and know well enough the depths of OSG build for > Android, you might want to see if and how this can be fixed... I am > available to run more tests of course, so let me know if there is anything > else I can do ok? > In the meantime I'll finally go on with some real coding :) > >Luca > > -- > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=42182#42182 > > > > > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Big news!!! I've been able to make the OSG example run on my Honeycomb device! Now, something is definitely screwed somewhere... here is what I did in order to make it run. - build for API platform 9 both OSG and the examples - build the example ONLY for armeabi, not armeabi-v7 - target Android 2.3.3 in the Eclipse build. By doing that the code runs fine. To me, this means that previously my Honeycomb device (Acer IconiaTab) was trying to load the armeabi-v7 version, as it should support it, but for some reason that build is not working. Forcing it to use the armeabi (by removing completely the build for v7) works fine. This is where I have to stop though... if you guys are in touch with the Android Dev community and know well enough the depths of OSG build for Android, you might want to see if and how this can be fixed... I am available to run more tests of course, so let me know if there is anything else I can do ok? In the meantime I'll finally go on with some real coding :) Luca -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=42182#42182 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Well I have it for linux I can send it to you. If it really is a Bug they will solve in the next rc. 2011/8/22 Luca Vezzadini > Searching on many forums it comes out that other people do have similar > issues with that linker warning. Not sure if that is related to the > execution errors that we have (see original post from Ricky) but there might > be a relation. > Somewhere it was mentioned that the _dso_handle thing is defined in libc.a; > and if you look at the stack traces posted by Ricky the libc is always > mentioned on the very top of the list... > Could it be an issue with the new NDK release? Maybe something is messed up > in the way it handles these things? I'm not such an expert in gcc > compilation so I cannot help that much here, sorry about that... > > Luca > > PS > I'd like to try using the NDK r5b which apparently did not show these > warnings. Any idea where I could get it from? The official NDK page only > offers the link to the latest version > > -- > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=42166#42166 > > > > > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Searching on many forums it comes out that other people do have similar issues with that linker warning. Not sure if that is related to the execution errors that we have (see original post from Ricky) but there might be a relation. Somewhere it was mentioned that the _dso_handle thing is defined in libc.a; and if you look at the stack traces posted by Ricky the libc is always mentioned on the very top of the list... Could it be an issue with the new NDK release? Maybe something is messed up in the way it handles these things? I'm not such an expert in gcc compilation so I cannot help that much here, sorry about that... Luca PS I'd like to try using the NDK r5b which apparently did not show these warnings. Any idea where I could get it from? The official NDK page only offers the link to the latest version -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=42166#42166 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
I actually used 10 for both (OSG and the example). Luca -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=42162#42162 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
10 doesn't matter for Ndk because the last change was in 9 (if you look there's no platform 10). If you already did this don't botter. But just to be sure you did it with the OSG library not the example, right? 2011/8/22 Luca Vezzadini > Hi Jorge, > Actually I did already change that, so to compile for platform-10 even. But > no success either. > If you think that it might help to try explicitly with platform 9 I will. > But I seriously doubt that it can help... > > Luca > > -- > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=42159#42159 > > > > > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Hi Jorge, Actually I did already change that, so to compile for platform-10 even. But no success either. If you think that it might help to try explicitly with platform 9 I will. But I seriously doubt that it can help... Luca -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=42159#42159 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Luca, I'm writing some documentation for my work and I thought that maybe (It shouldn't) you should compile the OSG library for platform 9. Just change the variable "APP_PLATFORM" to android-9 in the file Application.mk of the OSG build dir. Also change accordingly the rest of the files that refer to platform 5. I doubt it can be so simple but i've seen weirder things in Android. 2011/8/19 Jorge Izquierdo Ciges > Ok I'll check if those warnings come in the r6 in my Linux. > > 2011/8/19 Luca Vezzadini > >> I'm on NDK r6, the latest one. >> >> Luca >> >> -- >> Read this topic online here: >> http://forum.openscenegraph.org/viewtopic.php?p=42129#42129 >> >> >> >> >> >> ___ >> osg-users mailing list >> osg-users@lists.openscenegraph.org >> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org >> > > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Ok I'll check if those warnings come in the r6 in my Linux. 2011/8/19 Luca Vezzadini > I'm on NDK r6, the latest one. > > Luca > > -- > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=42129#42129 > > > > > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
I'm on NDK r6, the latest one. Luca -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=42129#42129 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
What's your Ndk version? There are people that claim that those warnings weren't before. 2011/8/19 Luca Vezzadini > Hi, > Yes, I will run more tests on this next week, trying to figure it out. > In the meantime, I noticed one warning when linking the JNI part of the > GLES1 example (libosgNativeLib.so) for armeabi-v7a: > > Code: > ld.exe: warning: type and size of dynamic symbol `__dso_handle' are not > defined > > > > Searching a bit on this it looks like a potential problem. I have no idea > if this gives any clue to people, I just wanted to report it for now. Will > dig more into it... > Thanks! > > Luca > > -- > Read this topic online here: > http://forum.openscenegraph.org/viewtopic.php?p=42125#42125 > > > > > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Hi, Yes, I will run more tests on this next week, trying to figure it out. In the meantime, I noticed one warning when linking the JNI part of the GLES1 example (libosgNativeLib.so) for armeabi-v7a: Code: ld.exe: warning: type and size of dynamic symbol `__dso_handle' are not defined Searching a bit on this it looks like a potential problem. I have no idea if this gives any clue to people, I just wanted to report it for now. Will dig more into it... Thanks! Luca -- Read this topic online here: http://forum.openscenegraph.org/viewtopic.php?p=42125#42125 ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Ok. I was thinking that the error would be in the JNI bridge. I think there are some changes about how android treats native processes. There is a mode to debug JNI by the way which gives more info. 2011/8/18 Riccardo Corsi > > One more info is that I've tried to build all the example with both 2.3.3 > and 3.1 target profile, and they show the same problem on the tablet. > I've tried as well to modify manually the minimum and target version of the > sdk in the AndroidManifest.xml but nothing changed. > > Ricky > > P.S. I'll run some more tests today, but since tomorrow I'll be in vacation > for a couple of weeks =) My colleague Luca will run some tests and follow > this thread while I'm away. Thank you. ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Hi Jorge,
I've added some Log as you suggested, and apparently as soon as it construct
the osgMainApp native object, something goes wrong.
In particular I've added a log in the EGL Renderer like this:
public void onSurfaceChanged(GL10 gl, int width, int height)
{
Log.e(TAG, "BEFORE_NATIVE_INIT");
osgNativeLib.init(width, height);
}
and one log in the init() method of osgNativeLib.cpp like this:
JNIEXPORT void JNICALL Java_osg_AndroidExample_osgNativeLib_init(JNIEnv *
env, jobject obj, jint width, jint height)
{
__android_log_write(ANDROID_LOG_ERROR, "OSGANDROID",
"NATIVE_INIT_CODE");
mainApp.initOsgWindow(0,0,width,height);
}
Finally, I've also added 2 log as first and last instruction of the
OsgMainApp ctor (in the cpp file):
OsgMainApp::OsgMainApp()
{
__android_log_write(ANDROID_LOG_ERROR, "OSGANDROID",
"OSG_MAIN_APP_CTOR_BEGIN");
_lodScale = 1.0f;
_prevFrame = 0;
_initialized = false;
_clean_scene = false;
__android_log_write(ANDROID_LOG_ERROR, "OSGANDROID",
"OSG_MAIN_APP_CTOR_END");
}
What happens is that "BEFORE_NATIVE_INIT" is printed, the 2 logs in the cpp
ctor are printed as well,
but NOT he "NATIVE_INIT_CODE", which suggests some error related to the
native object right after it's created.
The new logcat is in attach.
One more info is that I've tried to build all the example with both 2.3.3
and 3.1 target profile, and they show the same problem on the tablet.
I've tried as well to modify manually the minimum and target version of the
sdk in the AndroidManifest.xml but nothing changed.
Ricky
P.S. I'll run some more tests today, but since tomorrow I'll be in vacation
for a couple of weeks =) My colleague Luca will run some tests and follow
this thread while I'm away. Thank you.
On Wed, Aug 17, 2011 at 18:44, Jorge Izquierdo Ciges wrote:
> Well it's easy... there are some debug messages that I put them because the
> obscureness and difficulties to find some error so usually the trace should
> have first the message:
>
>
> 08-16 15:59:29.331: WARN/EGLview(4904): creating OpenGL ES 1 context
>
> That message come from the EGL java class WHICH initializes the GLES context.
> That message is in your logs. Now when you have the context the method
> onSurfaceCreated is called and it's a do nothing in my example, for more
> complex programs the correct initialization should be done there and the
> surfaceChange have another initialization code, because that function is
> called when you change possition on your device and on other places. So they
> should not have the same code.
>
> When you enter onSurfaceChanged, that's when OSG initializes everything so
> NOW we have the BIG trouble. the init function goes through a JNI bridge
> (init function) and calls the function initOsgWindow of mainApp.
>
> Inside initOsgWindow there's another LOG that should write. It's the FIRST
> instruction and it is dispatchet through __android_log_write so that will
> ever show whathever happens.
>
> "Initializing geometry"
>
> If you don't see that Log entry then no OSG code has been used.
>
> Then Maybe it's the same strange bug that Rafa told me once (he could fix it
> for his samsung) but since 2.2 that bug dissapeared... And i don't know e
>
> Now you should test if the application can reach the JNI bridge, put some
> __android_log_write in the osgNativeLib in the init function after and before
> calling the osg function.
>
> And about the log that's just the full heap. There are some manuals if you
> want to know more to translate to code but just by looking what I've said to
> you... it's a strange bug.
>
>
>
> 2011/8/17 Riccardo Corsi
>
>> Hi Jorge,
>>
>> I've set some breakpoints and stepped into the java code with the
>> debugger.
>> From the GLThread these methods are called in this order:
>> - EGLView::onSurfaceCreated()
>> - EGLView::onSurfaceChanged()
>>
>> Inside onSurfaceChanged(), when it calls the native method
>> osgNativeLib.init(), it exits and thus never reaches onDrawFrame() is never
>> reached.
>>
>> Ricky
>>
>> P.S. By the way... how did you get that info from the logcat? it's still
>> quite obscure to me...
>>
>>
>>
>>
>>
>> On Wed, Aug 17, 2011 at 16:38, Jorge Izquierdo Ciges
>> wrote:
>>
>>> That's one hell of a error. ¡He doesn't reach the initOsgWindow function!
>>>
>>> I've never seen it, but I have some ideas. First it's the same error for
>>> both of them. This must be some Native compatibility isue. I've heard that
>>> the EglView in which the example is based has some troubles with some
>>> devices but no one has said something solid about it. So, try to check by
>>> debug or just Log (quicker) which methods of the EGLview.java if onDraw is
>>> calle before onSurfaceChange or if onSurfaceChanged is never called we'll
>>> hace to see if there has been some changes in the API that requiere a SDK
>>> target version on the manifest. If not... well we'll see.
>>>
>>> 2011/8/17 Riccar
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Well it's easy... there are some debug messages that I put them because the obscureness and difficulties to find some error so usually the trace should have first the message: 08-16 15:59:29.331: WARN/EGLview(4904): creating OpenGL ES 1 context That message come from the EGL java class WHICH initializes the GLES context. That message is in your logs. Now when you have the context the method onSurfaceCreated is called and it's a do nothing in my example, for more complex programs the correct initialization should be done there and the surfaceChange have another initialization code, because that function is called when you change possition on your device and on other places. So they should not have the same code. When you enter onSurfaceChanged, that's when OSG initializes everything so NOW we have the BIG trouble. the init function goes through a JNI bridge (init function) and calls the function initOsgWindow of mainApp. Inside initOsgWindow there's another LOG that should write. It's the FIRST instruction and it is dispatchet through __android_log_write so that will ever show whathever happens. "Initializing geometry" If you don't see that Log entry then no OSG code has been used. Then Maybe it's the same strange bug that Rafa told me once (he could fix it for his samsung) but since 2.2 that bug dissapeared... And i don't know e Now you should test if the application can reach the JNI bridge, put some __android_log_write in the osgNativeLib in the init function after and before calling the osg function. And about the log that's just the full heap. There are some manuals if you want to know more to translate to code but just by looking what I've said to you... it's a strange bug. 2011/8/17 Riccardo Corsi > Hi Jorge, > > I've set some breakpoints and stepped into the java code with the debugger. > From the GLThread these methods are called in this order: > - EGLView::onSurfaceCreated() > - EGLView::onSurfaceChanged() > > Inside onSurfaceChanged(), when it calls the native method > osgNativeLib.init(), it exits and thus never reaches onDrawFrame() is never > reached. > > Ricky > > P.S. By the way... how did you get that info from the logcat? it's still > quite obscure to me... > > > > > > On Wed, Aug 17, 2011 at 16:38, Jorge Izquierdo Ciges wrote: > >> That's one hell of a error. ¡He doesn't reach the initOsgWindow function! >> >> I've never seen it, but I have some ideas. First it's the same error for >> both of them. This must be some Native compatibility isue. I've heard that >> the EglView in which the example is based has some troubles with some >> devices but no one has said something solid about it. So, try to check by >> debug or just Log (quicker) which methods of the EGLview.java if onDraw is >> calle before onSurfaceChange or if onSurfaceChanged is never called we'll >> hace to see if there has been some changes in the API that requiere a SDK >> target version on the manifest. If not... well we'll see. >> >> 2011/8/17 Riccardo Corsi >> >>> Hi Jorge and all, >>> >>> I start a new thread as I don't want to pollute the other - >>> http://forum.openscenegraph.org/viewtopic.php?t=8917&start=17 >>> which mainly talks about building for Android. >>> >>> Basically the issue is that osgAndroidExampleGLES1 doesn't work when >>> deployed to my Acer Iconia A500, running Android 3.1 >>> (while the same build works just fine on the Nexus S running Android >>> 2.3.4) >>> >>> The application starts up for a couple of seconds, and when it's about to >>> create the osg blue screen it dies. >>> I attach the logcat, even though I couldn't understand from it what's the >>> reason that makes the viewer die... >>> >>> The same happens with the GLES2 example, for which I attach the logcat as >>> well. >>> >>> I'm still investigating and will post if I have some news, >>> any hint is appreciated. >>> Ricky >>> >>> ___ >>> osg-users mailing list >>> osg-users@lists.openscenegraph.org >>> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org >>> >>> >> >> ___ >> osg-users mailing list >> osg-users@lists.openscenegraph.org >> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org >> >> > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
Hi Jorge, I've set some breakpoints and stepped into the java code with the debugger. >From the GLThread these methods are called in this order: - EGLView::onSurfaceCreated() - EGLView::onSurfaceChanged() Inside onSurfaceChanged(), when it calls the native method osgNativeLib.init(), it exits and thus never reaches onDrawFrame() is never reached. Ricky P.S. By the way... how did you get that info from the logcat? it's still quite obscure to me... On Wed, Aug 17, 2011 at 16:38, Jorge Izquierdo Ciges wrote: > That's one hell of a error. ¡He doesn't reach the initOsgWindow function! > > I've never seen it, but I have some ideas. First it's the same error for > both of them. This must be some Native compatibility isue. I've heard that > the EglView in which the example is based has some troubles with some > devices but no one has said something solid about it. So, try to check by > debug or just Log (quicker) which methods of the EGLview.java if onDraw is > calle before onSurfaceChange or if onSurfaceChanged is never called we'll > hace to see if there has been some changes in the API that requiere a SDK > target version on the manifest. If not... well we'll see. > > 2011/8/17 Riccardo Corsi > >> Hi Jorge and all, >> >> I start a new thread as I don't want to pollute the other - >> http://forum.openscenegraph.org/viewtopic.php?t=8917&start=17 >> which mainly talks about building for Android. >> >> Basically the issue is that osgAndroidExampleGLES1 doesn't work when >> deployed to my Acer Iconia A500, running Android 3.1 >> (while the same build works just fine on the Nexus S running Android >> 2.3.4) >> >> The application starts up for a couple of seconds, and when it's about to >> create the osg blue screen it dies. >> I attach the logcat, even though I couldn't understand from it what's the >> reason that makes the viewer die... >> >> The same happens with the GLES2 example, for which I attach the logcat as >> well. >> >> I'm still investigating and will post if I have some news, >> any hint is appreciated. >> Ricky >> >> ___ >> osg-users mailing list >> osg-users@lists.openscenegraph.org >> http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org >> >> > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
Re: [osg-users] osgAndroidExampleGLES1 and 2 error on Android 3.x device
That's one hell of a error. ¡He doesn't reach the initOsgWindow function! I've never seen it, but I have some ideas. First it's the same error for both of them. This must be some Native compatibility isue. I've heard that the EglView in which the example is based has some troubles with some devices but no one has said something solid about it. So, try to check by debug or just Log (quicker) which methods of the EGLview.java if onDraw is calle before onSurfaceChange or if onSurfaceChanged is never called we'll hace to see if there has been some changes in the API that requiere a SDK target version on the manifest. If not... well we'll see. 2011/8/17 Riccardo Corsi > Hi Jorge and all, > > I start a new thread as I don't want to pollute the other - > http://forum.openscenegraph.org/viewtopic.php?t=8917&start=17 > which mainly talks about building for Android. > > Basically the issue is that osgAndroidExampleGLES1 doesn't work when > deployed to my Acer Iconia A500, running Android 3.1 > (while the same build works just fine on the Nexus S running Android 2.3.4) > > The application starts up for a couple of seconds, and when it's about to > create the osg blue screen it dies. > I attach the logcat, even though I couldn't understand from it what's the > reason that makes the viewer die... > > The same happens with the GLES2 example, for which I attach the logcat as > well. > > I'm still investigating and will post if I have some news, > any hint is appreciated. > Ricky > > ___ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org > > ___ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
