Android TODO list & poll

As I’m away next week and will not be able to work on jme3 ill take the chance and list my TODOs, as far as i am aware of them.

You can help me in the decision what is up next when i come back, or yet better post what you think is needed.



Android related TODOs as of 02.07.2011:



add Framebuffer support in OGLESShaderRenderer



Post processors like Shadow, Water(?), Fog run, but they don’t work correct.



Native bullet (this is a big one)



Move app to external sd



Terrain Demos (should work with raw height maps, not tested), need more demos



Texture handling (design decision, all android texture images are in (Bitmap)Image.getEfficentData() - which is non-standard)



Audio support is basic. You can use streaming & non-streaming AudioNodes, only positional or non-positional is supported.

No filters or effects like reverb, aso. I dont think this will get any better on Android 2.2. I heard about a real 3D audio interface

in Android 2.3, but did not check yet.



Would you think it could be that at some point we have to support more than one android version?

The compile process could create jmonkeyengine3-android22.jar, jmonkeyengine3-android23.jar, …

Is this desirable?



So long, please post your thoughts !

2 Likes
Post processors like Shadow, Water(?), Fog run, but they don't work correct.
I'd say it's low priority. To hit the lowest 2.2 spec possible, wouldn't most developers rather invest their time in visual treats better optimized for mobile devices? I don't have the technical savvy to back it up, but I don't see a place for advanced shader graphics on mobile until dual-core is nearing mainstream.

Native bullet (this is a big one)
Similar arguments as above, not to mention native Bullet isn't even default in jME3 yet, so jME3-Android should be in no rush.

Move app to external sd
What would this involve? You mean the app would install on the SD card by default (or required)? Is our download size going to increase by a lot?

Terrain Demos (should work with raw height maps, not tested), need more demos
Yeh, terrain is shining lately. Getting one or two terraindemos on Android would be a blast. I imagine it also makes for a really flexible performance test, seeing as the scope of the terrain can be easily changed from lightweight to behemoth with simple parameters.

That said, I think you will learn a lot more about what needs doing once the new .apk is made available on the Android Market. You've probably done more work on jME3-Android over the last few months than what was done over the year leading up to now. I'm confident you'll get at the very least twice the amount of testers and contributors that you have currently, as soon as the new App is published and announced.

Enjoy your time away mate! wherever that might be..
1 Like

Getting native bullet working would be an incredible help to me.

1 Like

Thanks for answers.



I also think that getting nice demos (including terrain and cinematics) is currently more important than adding new features. I also hope

this will attract more testers which is really needed as there are so many different devices out there.



More testers - more devices - more stability.



So my next task is to create a new demo app.

1 Like

Perhaps getting nifty to work well would be great. Especially the whole resolution-incorrect-causing-missed-clicks situation.

Also, I think native bullet would be quite useful, as many people are actually using it for simple things like collision detection/handling so it is crucial for games especially.

@Momoko_Fan

Screen resolution fix and flipable Mouse X/Y in AndroidHarness is committed.

I would like to say that the features that I would appreciate the most would be native bullet and nifty. Running apps with JBullet on the phone has been noticably slower than native bullet, however, without collision callbacks native bullet isn’t really useful yet. I would like to use nifty to get some events to fire when buttons are pressed on the screen to handle motion of the camera (other than rotating). Today I am placing pictures in guiNode and reading the touch event position to determine if a picture was pressed to fire my events.



Just my 2 cents.

I tested terrain with android 2.3 on a Nexus 1, the lighting terrain shader fail, and if i use another material you can see the terrain patches flickering when moving the cam.

So there is work to do i guess on the terrain side for android.



Also i agree it’s not a priority.



I’ll look into post processors and framebuffers if you want, but i think my phone does not support npot textures…



I think native bullet on android should be a priority, jbullet seems to be really slow on it, but native bullet seems ok (i remember reading a post about it)

Hi,



Newbie comments, so dont shoot me down in flames :slight_smile:



I can’t find anything about debugging directly on the android device. I could be wrong!



I’d like the PLAY/DEBUG buttons in jmonkeyPlatform to not just install, but run the app, as it does in Eclipse.



I’d also appreciate a demo on android touch input, this would also capture mobile keys, SEARCH, BACK, MENU etc. If it could be made to be emulated on pc, so much the better (ie, b key sends BACK button event).



thx

I think the native bullet on android is a true winner and I think we’ve got it cracked!



http://hub.jmonkeyengine.org/groups/android/forum/topic/is-android-viable/?topic_page=3&num=15#post-145046