CRASH: Arraylist issue in mTextureSlots.get(0); about rajawali HOT 5 CLOSED

Thanks, I'll look into it

from rajawali.

Thanks Dennis!
I had a sgs2 crash report too, but noticing there was a fix for that, I updated.
I've now run into a problem where the fix is causing an exception upon reload because it's getting into the conditional in geometry3d because there's a pending ogl error (1280). I can hack around this by changing the conditional and testing if there's an error and its not 1280, but that's pretty kludgey :D

According to the internet, 1280 is a texture state error... one person noted its caused by calling the enable function on texture types in opengl es20, which isn't necessary anymore (was only relevant to fixed function apparently?), another said its caused by texture loading with the input stream instead of decode function... I've no idea at the moment where it's coming from but there's a few leads for now :)

----- Reply message -----
From: "Dennis Ippel" [email protected]
Date: Mon, May 14, 2012 08:06
Subject: [Rajawali] CRASH: Arraylist issue in mTextureSlots.get(0); (#24)
To: "jistyles" [email protected]

Thanks, I'll look into it

Reply to this email directly or view it on GitHub:
#24 (comment)

from rajawali.

Cheers Joel,
Yes, ignoring that error for now. I'm looking at a proper fix.
The last commit has some changes as well. I've removed all default
materials because they were mostly instantiated but never used.
You might double check whether you're using default materials or not :-)

On 16 May 2012 09:40, jistyles <
[email protected]

wrote:

Thanks Dennis!
I had a sgs2 crash report too, but noticing there was a fix for that, I
updated.
I've now run into a problem where the fix is causing an exception upon
reload because it's getting into the conditional in geometry3d because
there's a pending ogl error (1280). I can hack around this by changing the
conditional and testing if there's an error and its not 1280, but that's
pretty kludgey :D

According to the internet, 1280 is a texture state error... one person
noted its caused by calling the enable function on texture types in opengl
es20, which isn't necessary anymore (was only relevant to fixed function
apparently?), another said its caused by texture loading with the input
stream instead of decode function... I've no idea at the moment where it's
coming from but there's a few leads for now :)

----- Reply message -----
From: "Dennis Ippel" [email protected]
Date: Mon, May 14, 2012 08:06
Subject: [Rajawali] CRASH: Arraylist issue in mTextureSlots.get(0); (#24)
To: "jistyles" [email protected]

Thanks, I'll look into it

---

Reply to this email directly or view it on GitHub:
#24 (comment)

---

Reply to this email directly or view it on GitHub:
#24 (comment)

from rajawali.

back on topic...
this texture crash looks like it's caused by:

GLES20.glGetIntegerv(GLES20.GL_MAX_TEXTURE_IMAGE_UNITS, numTexUnits, 0);

...Not returning a valid response, OR enough TIU for how many the app's requested. My current request amount is 11 (tested by hard coding the resulting mMaxTextures to 10 which crashes, and 11 which is fine).

Now I'm used to PC development and I'm also not overly familiar with the Rajawali texture manager so I may be talking be off the mark, but I thought the GL_MAX_TEXTURE_IMAGE_UNITS was the maximum bound samplers per pixel fragment, not the total in the scene which seems to be what the texture manager is tracking here.
Running with this assumption I hard coded the mMaxTextures value to 128, and then tested by increasing my total scene textures to >16 (which is the GL_MAX_TEXTURE_IMAGE_UNITS for my htc desire) which seems to work fine.

hope this helps!

from rajawali.

ps. with that test, I expect it will fail if I bind more than 16 texture samplers to one material, but I hadn't got far enough to test that out :)

from rajawali.