[C] 'spRGB2Timeline_setFrame' Xcode reports "Bad access"

[pbk20191]

This extends '#2863'

I do actually find "bad_access" to the memory

this line cause the bad access

	frames[frame + COLOR2_B2] = b2;

In my understanding

static const int RGB2_ENTRIES = 7, COLOR2_R2 = 5, COLOR2_G2 = 6, COLOR2_B2 = 7;

should be like

static const int RGB2_ENTRIES = 7, COLOR2_R2 = 4, COLOR2_G2 = 5, COLOR2_B2 = 6;

because

each entry has 7 slot
0-> time
1 -> light-red
2 -> light-green
3 -> light-blue
4-> dark-red
5-> dark-green
6 -> dark-blue

7-> bad access

If i change the entries to

static const int RGB2_ENTRIES = 7, COLOR2_R2 = 4, COLOR2_G2 = 5, COLOR2_B2 = 6;

spine files is correctly parsed and renders expected result

You can actaully regenerate the issue using

https://github.com/pbk20191/spine-runtimes/tree/spinecios

with "Spine iOS example" project with "Spine C Example Target"

This only happens when spine file contains rgb2 not rgba2 so you can not check using "tank" and "coin" spine example.

And only occurs when using spine binary file
You can find It when you enable "Guard Malloc" option in xcode debugger

In Debug Build current status has no problem, But in Release Build with full optimization, RGB2 renders unexpected result

[pbk20191]

If you need full working sample, plz let me know where should I email it

[lukeeingram]

@pbk20191 given that I've fixed the RGB2 timeline issue, are you still running into the bad access error in Xcode?

[pbk20191]

@pbk20191 given that I've fixed the RGB2 timeline issue, are you still running into the bad access error in Xcode?

In my experience, this issue is caused by wrong indices for "RGB2" which can happen if the spine-c runtime create spine model from binary file not json file.
You can reproduce this in any environment which has "Guard Malloc" (libgmalloc)

• libgmalloc interrupt the program if the program access the memory which was not allocated.

also see

https://github.com/EsotericSoftware/spine-runtimes/pull/2863/files

but I don't understand why the PR is rejected. Since the below libgdx version indicates spine-c version do has wrong indices for RGB2 timeline.

https://github.com/EsotericSoftware/spine-runtimes/blob/4.2/spine-libgdx/spine-libgdx/src/com/esotericsoftware/spine/Animation.java#L1358C1-L1393C4

[pbk20191]

Also spine-cpp version RGB2 timeline seems to be damaged for "4.2"

see
#2886
#2885

so referring spine-cpp version is not a good choice in my opinion.