It seems as though the logic by which the runtime avoids unnecessarily recalculating the indices of a skeleton's mesh is slightly flawed.

We updated to the latest version of the runtime and began getting errors regarding invalid indices. After some digging we found lines 213 to 219 inclusive, where the indices' recalculation is skipped only based on whether the quantity of indices is sufficient. It seems like this is too simple a check to support the continuously changing order, content, and quantity of the vertices.

By commenting out these lines we've been able to get things working again, and we thought it'd be wise to bring this to your attention and leave a proper fix in your hands, especially as we could've just misunderstood the whole thing!

Thanks for your time. Other than the occasional transient hiccup, Spine has been instrumental in the smooth development of our game. It's hugely appreciated.

Thanks mate. I've tried your update and unfortunately I'm getting the original errors again. I've run out of time today before getting a workaround in place, so I'll get back to you when I do, assuming you don't beat me to it.

I've restored normal operation by substituting in the previous version of the "addSubmesh" function for now, including my comment-out. Crude, I know. And, I'm not sure what salient repro steps I could give, short of listing everything Spine-related about our project.

However, I have noticed the "Unity 4.2+" requirement on the runtime's home page, where we're wearing 4.0. Could that be the source of such a problem?