Yeah, its random..doesnt do it every time either..and its independent of how much geometry is there (it will do it on an empty model).
Yeah, I wouldnt hold my breath...ive been reporting it for years, and theres been no movement.
I imagine its probably due to some rather ancient coding...or gremlins..lol
What are the details for the appcrash?
If it's a memory corruption heisenbug, you'll see lots of false positive 'makes it go away' triggers that don't really do anything of the sort long-term. All those are doing is changing the memory arrangement and moving the "pt of corruption" to another location (versus the current location where it triggers the crash). That in no way means it won't cause a crash any more, instead
users no longer know when & where it will crash or cause other problems. That's not an improvement -- net result is worse, not better.
Too many among LW customers* believe finding a workaround that temporarily appears to "stop" a problem from occurring is somehow "as good as" finding the root cause of the problem and from there finding a workaround. The two paths are not equivalent, the former is actually worse than the latter. The latter qualifies and quantifies the bug, allowing actual impact of any workaround to be evaluated. The former leaves the current bug unqualified/unquantified, and in preventing it from occurring in one spot, is likely just causing it to manifest later elsewhere, likely with greater harm (greater systemic entropy).
*: I'm not saying that happened here, just pointing out it IS a recurrent theme in these forums.