Open starseeker opened 2 hours ago
I suspect that this is related to #970, as that is also with offset and has similar behavior as this one.
No idea if this info is significant (may just be a quirk of the local machine and/or specifics of the test) but just in case I'll add it here:
When I checked Status() after processing N triangle generated solids in the last loop of the test, at N<15 things passed, N==15 it sometimes passed and sometimes failed, and N>15 seemed to reliably fail.
When working on #998, a failure (see below) was observed in triangulation when moving the Status() checks of the test up out of the processing loops: https://github.com/elalish/manifold/pull/998/commits/1f47f80305267429e91843d7819c2f65d14a61a1
Subsequent testing indicated that the issue was only observed (at least, on Ubuntu Linux) when tbb was enabled.
Needs additional debugging setup to capture the precise problematic inputs - just setting up this issue here to capture the info buried in the #998 discussion.