Vec out of Range about manifold

[pca006132]

do you have any example that can reproduce the error? we never received any bug report like this from other users, but probably because hull is not used much in openscad, which uses other hull implementations.

[pca006132]

because tbb will mess up with exception stack (they rethrow the error in the main calling thread), the simplest way to debug this kind of issue would be to set debugger breakpoints.

[cartesian-theatrics]

I threw up a branch with an example that loads two models that fail to subtract. Not sure if it's useful or not. Model looks like this, mask is similar shape but slightly smaller to make the result hollow when subtracted. I'll try to created a generated test case when I get the chance (this one is taken from a non-trivial model created with my Clojure bindings). I'm quite confident there's no crazy degeneracy in the mesh.

https://github.com/elalish/manifold/compare/master...SovereignShop:manifold:vec-out-of-range-error?expand=1

[pca006132]

Thanks, will have a look at it once I have time.

[cartesian-theatrics]

Uploaded another example here of a simple model that fails to import:

https://www.printables.com/model/837410-aeroponic-mister-fitting

[pca006132]

Program received signal SIGTRAP, Trace/breakpoint trap.
manifold::VecView<glm::vec<3, float, (glm::qualifier)0> const>::operator[] (i=7391, this=<optimized out>) at /home/pca006132/code/manifold/src/utilities/include/vec_view.h:59
59	     throw std::out_of_range("Vec out of range");
(gdb) bt
#0  manifold::VecView<glm::vec<3, float, (glm::qualifier)0> const>::operator[] (i=7391, this=<optimized out>) at /home/pca006132/code/manifold/src/utilities/include/vec_view.h:59
#1  (anonymous namespace)::FaceMortonBox::operator() (inout=..., this=<optimized out>) at /home/pca006132/code/manifold/src/manifold/src/sort.cpp:102
#2  thrust::detail::wrapped_function<(anonymous namespace)::FaceMortonBox, void>::operator()<thrust::detail::tuple_of_iterator_references<unsigned int&, manifold::Box&, int> > (this=<optimized out>, x=...)
    at _deps/thrust-src/thrust/detail/function.h:125
#3  thrust::system::detail::sequential::for_each_n<thrust::system::cpp::detail::par_t, thrust::zip_iterator<thrust::tuple<unsigned int*, manifold::Box*, thrust::counting_iterator<int, thrust::use_default, thrust::use_default, thrust::use_default>, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type> >, int, (anonymous namespace)::FaceMortonBox> (
    first=..., n=<optimized out>, f=<error reading variable: That operation is not available on integers of more than 8 bytes.>) at _deps/thrust-src/thrust/system/detail/sequential/for_each.h:82
#4  thrust::for_each_n<thrust::system::cpp::detail::par_t, thrust::zip_iterator<thrust::tuple<unsigned int*, manifold::Box*, thrust::counting_iterator<int, thrust::use_default, thrust::use_default, thrust::use_default>, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type> >, int, (anonymous namespace)::FaceMortonBox> (exec=..., first=..., 
    n=<optimized out>, f=...) at _deps/thrust-src/thrust/detail/for_each.inl:67
#5  0x00007ffff7e8a2b7 in manifold::for_each_n<thrust::zip_iterator<thrust::tuple<unsigned int*, manifold::Box*, thrust::counting_iterator<int, thrust::use_default, thrust::use_default, thrust::use_default>, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type, thrust::null_type> >, int, (anonymous namespace)::FaceMortonBox> (args=..., args=..., args=..., 
    policy=<optimized out>) at /home/pca006132/code/manifold/src/utilities/include/par.h:170
#6  manifold::Manifold::Impl::GetFaceBoxMorton (this=this@entry=0x7fffffff4df0, faceBox=..., faceMorton=...) at /home/pca006132/code/manifold/src/manifold/src/sort.cpp:386
#7  0x00007ffff7e880ca in manifold::Manifold::Impl::Finish (this=0x7fffffff4df0) at /home/pca006132/code/manifold/src/manifold/src/sort.cpp:256
#8  0x00007ffff7e2ec61 in manifold::Boolean3::Result (this=0x7fffffff4f80, op=manifold::OpType::Subtract) at /home/pca006132/code/manifold/src/manifold/src/boolean_result.cpp:813
#9  0x00007ffff7e421ff in manifold::CsgOpNode::ToLeafNode (this=0x3a7ee1c0fb0) at /home/pca006132/code/manifold/src/manifold/src/csg_tree.cpp:435
#10 0x00007ffff7e662d1 in manifold::Manifold::GetCsgLeafNode (this=0x7fffffff5170) at /home/pca006132/code/manifold/src/manifold/src/manifold.cpp:110
#11 0x00007ffff7e669ca in manifold::Manifold::GetMesh (this=0x116e) at /home/pca006132/code/manifold/src/manifold/src/manifold.cpp:159
#12 0x00005555555f28d1 in Manifold_HullFail_Test::TestBody (this=<optimized out>) at /home/pca006132/code/manifold/test/manifold_test.cpp:350

It seems that the commit causing issue is 32eefe9. Still need further debugging though.

[pca006132]

@elalish I think some of the loops have different behavior, for example for DedupeEdge

halfedge_[halfedge_[current].pairedHalfedge].startVert = newVert;

is different from

      current = NextHalfedge(current);
      halfedge_[current].startVert = newVert;

[elalish]

Good call - okay I'll revert #778 and reopen it so we can do a more serious review. Does anyone have an idea for a TEST we can add to catch this kind of thing?