Open HendrikBrueckler opened 1 week ago
I see now that you are supposed to use permutations for interfacing with polyscope, and that the 1-to-1 mapping of corners to halfedges is probably just the least cumbersome way to implement corner-based properties for modifiable polygonal meshes, despite the memory overhead.
I guess this can be closed, unless someone sees actual benefit in making CornerData::size()
fit SurfaceMesh::nCorners()
tightly.
Hi :) I encountered a small issue trying to use geometry-central with polyscope.
Currently
SurfaceMesh::nCorners()
returnsnInteriorHalfedgesCount
whereas the capacity check forCornerData
usesnHalfedgesCapacity
. On a compressed mesh with boundary, the two don't match; biggerCornerData
buffers than necessary are allocated andCornerData::size()
is larger thanSurfaceMesh::nCorners()
.For compressed meshes with boundary this creates issues when trying to combine geometry-central with polyscope and trying to add a corner-based parameterization from geometry-central as a property to the polyscope-mesh (polyscope essentially expects a buffer of size
nInteriorHalfedgesCount
).I would expect that intended behaviour here is that exterior halfedges do not contribute to "corners" and
CornerData
capacity should actually be equal tonInteriorHalfedgesCount
for a compressed mesh?