Opened 19 years ago

Closed 19 years ago

#23 closed defect (fixed)

bad mesh generated.

Reported by: duncan Owned by: duncan
Priority: normal Milestone: Mesh generation maintenance
Component: Functionality and features Version: 1.0
Severity: normal Keywords:
Cc: duncan

Description (last modified by duncan)

This is from ticket #9, which was overloaded.

here's an error from least-squares.py

v 6
self.mesh.vertexlist[v] None
Traceback (most recent call last):
  File "I:\inundation\debug\mesh_error_reporting\show_mesh_bug.py", li
ne 123, in ?
    verbose=True)
  File "I:\inundation\pyvolution\least_squares.py", line 211, in fit_t
o_mesh
    precrop = precrop)
  File "I:\inundation\pyvolution\least_squares.py", line 360, in __ini
t__
    precrop = precrop)
  File "I:\inundation\pyvolution\least_squares.py", line 545, in build
_interpolation_matrix_A
    element_found, sigma0, sigma1, sigma2, k = \
  File "I:\inundation\pyvolution\least_squares.py", line 608, in searc
h_triangles_of_vertices
    for k, _ in self.mesh.vertexlist[v]:
TypeError: iteration over non-sequence

It's caused since vertex 6 is not part of any triangle.

I:\inundation\debug\mesh_error_reporting\show_bad_mesh.py (changeset:1954) produces a different mesh that shows the same error. Vertex 10 is not associated with any triangle. This is evident in the .tsh file.

See show_mesh_bug.py (changeset:1878) for 1st report of bug.

Currently least squares assumes a mesh is ok. Should this be changed? probably.

I just had a look at the mesh.py, at the triang.genMesh call. The error seems to be coming from triangle.

Change History (8)

comment:1 Changed 19 years ago by duncan

  • Description modified (diff)
  • Summary changed from crash when generating a mesh to bad mesh generated.

comment:2 Changed 19 years ago by duncan

  • Description modified (diff)

comment:3 Changed 19 years ago by duncan

  • Description modified (diff)

comment:4 Changed 19 years ago by ole

  • Component changed from Architecture and API to Functionality and features

comment:5 Changed 19 years ago by ole

  • Owner changed from somebody to duncan

Duncan said: "Currently least squares assumes a mesh is ok. Should this be changed? probably."

Perhaps that test should be part of check_integrity() in mesh ?

comment:6 Changed 19 years ago by duncan

  • Status changed from new to assigned

comment:7 Changed 19 years ago by duncan

  • Milestone changed from Mesh generation finished to Mesh generation maintanence

comment:8 Changed 19 years ago by duncan

  • Resolution set to fixed
  • Status changed from assigned to closed

All user points are turned into triangulation vertices in triangle. added code in pmesh.mesh and triagle so vertices without a triangle are removed.

Note: See TracTickets for help on using tickets.