jorgensd
fails to build with with GCC-15
A debian bug at https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1096535 reports that dolfinx_mpc fails to build (fails tests) with gcc-15: Build log at http://qa-logs.debian.net/2025/02/16/amd64exp/dolfinx-mpc_0.9.0-2_unstable_gccexp.log.gz
There is a gcc-15 porting guide at http://gcc.gnu.org/gcc-15/porting_to.html
The report is for v0.9.0. I haven't tested independently, I don't know if it's fixed in 0.9.1.
sample error:
_________________________ test_cube_contact[True-C++] __________________________
generate_hex_boxes = (<dolfinx.mesh.Mesh object at 0x7f3c6595ae40>, <dolfinx.mesh.MeshTags object at 0x7f3c7748cb90>)
nonslip = True
get_assemblers = (<function assemble_matrix at 0x7f3c5d5e9d00>, <function assemble_vector at 0x7f3c5d5eb7e0>)
@pytest.mark.parametrize("get_assemblers", ["C++"], indirect=True)
@pytest.mark.parametrize("nonslip", [True, False])
def test_cube_contact(generate_hex_boxes, nonslip, get_assemblers): # noqa: F811
assemble_matrix, assemble_vector = get_assemblers
comm = MPI.COMM_WORLD
root = 0
# Generate mesh
mesh_data = generate_hex_boxes
mesh, mt = mesh_data
fdim = mesh.topology.dim - 1
# Create functionspaces
V = fem.functionspace(mesh, ("Lagrange", 1, (mesh.geometry.dim,)))
# Helper for orienting traction
# Bottom boundary is fixed in all directions
u_bc = fem.Function(V)
with u_bc.x.petsc_vec.localForm() as u_local:
u_local.set(0.0)
u_bc.x.petsc_vec.destroy()
bottom_dofs = fem.locate_dofs_topological(V, fdim, mt.find(5))
bc_bottom = fem.dirichletbc(u_bc, bottom_dofs)
g_vec = [0, 0, -4.25e-1]
if not nonslip:
# Helper for orienting traction
r_matrix = dolfinx_mpc.utils.rotation_matrix([1 / np.sqrt(2), 1 / np.sqrt(2), 0], -theta)
# Top boundary has a given deformation normal to the interface
g_vec = np.dot(r_matrix, [0, 0, -4.25e-1])
# Top boundary has a given deformation normal to the interface
def top_v(x):
values = np.empty((3, x.shape[1]))
values[0] = g_vec[0]
values[1] = g_vec[1]
values[2] = g_vec[2]
return values
u_top = fem.Function(V)
u_top.interpolate(top_v)
top_dofs = fem.locate_dofs_topological(V, fdim, mt.find(3))
bc_top = fem.dirichletbc(u_top, top_dofs)
bcs = [bc_bottom, bc_top]
# Elasticity parameters
E = 1.0e3
nu = 0
mu = fem.Constant(mesh, default_scalar_type(E / (2.0 * (1.0 + nu))))
lmbda = fem.Constant(mesh, default_scalar_type(E * nu / ((1.0 + nu) * (1.0 - 2.0 * nu))))
# Stress computation
def sigma(v):
return 2.0 * mu * ufl.sym(ufl.grad(v)) + lmbda * ufl.tr(ufl.sym(ufl.grad(v))) * ufl.Identity(len(v))
# Define variational problem
u = ufl.TrialFunction(V)
v = ufl.TestFunction(V)
a = ufl.inner(sigma(u), ufl.grad(v)) * ufl.dx
rhs = ufl.inner(fem.Constant(mesh, default_scalar_type((0, 0, 0))), v) * ufl.dx
bilinear_form = fem.form(a)
linear_form = fem.form(rhs)
# Create LU solver
solver = PETSc.KSP().create(comm)
solver.setType("preonly")
solver.setTolerances(rtol=1.0e-14)
solver.getPC().setType("lu")
# Create MPC contact condition and assemble matrices
mpc = dolfinx_mpc.MultiPointConstraint(V)
if nonslip:
with Timer("~Contact: Create non-elastic constraint"):
> mpc.create_contact_inelastic_condition(mt, 4, 9, eps2=500 * np.finfo(default_scalar_type).resolution)
test_cube_contact.py:242:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <dolfinx_mpc.multipointconstraint.MultiPointConstraint object at 0x7f3c775fece0>
meshtags = <dolfinx.mesh.MeshTags object at 0x7f3c7748cb90>, slave_marker = 4
master_marker = 9, eps2 = 5e-13
def create_contact_inelastic_condition(
self,
meshtags: _cpp.mesh.MeshTags_int32,
slave_marker: int,
master_marker: int,
eps2: float = 1e-20,
):
"""
Create a contact inelastic condition between two sets of facets marker with individual markers.
The interfaces should be within machine precision of eachother, but the vertices does not need to align.
The condition created is :math:`u_s = u_m` where `s` is the restriction to the
slave facets, `m` to the master facets.
Args:
meshtags: The meshtags of the set of facets to tie together
slave_marker: The marker of the slave facets
master_marker: The marker of the master facets
eps2: The tolerance for the squared distance between cells to be considered as a collision
"""
if isinstance(eps2, numpy.generic): # nanobind conversion of numpy dtypes to general Python types
eps2 = eps2.item() # type: ignore
> mpc_data = dolfinx_mpc.cpp.mpc.create_contact_inelastic_condition(
self.V._cpp_object, meshtags._cpp_object, slave_marker, master_marker, eps2
)
E TypeError: create_contact_inelastic_condition(): incompatible function arguments. The following argument types are supported:
E 1. create_contact_inelastic_condition(arg0: dolfinx::fem::FunctionSpace<float>, arg1: dolfinx::mesh::MeshTags<int>, arg2: int, arg3: int, arg4: float, /) -> dolfinx_mpc.cpp.mpc.mpc_data_float
E 2. create_contact_inelastic_condition(arg0: dolfinx::fem::FunctionSpace<float>, arg1: dolfinx::mesh::MeshTags<int>, arg2: int, arg3: int, arg4: float, /) -> dolfinx_mpc.cpp.mpc.mpc_data_complex_float
E 3. create_contact_inelastic_condition(arg0: dolfinx::fem::FunctionSpace<double>, arg1: dolfinx::mesh::MeshTags<int>, arg2: int, arg3: int, arg4: float, /) -> dolfinx_mpc.cpp.mpc.mpc_data_double
E 4. create_contact_inelastic_condition(arg0: dolfinx::fem::FunctionSpace<double>, arg1: dolfinx::mesh::MeshTags<int>, arg2: int, arg3: int, arg4: float, /) -> dolfinx_mpc.cpp.mpc.mpc_data_complex_double
E
E Invoked with types: dolfinx.cpp.fem.FunctionSpace_float64, dolfinx.cpp.mesh.MeshTags_int32, int, int, float
../../debian/python3-dolfinx-mpc/usr/lib/python3.13/dist-packages/dolfinx_mpc/multipointconstraint.py:461: TypeError
---------------------------- Captured stdout setup -----------------------------
Info : Clearing all models and views...
Info : Done clearing all models and views
at first glance, this looks like an issue with using different compilers for dolfinx and dolfimx_mpc, on the nanobind/python interface level. I can’t download the logs on my phone, but is dolfinx built with same compiler, same nanobind version?
Good question, that may well be the issue, not gcc-15 itself. I'll keep monitoring it.
The versions in the reported log are: gcc-15 15-20250213-1 nanobind 2.5.0-1 dolfinx 1:0.9.0-6+b2 python 3.13.2-1
dolfinx 1:0.9.0-6+b2 was built against nanobind 2.4.0-1 and python3 3.13.1-3.
So likely you're right, it's the mismatch between nanobind 2.4.0 and 2.5.0.
The debian package dependency mechanism expects incompatibility only with the major version, not the minor version. Perhaps that needs to be tightened up, if the problem really is nanobind 2.4 vs 2.5.
The debian package dependency mechanism expects incompatibility only with the major version, not the minor version. Perhaps that needs to be tightened up, if the problem really is nanobind 2.4 vs 2.5.
Nanobind is very strict with the versions: https://nanobind.readthedocs.io/en/latest/faq.html#how-can-i-avoid-conflicts-with-other-projects-using-nanobind
use the same nanobind ABI version (see the Changelog for details). use the same compiler (extensions built with GCC and Clang are isolated from each other). use ABI-compatible versions of the C++ library. use the stable ABI interface consistently (stable and unstable builds are isolated from each other). use debug/release mode consistently (debug and release builds are isolated from each other).
Allthough the ABI version is the same between 2.4 and 2.5 https://nanobind.readthedocs.io/en/latest/changelog.html#version-2-2-0-october-3-2024
I guess the only thing to do at this point is check if dolfinx_mpc generally builds successfully with gcc-15. Is it easy to configure docker to use gcc-15?