Bug#1017598: freecad: autopkgtest fails on serveral archs (femtest.app.test_solver_calculix.TestSolverCalculix)

[Tobias Frost]

Package: freecad
Version: 0.20+dfsg1-2
Severity: serious
Control: tags -1 upstream
Control: forwarded -1 https://github.com/FreeCAD/FreeCAD/issues/7188

Dear Maintainer,

Since 0.20+dfsg1-2, autopkgtests fails the test
  femtest.app.test_solver_calculix.TestSolverCalculix
on the architectures
  arm64, i386, ppc64el and s390x


with this error:

======================================================================
FAIL: test_constraint_transform_torque (femtest.app.test_solver_calculix.TestSolverCalculix)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/usr/share/freecad/Mod/Fem/femtest/app/test_solver_calculix.py", line 278, in test_constraint_transform_torque
    self.input_file_writing_test(get_namefromdef("test_"))
  File "/usr/share/freecad/Mod/Fem/femtest/app/test_solver_calculix.py", line 397, in input_file_writing_test
    self.assertFalse(
AssertionError: 'Comparing /usr/lib/freecad/Mod/Fem/femtest/data/calculix/constraint_transform_torque.inp to /tmp/FEM_unittests/solver_calculix_constraint_transform_torque_e7672d59b8d6/Mesh.inp failed!\n--- \n+++ \n@@ -11717 +11717 @@\n-2505,2,-9.5670268990152E-01\n+2505,2,-9.5670268990153E-01\n@@ -12539 +12539 @@\n-3327,2,-2.2462134450621E+00\n+3327,2,-2.2462134450620E+00\n' is not false : CalculiX write_inp_file for constraint_transform_torque test failed.
Comparing /usr/lib/freecad/Mod/Fem/femtest/data/calculix/constraint_transform_torque.inp to /tmp/FEM_unittests/solver_calculix_constraint_transform_torque_e7672d59b8d6/Mesh.inp failed!
--- 
+++ 
@@ -11717 +11717 @@
-2505,2,-9.5670268990152E-01
+2505,2,-9.5670268990153E-01
@@ -12539 +12539 @@
-3327,2,-2.2462134450621E+00
+3327,2,-2.2462134450620E+00

(snippet from https://ci.debian.net/data/autopkgtest/testing/ppc64el/f/freecad/24870950/log.gz)

The upstream bug tracker suggests that this is a problem with the test, as it seems
floats for the input are very similiar, only differing by the LSB.

I'll disable the test for the time being.

--
tobi