Restriction lambda containing captured variables do not work

[stijnh]

It seems that the following restriction:

n = 1024 * 1024 * 64
restrictions = [
    lambda p: n % calculate_chunk(p["vecsize"], p["nthreads"]) == 0,
]

tune_kernel("vector_add", kernel_string, size, args, tune_params, answer=answer, restrictions=restrictions)

Gives the following error:

  File "constraint/problem.py", line 249, in constraint.problem.Problem.getSolutionsAsListDict
  File "constraint/problem.py", line 239, in constraint.problem.Problem.getSolutionsOrderedList
  File "constraint/problem.py", line 214, in constraint.problem.Problem.getSolutions
  File "constraint/solvers.py", line 404, in constraint.solvers.OptimizedBacktrackingSolver.getSolutions
  File "constraint/solvers.py", line 392, in constraint.solvers.OptimizedBacktrackingSolver.getSolutionsList
  File "constraint/constraints.py", line 188, in constraint.constraints.FunctionConstraint.__call__
  File "<string>", line 1, in r
NameError: name 'n' is not defined

Additionally, the following restriction:

restrictions = [
    lambda p: n % calculate_chunk(p) == 0
]

Results in the following error:

  File "constraint/problem.py", line 249, in constraint.problem.Problem.getSolutionsAsListDict
  File "constraint/problem.py", line 239, in constraint.problem.Problem.getSolutionsOrderedList
  File "constraint/problem.py", line 214, in constraint.problem.Problem.getSolutions
  File "constraint/solvers.py", line 404, in constraint.solvers.OptimizedBacktrackingSolver.getSolutions
  File "constraint/solvers.py", line 392, in constraint.solvers.OptimizedBacktrackingSolver.getSolutionsList
  File "constraint/constraints.py", line 188, in constraint.constraints.FunctionConstraint.__call__
TypeError: r() takes 0 positional arguments but 2 were given