Closed saidctb closed 4 years ago
@saidctb Actually Codacy's complaint may have uncovered a conceptual issue.
You have added an import of VectorFunctionSpace
, but it is never used. Instead, in _split_test_function(expr)
you create a ScalarFunctionSpace
for each component of expr
. Is this because we are implicitly assuming that any VectorFunctionSpace
is a tensor product of ScalarFunctionSpaces
?
I think that such an assumption belongs to psydac
, which uses tensor-product splines, but not to sympde
, which does not know anything about the discretization.
Codacy complains about an unused import. I can take care of it.