Nullable traits and default value (WIP)

numtraits.py

@@ -38,16 +38,23 @@
 class NumericalTrait(TraitType):
     info_text = 'a numerical trait, either a scalar or a vector'
     def __init__(self, ndim=None, shape=None, domain=None,
-                 default=None, convertible_to=None):
-        super(NumericalTrait, self).__init__()
+                 default=None, convertible_to=None, nullable=False):
+        # We use the default value for the traitlet in the following cases:
+        # - default is not None,
+        # - default is None and the trait is nullable.
+        # In the other cases, we assume that a None default means that
+        # there is no default value defined.
+        if (default is None and nullable) or default is not None:
+            super(NumericalTrait, self).__init__(default_value=default)
+        else:
+            super(NumericalTrait, self).__init__()
 
         # Just store all the construction arguments.
         self.ndim = ndim
         self.shape = shape
         self.domain = domain
-        # TODO: traitlets supports a `default` argument in __init__(), we should
-        # probably link them together once we start using this.
         self.default = default
+        self.nullable = nullable
         self.target_unit = convertible_to
 
         if self.target_unit is not None:
@@ -65,6 +72,10 @@ def _check_args(self):
                     raise TraitError("shape={0} and ndim={1} are inconsistent".format(self.shape, self.ndim))
 
     def validate(self, obj, value):
+        # If the trait is nullable and the value is None,
+        # then we have nothing to check.
+        if self.nullable and value is None:
+            return value
 
         # We proceed by checking whether Numpy tells us the value is a
         # scalar. If Numpy isscalar returns False, it could still be scalar

test_numtraits.py

@@ -13,6 +13,9 @@ class ScalarProperties(HasTraits):
     d = NumericalTrait(ndim=0, domain='negative')
     e = NumericalTrait(ndim=0, domain='strictly-negative')
     f = NumericalTrait(ndim=0, domain=(3, 4))
+    g = NumericalTrait(ndim=0, nullable=True)
+    h = NumericalTrait(ndim=0, nullable=True, default=1.23)
+    i = NumericalTrait(ndim=0, default=4.56)
 
 class TestScalar(object):
 
@@ -79,6 +82,20 @@ def test_range(self):
             self.sp.f = 7
         assert exc.value.args[0] == "f should be in the range [3:4]"
 
+    def test_nullable_default(self):
+        assert self.sp.g is None
+        assert self.sp.h == 1.23
+        assert self.sp.i == 4.56
+        self.sp.g = 1.2
+        assert self.sp.g == 1.2
+        self.sp.h = None
+        assert self.sp.h is None
+        self.sp.i = 1.23
+        assert self.sp.i == 1.23
+        with pytest.raises(TraitError) as exc:
+            self.sp.i = None
+        assert exc.value.args[0] == "i should be a scalar value"
+
 
 class ArrayProperties(HasTraits):