Merge pull request #15534 from seberg/deprecate-abstract-scalar-types

DEP: Do not allow "abstract" dtype conversion/creation

2 files changed, 25 insertions, 2 deletions

diff --git a/doc/release/upcoming_changes/15534.deprecation.rst b/doc/release/upcoming_changes/15534.deprecation.rst
new file mode 100644
index 000000000..243e224ba
--- /dev/null
+++ b/doc/release/upcoming_changes/15534.deprecation.rst
@@ -0,0 +1,11 @@
+Converting certain types to dtypes is Deprecated
+------------------------------------------------
+The super classes of scalar types, such as ``np.integer``, ``np.generic``,
+or ``np.inexact`` will now give a deprecation warning when converted
+to a dtype (or used in a dtype keyword argument).
+The reason for this is that `np.integer` is converted to ``np.int_``,
+while it would be expected to represent *any* integer (e.g. also
+``int8``, ``int16``, etc.
+For example, ``dtype=np.floating`` is currently identical to
+``dtype=np.float64``, even though also ``np.float32`` is a subclass of
+``np.floating``.
diff --git a/doc/source/reference/arrays.dtypes.rst b/doc/source/reference/arrays.dtypes.rst
index e97ee3c3a..f8d40b13c 100644
--- a/doc/source/reference/arrays.dtypes.rst
+++ b/doc/source/reference/arrays.dtypes.rst
@@ -156,6 +156,14 @@ Array-scalar types
 
 Generic types
 
+    .. deprecated NumPy 1.19::
+
+        The use of generic types is deprecated. This is because it can be
+        unexpected in a context such as ``arr.astype(dtype=np.floating)``.
+        ``arr.astype(dtype=np.floating)`` which casts an array of ``float32``
+        to an array of ``float64``, even though ``float32`` is a subdtype of
+        ``np.floating``.
+
     The generic hierarchical type objects convert to corresponding
     type objects according to the associations:
 
@@ -179,8 +187,7 @@ Built-in Python types
     :class:`float`    :class:`float\_`
     :class:`complex`  :class:`cfloat`
     :class:`bytes`    :class:`bytes\_`
-    :class:`str`      :class:`bytes\_` (Python2) or :class:`unicode\_` (Python3)
-    :class:`unicode`  :class:`unicode\_`
+    :class:`str`      :class:`str\_`
     :class:`buffer`   :class:`void`
     (all others)      :class:`object_`
     ================  ===============
@@ -196,6 +203,11 @@ Built-in Python types
        >>> dt = np.dtype(int)     # Python-compatible integer
        >>> dt = np.dtype(object)  # Python object
 
+    .. note::
+
+        All other types map to ``object_`` for convenience. Code should expect
+        that such types may map to a specific (new) dtype in future the future.
+
 Types with ``.dtype``
 
     Any type object with a ``dtype`` attribute: The attribute will be