fix

Author: crusaderky

.github/workflows/tests.yml

@@ -40,8 +40,12 @@ jobs:
             python -m pip install 'numpy==1.26.*'
           else
             # Don't `pip install .[dev]` as it would pull in the whole torch cuda stack
-            python -m pip install array-api-strict dask[array] jax[cpu] ndonnx numpy sparse
+            python -m pip install array-api-strict dask[array] jax[cpu] numpy sparse
             python -m pip install torch --index-url https://download.pytorch.org/whl/cpu
+            if [ "${{ matrix.python-version }}" != "3.13" ]; then
+              # onnx wheels are not available on Python 3.13 at the moment of writing
+              python -m pip install ndonnx
+            fi
           fi
 
       - name: Dump pip environment

docs/supported-array-libraries.md

@@ -36,8 +36,6 @@ deviations from the standard should be noted:
   50](https://numpy.org/neps/nep-0050-scalar-promotion.html) and
   https://github.com/numpy/numpy/issues/22341)
 
-- `asarray()` does not support `copy=False`.
-
 - Functions which are not wrapped may not have the same type annotations
   as the spec.
 
@@ -47,12 +45,9 @@ The minimum supported NumPy version is 1.22. However, this older version of
 NumPy has a few issues:
 
 - `unique_*` will not compare nans as unequal.
-- `finfo()` has no `smallest_normal`.
 - No `from_dlpack` or `__dlpack__`.
 - `argmax()` and `argmin()` do not have `keepdims`.
 - `qr()` doesn't support matrix stacks.
-- `asarray()` doesn't support `copy=True` (as noted above, `copy=False` is not
-  supported even in the latest NumPy).
 - Type promotion behavior will be value based for 0-D arrays (and there is no
   `NPY_PROMOTION_STATE=weak` to disable this).
 
@@ -72,8 +67,8 @@ version.
   attribute in the spec. Use the {func}`~.size()` helper function as a
   portable workaround.
 
-- PyTorch does not have unsigned integer types other than `uint8`, and no
-  attempt is made to implement them here.
+- PyTorch does has incomplete support for unsigned integer types other
+  than `uint8`, and no attempt is made to implement it here.
 
 - PyTorch has type promotion semantics that differ from the array API
   specification for 0-D tensor objects. The array functions in this wrapper
@@ -100,8 +95,6 @@ version.
 - As with NumPy, type annotations and positional-only arguments may not
   exactly match the spec for functions that are not wrapped at all.
 
-The minimum supported PyTorch version is 1.13.
-
 (jax-support)=
 ## [JAX](https://jax.readthedocs.io/en/latest/)
 
@@ -131,8 +124,6 @@ For `linalg`, several methods are missing, for example:
 - `matrix_rank`
 Other methods may only be partially implemented or return incorrect results at times.
 
-The minimum supported Dask version is 2023.12.0.
-
 (sparse-support)=
 ## [Sparse](https://sparse.pydata.org/en/stable/)
 

numpy-1-22-xfails.txt

@@ -42,9 +42,6 @@ array_api_tests/meta/test_hypothesis_helpers.py::test_symmetric_matrices
 # NumPy 1.22 specific XFAILS
 ############################
 
-# finfo has no smallest_normal
-array_api_tests/test_data_type_functions.py::test_finfo
-
 # dlpack stuff
 array_api_tests/test_has_names.py::test_has_names[creation-from_dlpack]
 array_api_tests/test_has_names.py::test_has_names[array_method-__dlpack__]