Skip to content

Commit 4bdced5

Browse files
asmeurerasmeurer
committed
Move the changelog into the docs
1 parent 189973f commit 4bdced5

File tree

3 files changed

+165
-163
lines changed

3 files changed

+165
-163
lines changed

‎CHANGELOG.md

-163
This file was deleted.

‎CHANGELOG.md

+1
Original file line numberDiff line numberDiff line change
@@ -0,0 +1 @@
1+
docs/changelog.md

‎docs/changelog.md

+163
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,163 @@
1+
# Changelog
2+
3+
## 1.5 (2024-03-07)
4+
5+
### Major Changes
6+
7+
- Add support for Dask ([@lithomas1](https://github.com/lithomas1)).
8+
9+
- Add support for JAX. Note that unlike other array libraries,
10+
array-api-compat does not contain any wrappers for JAX functions. All JAX
11+
array API support is in JAX itself. Thus, there is no `array_api_compat.jax`
12+
submodule, and `array_namespace(<JAX array>)` returns the
13+
`jax.experimental.array_api` module.
14+
15+
- The functions `is_numpy_array(x)`, `is_cupy_array(x)`, `is_torch_array(x)`,
16+
`is_dask_array(x)`, `is_jax_array(x)` are now part of the public
17+
`array_api_compat` API.
18+
19+
- Add wrappers for the `fft` extension module for NumPy, CuPy, and PyTorch.
20+
21+
### Minor Changes
22+
23+
- Allow `'2022.12'` as the `api_version` in `array_namespace()`. `'2021.12'`
24+
is also supported but will issue a warning since the returned namespace will
25+
still be a 2022.12 compliant one.
26+
27+
- Add wrapper for numpy.linalg.solve, which broadcasts the inputs according to
28+
the standard.
29+
30+
- Add wrappers for various PyTorch linalg functions.
31+
32+
- Fix a bug with `numpy.linalg.vector_norm(keepdims=True)`.
33+
34+
- BREAKING: Update `vecdot` wrappers to apply `axes` before broadcasting, not
35+
after. This matches the updated 2023.12 standard wording, and also the
36+
behavior of the new `numpy.vecdot` gufunc in NumPy 2.0.
37+
38+
- Fix some linalg functions which were supposed to be in both the main
39+
namespace and the linalg extension namespace.
40+
41+
- Add Ruff to CI. ([@adonath](https://github.com/adonath))
42+
43+
- Test that internal definitions of `__all__` are self-consistent, which
44+
should help to avoid issues where wrappers are accidentally not exported to
45+
the compat namespaces properly.
46+
47+
## 1.4.1 (2024-01-18)
48+
49+
### Minor Changes
50+
51+
- Add support for the upcoming NumPy 2.0 release.
52+
53+
- Added a torch wrapper for `trace` (`torch.trace` doesn't support the
54+
`offset` argument or stacking)
55+
56+
- Wrap numpy, cupy, and torch `nonzero` to raise an error for zero-dimensional
57+
input arrays.
58+
59+
- Add torch wrapper for `newaxis`.
60+
61+
- Improve error message for `array_namespace`
62+
63+
- Fix linalg.cholesky returning the conjugate of the expected upper
64+
decomposition for numpy and cupy.
65+
66+
## 1.4 (2023-09-13)
67+
68+
### Major Changes
69+
70+
- Releases are now made with GitHub Actions (thanks
71+
[@matthewfeickert](https://github.com/matthewfeickert)).
72+
73+
### Minor Changes
74+
75+
- Fix `torch.result_type()` cross-kind promotion
76+
([@lucascolley](https://github.com/lucascolley)).
77+
78+
- Fix the torch.take() wrapper to make axis optional for ndim = 1.
79+
80+
- Add requires-python metadata to the package
81+
([@matthewfeickert](https://github.com/matthewfeickert)).
82+
83+
## 1.3 (2023-06-20)
84+
85+
### Major Changes
86+
87+
- Add [2022.12](https://data-apis.org/array-api/2022.12/) standard support.
88+
This includes things like adding complex dtype support, adding the new
89+
`take` function, and various minor changes in the specification.
90+
91+
### Minor Changes
92+
93+
- Support `"cpu"` in CuPy `to_device()`.
94+
95+
- Return a new array in NumPy/CuPy `reshape(copy=False)`.
96+
97+
- Fix signatures for PyTorch `broadcast_to` and `permute_dims`.
98+
99+
## 1.2 (2023-04-03)
100+
101+
### Major Changes
102+
103+
- Support the linalg extension in the `array_api_compat.torch` namespace.
104+
105+
- Add `isdtype()`.
106+
107+
### Minor Changes
108+
109+
- Fix the `k` keyword argument to `tril` and `triu` in `torch`.
110+
111+
## 1.1.1 (2023-03-10)
112+
113+
### Major Changes
114+
115+
- Rename `get_namespace()` to `array_namespace()` (`get_namespace()` is
116+
maintained as a backwards compatible alias).
117+
118+
### Minor Changes
119+
120+
- The minimum supported NumPy version is now 1.21. Fixed a few issues with
121+
NumPy 1.21 (with `unique_*` and `asarray`), although there are also a few
122+
known issues with this version (see the README).
123+
124+
- Add `api_version` to `get_namespace()`.
125+
126+
- `array_namespace()` (*née* `get_namespace()`) now works correctly with
127+
`torch` tensors.
128+
129+
- `array_namespace()` (*née* `get_namespace()`) now works correctly with
130+
`numpy.array_api` arrays.
131+
132+
- `array_namespace()` (*née* `get_namespace()`) now raises `TypeError` instead
133+
of `ValueError`.
134+
135+
- Fix the `torch.std` wrapper.
136+
137+
- Add `torch` wrappers for `ones`, `empty`, and `zeros` so that `shape` can be
138+
passed as a keyword argument.
139+
140+
## 1.1 (2023-02-24)
141+
142+
### Major Changes
143+
144+
- Added support for PyTorch.
145+
146+
- Add helper function `size()` (required if torch is used as
147+
`torch.Tensor.size` is a method that is incompatible with the array API
148+
[`.size`](https://data-apis.org/array-api/latest/API_specification/generated/array_api.array.size.html#array_api.array.size)).
149+
150+
- All wrapper functions that wrap existing library functions now pass through
151+
arbitrary `**kwargs`.
152+
153+
### Minor Changes
154+
155+
- Added CI to run against the [array API testsuite](https://github.com/data-apis/array-api-tests).
156+
157+
- Fix `sort(stable=False)` and `argsort(stable=False)` with CuPy.
158+
159+
## 1.0 (2022-12-05)
160+
161+
### Major Changes
162+
163+
- Initial release. Includes support for NumPy and CuPy.

‎docs/index.md

+1
Original file line numberDiff line numberDiff line change
@@ -182,5 +182,6 @@ libraries.
182182
183183
helper-functions.rst
184184
supported-array-libraries.md
185+
changelog.md
185186
dev/index.md
186187
```

0 commit comments

Comments
 (0)