Skip to content

Commit 336d38e

Browse files
committed
DOC: start adding a changelog
1 parent 590a2de commit 336d38e

File tree

1 file changed

+49
-0
lines changed

1 file changed

+49
-0
lines changed

docs/changelog.md

+49
Original file line numberDiff line numberDiff line change
@@ -1,5 +1,54 @@
11
# Changelog
22

3+
## 2.3 (2025-XX-XX)
4+
5+
### Major Changes
6+
7+
- Binary elementwise functions now accept python scalars: the only requirement is that
8+
at least one of the arguments must be an array; the other argument may be either
9+
a python scalar or an array. Python scalars are handled in accordance with the
10+
type promotion rules, as specified by the standard.
11+
This change unifies the behavior of binary functions and their matching operators,
12+
(where available), such as `multiply(x1, x2)` and `__mul__(self, other)`.
13+
14+
``where`` accepts array or scalars as its 2nd and 3rd arguments, ``x1`` and ``x2``. The first argument, ``condition``, must be an array.
15+
16+
- The following functions, new in the 2024.12 standard revision, are implemented:
17+
18+
- `count_nonzero`
19+
- `cumulative_prod`
20+
21+
22+
### Minor Changes
23+
24+
- `vecdot` now conjugates the first argument, in accordance with the standard.
25+
26+
- `astype` now raises a `TypeError` instead of casting a complex floating-point
27+
array to a real-valued or an integral data type.
28+
29+
- `where` requires that its first argument, `condition` has a boolean data dtype,
30+
and raises a `TypeError` otherwise.
31+
32+
- `isdtype` raises a `TypeError` is its argument is not a dtype object.
33+
34+
- arrays created with `from_dlpack` now correctly set their `device` attribute.
35+
36+
- `mean` accepts complex floating-point arrays.
37+
38+
39+
### Contributors
40+
41+
The following users contributed to this release:
42+
43+
Aaron Meurer
44+
Clément Robert
45+
Guido Imperiale
46+
Evgeni Burovski
47+
Lucas Colley
48+
Tim Head
49+
50+
51+
352
## 2.2 (2024-11-11)
453

554
### Major Changes

0 commit comments

Comments
 (0)