|
1 |
| -Announcing Python-Blosc2 3.8.0 |
| 1 | +Announcing Python-Blosc2 3.9.1 |
2 | 2 | ==============================
|
3 | 3 |
|
4 |
| -This is a minor version release where we have aimed to make the first steps |
5 |
| -towards complying fully with the array-api standard: |
6 |
| - |
7 |
| -✅ C-Blosc2 internal library updated to latest 2.21.3. |
8 |
| - |
9 |
| -✅ numexpr version requirement pushed to 2.3.0 to incorporate |
10 |
| -``round``, ``sign``, ``signbit``, ``copysign``, ``nextafter``, ``hypot``, |
11 |
| -``maximum``, ``minimum``, ``trunc``, ``log2``, ``//`` and bitwise AND/OR/XOR/NOT. |
12 |
| - |
13 |
| -✅ Enhanced linear algebra via ``tensordot``, extended ``matmul``, ``vecdot``, ``diagonal``, |
14 |
| -``outer``, as well as ``broadcast_to`` and ``meshgrid``. |
15 |
| - |
16 |
| -✅ Roughly 60 attributes and methods for the ``NDArray`` class, as well as functions, have |
17 |
| -been added to increase compliance with the array-api standard, with ufuncs now very well supported. |
18 |
| - |
19 |
| -✅ A subtle bug which caused ``arr = blosc2.expand_dims(arr)`` to fail has been fixed. |
| 4 | +This is a patch release where we have aimed to tidy up the code and improve robustness: |
20 | 5 |
|
| 6 | +✅ Bumped to numexpr 2.13.1 to incorporate new maximum/minimum NaN handling and +/* for booleans |
| 7 | +which matches NumPy behaviour. |
| 8 | +✅ Refactoring in order to ensure Blosc2 functions with NumPy 1.26. |
| 9 | +✅ Streamlined documentation by introducing Array Protocol |
21 | 10 |
|
22 | 11 | You can think of Python-Blosc2 3.x as an extension of NumPy/numexpr that:
|
23 | 12 |
|
|
0 commit comments