Compatibility release

This release brings ulab in line with micropython 1.23. In addition to the not-so-visible compatibility with the latest of micropython, bitwise operators have been implemented, the random module has been added and a number of bugs have been squashed.

What's Changed

• ulab.numpy: implement sinc for creating audio filters by @jepler in #617
• fix ndarray subscription method by @v923z in #619
• add the out keyword argument to universal functions by @v923z in #621
• fix fetching submodules (cp build process change) by @jepler in #627
• add bitwise operators by @v923z in #616
• Bitwise by @Derfies in #628
• Int overflow by @FelixNumworks in #629
• [ndarray] Fix ndarray_from_tuple reading out of _shape->items bounds by @FelixNumworks in #630
• Polyval handles non-array as second argument by @HugoNumworks in #601
• Fix -Wunused-variable diagnostic when !ULAB_SUPPORTS_COMPLEX by @jepler in #631
• allow function iterator in math functions by @v923z in #633
• Per github docs, run apt-get update by @jepler in #637
• numpy/vector.c: remove usage of fpclassify by @kbsriram in #636
• implement AND, OR, XOR binary operators by @v923z in #639
• Fix #643 by @qqice in #645
• Drop certain CircuitPython workarounds that are no longer needed by @jepler in #647
• Fix (u)int8 upcasting as per docs and numpy by @s-ol in #650
• fix the np.delete bug by @hiltay in #653
• add random module by @v923z in #654
• add random documentation rst file by @v923z in #655
• add link to random module documentation in readme by @v923z in #659
• fix reshape by @v923z in #660
• Update type annotations in compare.c and vector.c by @kbsriram in #663
• Remove the STATIC macro. by @Gadgetoid in #664
• fix loadtxt for the case, when built-in complexes are not supported by @v923z in #666

New Contributors

• @Derfies made their first contribution in #628
• @kbsriram made their first contribution in #636
• @qqice made their first contribution in #645
• @s-ol made their first contribution in #650
• @hiltay made their first contribution in #653

Full Changelog: 6.0.12...6.5.2

Compatibility release

This release brings ulab in line with the latest micropython, and fixes a number of smaller bugs.

Move `spectrogram` to `utils` sub-module

This release moves the spectrogram function to the utils sub-module, adds the hash module constant, implements a great number of small bug fixes, and brings ulab in line with micropython.

upgrade with complex support

This release of ulab adds optional support for complex arrays, the compress function, the .tolist() array method, and the .imag, .real array properties.

Complex arrays can be sliced and iterated on, have the usual unary operators, and the +, *, /, -, ==, and != binary operators. In addition, the all, any, convolve, exp, sqrt functions accept complex arguments, and the conjugate, imag, real, sort_complex functions are implemented.

If complex support is enabled, the FFT routines can be configured to behave in a completely numpy-compatible way.

Fix compare/sum

This bug fix release rectifies compare's, and sum's handling of integer/Boolean types.

Documentation improvements

This release adds the readthedocs yaml file, so that documentation can be generated automatically.

Documentation improvements

With this release, documentation for version 0.54.6 can be generated automatically.

Documentation improvements

With this release, documentation for version 0.36 can automatically be generated.

new .flat property, numpy.linalg.qr function

This release implements the .dtype, .itemsize, .shape, .size, and .strides getter properties for micropython (these getters are already included in circuitpython), and adds the .T ndarray property for the transpose, the .shape setter (in micropython), and the flat iterator for higher dimensional arrays, accessible via the .flat property (in micropython).

In addition, the numpy.linalg module has a new function, qr for the QR decomposition. The bug in the calculation of the sum of 4D arrays has also been squashed.

Behind the scenes, the code base has been re-organised, so that the circuitpython documentation can be generated in a consistent way.

new module, properties implementation, and bug fixes

This release of ulab adds a new module, scipy.linalg, with two functions, and the empty array constructor function on the python side, and implements numpy-compatible handling of ndarray properties. In addition, it also fixes a number of bugs. More details can be found in the change log.