XXX version-specific blurb XXX
- Redesign the caterva guts and perform a code refactorization in order to simplify the code. This includes API renaming. A performance improvement is obtained as can be seen in FrancescAlted#69 (comment)
- Implement a set_slice for arrays backed by Blosc. This allows users to update the values in the array whenever and wherever they want.
- Implement constructors (empty, zeros, full) using the special-values features introduced in Blosc.
- Use the pydata_sphinx_theme in the documentation.
- API renaming. The function names simulate a namespace and some variable names have been changed to those used by the general community (part -> chunk).
- Add a new level of multi-dimensionality. As a result, unlike other libraries, Caterva supports two levels of multi-dimensional chunking (chunks and blocks).
- Improve library compilation to allow users to avoid building tests and examples.
- Simplify the test's suite for a proper integration in Windows.
- Update documentation to improve the library description and to add the Release notes and the Roadmap in a new section.
- Fixing that 0.3.1 and 0.3.2 tags were not made in master :-/
- Big code and API refactorization. As result, the API is more consistent and hopefully more intuitive to use. For more info on the new API, see https://caterva.readthedocs.io.
- Add a caterva_from_sframe() function.
- Both static and dynamic libraries are created by default now. If you want to disable the creation of one of them, just set the cmake options SHARED_LIB=OFF or STATIC_LIB=OFF.
- Add a copy parameter to caterva_from_file().