Version 2.8.0: support numpy >= 2
NNGT now equires matplotlib >= 3.6 and igraph >= 0.10.
**Improvements and bugfixes**
- improved colorbar for plots
- fix fallback if requested backend fails
- avoid zero width/height shapes
- improve shape falback if shapely is missing
**Compatibility**
- better matplotlib compatibility for graph-tool and igraph
- move to np.nan instead of NaN for numpy 2
- move to ``packaging`` instead of ``pkg_resources``
**Maintenance and cleanup**
- update build tests and args in setup.py and pyxbld
- remove use of old numpy random (randint/get_state)
- remove some unused imports
- ability to run compatibility tests for a subset of backends
- use venv for CI
- skip graph-tool in tests until it is fixed
- fix documentation
Single line description + ignore build exceptions
Catch compile errors and use modern OpenMP with MSVC
Make installation more resilient
Improved wheels and sdist builds for shipping
Version 2.7.1
Improved configuration (added save_config and reset_config functions)
Fix support for recent scipy and networkx version
Improved matplotlib support with graph-tool and igraph (cairo)
Fix closeness with networkx>=2.6
Doc: fix build
Update RtD conda environment and yml
Fix deprecated alltrue in nngt_config.py
Maintenance: scipy, graph-tool, mplbackend updates
* bump graph-tool minimum version to 2.46
* update Readthedocs environment
* move to scipy.signal.windows
* fix sparse matrix import
* improve mpl backend support for gt/igraph
Version 2.7.0
Support graph-tool >= 2.4.6
Support networkx >= 3
Move to pyproject.toml
Make project REUSE compliant
Fix bug for igraph node attributes
Bugfix: remove typa nattr edit in _IGraph.new_node
Readd NEST, update Readme and doc
Update documentation dependencies
Use pyproject.toml and make REUSE compliant
- setup.py now only for editable install
- update dependencies for doc
- update project data
- move to REUSE scheme for licenses
Networkx 3 and graph-tool 2.4.6 support
Fix new graph-tool import for remove_parallel_edges
Support networkx 3 and new scipy sparse arrays
Temporary fix for the build by forcing matplotlib < 3.7 and using LXML
Version 2.6.0
Include minor change to the NEST functions that trigger a warning plus
add an explanation on how to hide the warnings.
Maintenance: C++ tuple hash/equal functions, doc
Remove deprecated unary and binary function types from generation/func_connect.h
Update doc:
- restore utf-8 charset
- synchronize ReadTheDocs from SourceHut
- set dark/light mode for math
Docfix: remove sphinx-bootstrap-theme