fix(py-vbi): add version 0.1.3

That would be also nice to put cupy as default requirements, It's optional on original GitHub. Is it possible?

This release improves the project's usability for developers and researchers by enhancing GPU-accelerated model implementations (via CuPy/Numba), refining Docker-based deployment, and strengthening CI/CD pipelines for PyPI publishing. Documentation has been significantly polished for better accessibility, and new datasets expand the toolkit's capabilities

@ziaee did you release the same version (0.1.3) again? In this case, you need to delete the patch file and the patch() statement from package.py, since it can't be applied anymore.

If you need to add cuda-related dependencies, please use a cuda variant (or just use the CudaPackage Mixin class)

yes, I updated the tag in GitHub, can we just add py-cupy as dependency? https://packages.spack.io/package.html?name=py-cupy

added 2 commits

• aed29b42 - fix(py-vbi): remove obsolete installation patch
• 53a1909f - feat(py-vbi): add dependency on py-cupy for build and run

Compare with previous version

sure, but since CUDA is not available in all the environments where the ESD is deployed (e.g, the Lab environment does not have GPUs), we need a way to selectively depend on it only when it makes sense. The easiest way to do this is to use the CudaPackage class that provides a cuda variant - see for e.g. how py-elephant depends on py-cuda

added 1 commit

• 4c92b116 - Apply 1 suggestion(s) to 1 file(s)

Compare with previous version

resolved all threads

added 1 commit

• e10d43a2 - Apply 1 suggestion(s) to 1 file(s)

Compare with previous version

@ziaee it seems that one test is failing, you can see the logs here

added 2 commits

• 99a7e236 - update vbi for v0.1.3
• 4a5082ee - Merge branch 'add_vbi_0.1.3' of gitlab.ebrains.eu:ziaee/ebrains-spack-builds into add_vbi_0.1.3

Compare with previous version

resolved all threads

merged

@ziaee the changes have been deployed now, could you test again?

The dependecies are imported except cupy which is not accessible. So I could not test cupy models.

setup.py compile C++ codes in vbi.models.cpp._src

I see the compiled libraries here:

ls /srv/test-build-2502/ebrains-spack-builds/vendor/spack/var/spack/environments/test/.spack-env/view/lib/python3.11/site-packages/vbi/models/cpp/_src/*.so

_do.cpython-311-x86_64-linux-gnu.so       _jr_sde.cpython-311-x86_64-linux-gnu.so  _mpr_sde.cpython-311-x86_64-linux-gnu.so  _wc_ode.cpython-311-x86_64-linux-gnu.so
_jr_sdde.cpython-311-x86_64-linux-gnu.so  _km_sde.cpython-311-x86_64-linux-gnu.so  _vep.cpython-311-x86_64-linux-gnu.so

but not imported.

Sorry for the delay - yes, indeed, the problem is that vbi.__init__.py is not explicitly exposing the models subpackage:

import vbi
dir(vbi)

['LoadSample',
 '__builtins__',
 '__cached__',
 '__doc__',
 '__file__',
 '__loader__',
 '__name__',
 '__package__',
 '__path__',
 '__spec__',
 '__version__',
 '_version',
 'add_feature',
 'add_features_from_json',
 'calc_features',
 'display_time',
 'extract_features',
 'extract_features_df',
 'extract_features_list',
 'feature_extraction',
 'get_features_by_domain',
 'get_features_by_given_names',
 'get_version',
 'j2p',
 'make_mask',
 'os',
 'p2j',
 'posterior_peaks',
 'report_cfg',
 'test_imports',
 'tests',
 'timer',
 'update_cfg',
 'utils']

so import vbi.models fails

no worries, thanks

(fixed in !654 (merged))