Julien Jerphanion
Julien Jerphanion
Have you tried this snippet on newer versions of PyMesh?
Any ideas @qnzhou? 🙂
I've started #247 as a proposal.
Hello, I also get the same problem on Fedora 32 using: - python 3.8 (from package manager) - python 3.6, 3.7 from anaconda be it by installing it from source...
scikit-learn defintes [private class to parse versions](https://github.com/scikit-learn/scikit-learn/blob/5efcb10c3e9f87103c404c32a036beb623182c83/sklearn/externals/_packaging/version.py#L280). It might be possible to privately import this class (i.e. `sklearn.externals._packaging.Version` and expose it as a private symbol of `dirty_cat` (e.g `dirty_cat._packaging.Version`) to...
A possible solution is to vendor [a (modified) implementation of `Version` from `pypa/packaging` like scikit-learn did](https://github.com/scikit-learn/scikit-learn/blob/5efcb10c3e9f87103c404c32a036beb623182c83/sklearn/externals/_packaging/version.py#L1-L3).
I do obtain the same problem on Fedora 32 when building from source.
Thanks @akaszynski !
It looks like the project is unmaintained unfortunately.
Thanks for participating in this conversation, @betatim. > I like this way of thinking about this. In particular because you want to replace more than just one method (fit()) but...