Incompatible license with scipy about scikit-tensor

[ahwillia]

Sorry to dig up this old issue, but I want to give my +1 for an MIT license. Development on this repo doesn't seem too frequent and it could free up others to push methods in this space further.

[ogrisel]

+1

[zygmuntz]

or maybe LGPL

[ogrisel]

LGPL is not ideal either: as scipy community members we want to be able to copy snippets of code from downstream projects up to upstream projects (like SciPy, NumPy or even Python) when it make sense without having to ask the permission to re-license those code snippets under a license compatible with the upstream projects.

But off course the author is free to do what he wishes. Just be aware that many numpy scipy community members will not use that project or consider it as a possible dependency of their own projects because of the GPL or LGPL license.

[mnick]

Thank you all for the suggestions and the points brought up.

Converting to a sklearn-like API should be doable and I'll put it on my todo list.

The licensing question has already been brought up via email and I will reconsider it. As a researcher I like many things about the GPL such that I'm reluctant to move away from it. However, I understand that the GPL is suboptimal from a packaging/community point of view and this is indeed a strong argument for switching to a MIT/BSD license.