Numpy 2.0 update causing issues with other software

[Josh-XT]

Forcing numpy to 2.0+ forces breaking changes on other software. Unless it is absolutely required to use 2.0+, can this be reverted?

184e508

[rphellan]

Yes, also numpy 2.0.0 requires python 3.9. Then, the line numpy>=2.0.0; python_version < "3.9" leads to a bug.

[Darshan2104]

Yes it is creating issue. I am facing issue while working with keyphrase-vectorizers==0.0.10 where it is using spacy version >=3.0.0 and it is installing spacy==3.8.2 i'm using python3.8 with numpy==1.24.3 which is compatable but spacy3.8.2 is looking for numpy2.0+ and that is not compatable with python3.8.

[lsmith77]

adding fasttext to the list of libs not compatible (and unfortunately also not maintained). we have found it to be the best language detection for shorter text.

[simon-ging]

@lsmith77 off-topic but I uploaded a numpy2 version of fasttext, because I ran into the same problem, and it was just a one line fix. see https://pypi.org/project/fasttext-numpy2/

[honnibal]

There's no way to support both numpy v1 and numpy v2 in the same release. You'll need to pin your spaCy dependency to have an upper-bound that uses numpy v1.