Jianlin Shi

Thanks. What is PR proposing ?

BTW, @Angrynut , is this scala? How to configure it in Java? There is not a class named "DruidDataSource". Is this another dependency? Thanks

Yes. We would like too. However, spaCy versions run too fast, and has back compatibility issues. It's a hard decision to choose the best base version. Any suggestions are welcome.

I know spansgroup is an issue between early 3.x and later 3.x versions. It's one little thing, but affect a lot of places.

agree. But in some cases, it could be better in the other way. We could make this as an option to let user to choose.

Failed to install unqlite Building wheels for collected packages: unqlite Building wheel for unqlite (setup.py): started Building wheel for unqlite (setup.py): finished with status 'error' error: subprocess-exited-with-error × python setup.py...

Sorry, missed this thread. Is it still an issue with latest release?

Should have been fixed.

For the large files, do they fit into these limits? https://docs.github.com/en/repositories/working-with-files/managing-large-files/about-git-large-file-storage If so, I think making a copy of these models under the project repo can make it faster to...

Another option is to set these models as caches in CI https://docs.github.com/en/actions/using-workflows/caching-dependencies-to-speed-up-workflows I think this might be a better option