jasperroebroek/sklearn-quantile

Packaging & Testing

Closed this issue · 1 comments

Hey Jasper,

We communicated briefly on skgarden, where you pointed me here -- thanks for putting this together!

I had a couple of questions / comments / areas where I thought I could chip in, and was wondering what your thoughts were on the following:

  • small, nit thing: I see sklearn_quantile as the package name in the setup.py, but I see sklearn-quantile (which fits Python style guide) in PyPI. Am I just missing something because I'm tired (totally possible) or are you intended to have sklearn_quantile be the package name?
  • I don't see instructions to install via pip on the README -- is that intentional (i.e. is there anything special going on in the conda install) or should the pip install work fine?
  • it's great that you have some tests here -- would you be open to me adding more?

Best,

Abe

Hi Abe,

thanks for reaching out!

  • I intended to use sklearn-quantile as the package name, but use sklearn_quantile as the name for importing in python. This seems to be quite common approach (see for example flask-login and flask-sqlalchemy). Not sure if this is the clearest approach, I am open to suggestions.
  • Pip install should work fine
  • I am always in favor of adding more tests if that improves stability. I currently don't have much time to invest in expanding them specifically. Again, I am open to suggestions!

Cheers,
Jasper