search

lemma-osu / sknnr

scikit-learn compatible estimators for various kNN imputation methods
https://sknnr.readthedocs.io
0 stars 1 forks source link

Change installed name to `sknnr` #61

Closed aazuspan closed 8 months ago

aazuspan commented 1 year ago

Our previous plan was to use scikit-learn-knn-regression as the installed name of the package and sknnr as the imported name, but per the discussion in #59, we're now tentatively planning to use sknnr across the board.

Pros for switching to sknnr:

Cons for switching to sknnr:

Making the switch will be pretty straightforward, and should just require updating the pyproject.toml, the repository name on Github, and links to the repository throughout the docs (although Github will continue to redirect).

If we do make a final decision to go with sknnr on PyPI, I suggest we make an alpha release just to claim the package name (pre-releases can only be installed with pip install sknnr --pre, so it should be clear that it is not an official, stable release if anyone comes across it).

grovduck commented 1 year ago

If we do make a final decision to go with sknnr on PyPI, I suggest we make an alpha release just to claim the package name (pre-releases can only be installed with pip install sknnr --pre, so it should be clear that it is not an official, stable release if anyone comes across it).

This will be a brand new step in my evolution (posting a package to PyPI)! Sounds intimidating, but I'll follow your lead here and I'm fully supportive of snagging the name.

grovduck commented 8 months ago

@aazuspan, wondering if this is the right time to make this package name change now that we have at least three packages that are depending on it (sknnr-spatial, gee-knn-python, and synthetic-knn). I'm still in favor of making this change if you are. I'll probably want to spectate while you create the PyPI entry, though.

aazuspan commented 8 months ago

Yeah, great point. I think it's time! I'll rename the repo and make a quick PR with the pyproject.toml updates to get it ready for the alpha release.

I'll probably want to spectate while you create the PyPI entry, though.

Want to do it live over Zoom?

grovduck commented 8 months ago

Want to do it live over Zoom?

I can't tell if you're mocking me :wink:, but, yes, I'd love to see you do that ...

aazuspan commented 8 months ago

Haha, not at all! I can screenshare the process, as long as you don't mind watching me fumble my way through the authentication (they just switched how they handle credentials, so my last release took a while to get right!).

grovduck commented 8 months ago

@aazuspan, I'm @grovduck on PyPI. (You must have changed the repo name yesterday at some point as well?)

aazuspan commented 8 months ago

Perfect, just invited you.

You must have changed the repo name yesterday at some point as well?

Yes. Github redirects so it shouldn't be crucial to update anything, but I did update my remote URL with git remote set-url origin https://github.com/lemma-osu/sknnr, just to be on the safe side.

FYI I just found out pip will apparently install a pre-release without the --pre flag if it's the only release available (pypa/pip#5503), so sknnr is installable with just pip install sknnr. We could yank the release if it was a big concern, but I'm not too worried. The upside is we don't need to specify a tag in the downstream package dependencies.

grovduck commented 8 months ago

I did update my remote URL with git remote set-url origin https://github.com/lemma-osu/sknnr, just to be on the safe side

Ah, thanks for this tip. I would have not done this!

FYI I just found out pip will apparently install a pre-release without the --pre flag if it's the only release available (pypa/pip#5503), so sknnr is installable with just pip install sknnr. We could yank the release if it was a big concern, but I'm not too worried.

Agreed, I'm not too worried either.

grovduck commented 8 months ago

Just found a small formatting issue on PyPI with footnotes.

pypi From a quick search of an old SO post, it looks like PyPI now support markdown and is supposed to render exactly as it would on Github:

Currently, PyPI uses cmarkgfm as the markdown renderer, via the readme_renderer library (using readme_renderer.markdown.render(long_description) to produce HTML output). This means that your markdown documents will render exactly the same as on GitHub; it is essentially the same renderer.

It mentions twine here, but I think you said hatch should cover anything that twine does? Happy to open a new issue to track if you think it's worth it.