Antony Lee

For instructions on git/prs/github, consider checking matplotlib's contribution docs (https://matplotlib.org/stable/devel/contributing.html), including their gitter channel. I don't have much time right now for evaluating the details of the feature, but I...

1\) and 2) seem fine (though things may need to be clarified regarding how your "enabled" attribute differs from the already existing Cursor.enabled). Regarding 3) I think it may be...

> Regarding (2), I wrote ENABLED, but actually the new property name is ACTIVE, and is only relevant to bindings. If I am not mistaken, ENABLED = False also disables...

So I think there's at least another reason why auto-pip'ing is a bad idea: there's a bunch of modules where the module names does _not_ correspond to the PyPI name;...

I think you should also match against the _alias_ if present.

This doesn't match against "import foo as bar". A few more ideas: - I'd probably get rid of the common_others mapping (because the user's commonly used abbreviations are going to...

I assume it's not in a finished state yet? There's still a bunch of things apparently left to complete.

It doesn't, but I could probably accept a PR for that.

This was not supported due to https://github.com/pypa/pip/issues/3797 (and the limitation (or rather a similar one) is noted in https://github.com/anntzer/pypi2pkgbuild/blob/main/TODO.rst#issues); I guess I could consider a PR that fixes that.

Sounds reasonable to add a `-d/--depends python-pytorch` or similar flag to allow the end user to 1) manually specify additional depends and 2) resolve such ambiguities. A patch would be...