On Fri, Jun 29, 2018, 8:14 PM Andrei Kucharavy <andrei.kucharavy@gmail.com> wrote:

> Not all packages are within the numpy/scipy universe - Pandas and Seaborn are notable examples.

Huh?! Pandas is a thin wrapper around NumPy. To be fair, it is a wrapper that adds a huge number of wrapping methods and classes. Seaborn in turn has at least a soft dependency on Pandas (some of the charts really need a DataFrame to work from).

I like the idea of standardizing curation information. But it has little to do with Python itself. Getting the authors of scientific packages to agree on conventions is what needed, and doing that requires accurately determining their needs, not some mandate from Python itself. Nothing in the language needs to change to agree on some certain collection of names (perhaps dunders, perhaps not), and some certain formats for the data that might live inside them.

Down the road, if there gets to be widespread acceptance of these conventions, Python standard library might include a function or two to work with them. But the horse should go before the cart.