[Numpy-discussion] ENH: Proposal to add atleast_nd function

[Stephan Hoyer]

On Wed, Feb 17, 2021 at 2:37 AM Ralf Gommers <ralf.gommers at gmail.com> wrote:

>
>
> On Wed, Feb 17, 2021 at 12:26 AM Stefan van der Walt <stefanv at berkeley.edu>
> wrote:
>
>> On Tue, Feb 16, 2021, at 07:49, Joseph Fox-Rabinovitz wrote:
>>
>> I'm getting a generally lukewarm not negative response. Should we put it
>> to a vote?
>>
>>
>> Things here don't typically get decided by vote—I think you'll have to
>> build towards consensus.  It may be overkill to write a NEP, but outlining
>> a proposed solution along with pros and cons and getting everyone on board
>> is necessary.
>>
>> The API surface is a touchy issue, and so it is difficult to get new
>> features like these added.
>>
>
> This function is less bad than most similar utility functions, because it
> starts with atleast_ so from a "function browsing" end user perspective
> it's not much additional clutter. But it does still force other libraries
> to do work because they aim to be compatible to numpy's main namespace
> (e.g. see jax.numpy).
>
> And there's 6-7 maintainers all not strongly opposed but also not
> enthusiastic.
>

I agree with Ralf's assessment.

This function feels like a natural generalization of existing NumPy
functionality, but we don't expand NumPy's API without use-cases. That's
just a waste of time for everyone involved.

I am most moved by Juan's report that he has the "very distinct impression
of needing it repeatedly," but I would still love to see concrete examples
of where users have found this be helpful.

It is not a hard function to write, so if it was useful I would expect to
see some version of it in an existing open source project or at least on
StackOverflow.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.python.org/pipermail/numpy-discussion/attachments/20210218/bb8c830f/attachment.html>