[prev in list] [next in list] [prev in thread] [next in thread] 

List:       python-distutils-sig
Subject:    Re: [Distutils] new PyPI: a rant from a package maintainer
From:       Nick Coghlan <ncoghlan () gmail ! com>
Date:       2017-08-07 5:02:29
Message-ID: CADiSq7edA+5JqO75Lq1QsCmuYuk98ymmFc_bkD9rbcENC3V+GA () mail ! gmail ! com
[Download RAW message or body]

On 7 August 2017 at 12:19, Randy Syring <randy@thesyrings.us> wrote:
> Thanks Donald for the clear reply from which it was easy to create these
> separate issues.  I didn't realize you had so little paid time to work on
> packaging issues.  I'm surprised with all the large companies that use
> Python that the PSF can't secure more donations to invest in fixing these
> issues.  Or that someone like Redhat can't get a couple engineers to devote
> time for a couple years like they do to other OSS projects. (No criticism
> intended, just musing).

Please feel free to criticise Python's commercial redistributors
loudly and publicly on this point, as our collective failure to deal
with the situation appropriately is something our respective customers
*should* be questioning and challenging us over, especially when
support contracts are up for renegotiation :)

As far as I've been able to work out, the main problem appears to be
that the professional Python user base currently self-selects into two
primary categories:

1. "We're happy to rely on self-support and community support, so we
don't need a Python vendor"
2. "Our ops team pays a commercial Python runtime vendor, but our dev
teams have no ability to file support requests with that vendor"

In combination, these behaviours mean vendors' commercial demand
signals are thoroughly broken, creating an environment where it
becomes difficult to make the business case that increased upstream
investment will lead to proportional increases in revenue. (To help
illustrate the scale of the problem, while omitting names to protect
the guilty: "Do people in large organisations actually use Python for
web development?" is something I have had a redistributor ask as a
genuine question in the same month where Python topped IEEE Spectrum's
annual survey of language popularity for the first time. When we have
that level of disconnect in the perception of ecosystem adoption,
something significant is clearly amiss)

At the Python Software Foundation level, one of the things we've
realised we need to do is to provide more streamlined ways for
organisations using Python to direct funding specifically towards
infrastructure support for PyPI (with the Packaging Working Group then
being responsible for managing those funds), since "We should fund the
platform that hosts our Python dependencies" is a potentially easier
pitch for engineering teams to make than advocating for their
employers to become full PSF Sponsor members.

Towards that end, we've recently applied to Mozilla for a development
grant to help the PSF fund the design, development, and project
management efforts needed to get those improved infrastructure
sustainability measures in place, as well as getting to the point
where the legacy PyPI service can be shut down entirely, eliminating
one of the current major barriers to improving the shared
infrastructure (credit goes to Eric Holscher for getting that request
through to the grant submission stage - the Packaging WG had started
drafting it some time ago, but we'd stalled before reaching the point
of actually submitting it).

Cheers,
Nick.

-- 
Nick Coghlan   |   ncoghlan@gmail.com   |   Brisbane, Australia
_______________________________________________
Distutils-SIG maillist  -  Distutils-SIG@python.org
https://mail.python.org/mailman/listinfo/distutils-sig
[prev in list] [next in list] [prev in thread] [next in thread]